Sourcing Simulator Users Guide

Sourcing SimulatorVersion 2.1

Users’ Guide

Russell E. KingHenry L. W. Nuttle

CopyrightNorth Carolina State University

April 1999

i

1 INTRODUCTION................................................................................................................................. 1

1.1 OVERVIEW OF THE SOURCING SIMULATOR ...................................................................................... 11.2 INSTALLING THE SIMULATOR ON YOUR COMPUTER ......................................................................... 1

2 OVERVIEW OF THE SOURCING ANALYSIS AND DECISION SURFACE MODELS........... 2

2.1 OVERVIEW OF THE SOURCING ANALYSIS MODEL ............................................................................ 22.2 OVERVIEW OF DECISION SURFACE MODEL...................................................................................... 3

3 HOW TO RUN THE SOURCING SIMULATOR............................................................................. 4

3.1 RUNNING THE MODELS.................................................................................................................... 43.2 SAVING AND PRINTING DATA........................................................................................................... 7

4 SOURCING ANALYSIS MODEL INPUTS....................................................................................... 8

4.1 SOURCING ANALYSIS MODEL INPUTS .............................................................................................. 94.1.1 Buyer's Plan Tab (Figure 4.1) ................................................................................................ 94.1.2 Cost Data Tab (Figure 4.2)................................................................................................... 124.1.3 Consumer Demand Tab (Figure 4.3) .................................................................................... 144.1.4 Markdowns / Promotions Tab (Figure 4.4) .......................................................................... 164.1.5 Sourcing Strategy (Figure 4.5) ............................................................................................. 194.1.6 Vendor Specification (Figure 4.6)......................................................................................... 24

4.2 BATCH RUN INPUTS....................................................................................................................... 32

5 SOURCING ANALYSIS MODEL OUTPUTS ................................................................................ 35

5.1 SIMULATION RESULTS TABLE (FIGURE 5.1) ................................................................................... 355.1.1 Output Statistics .................................................................................................................... 35

Raw Material Inventory Measures .....................................................................................................................385.1.2 Displaying Results from Several Scenarios .......................................................................... 405.1.3 Editing and Tailoring the Results Table ............................................................................... 405.1.4 Saving and Printing (Tabular) Simulation Results ............................................................... 415.1.5 Viewing Inputs ...................................................................................................................... 41

5.2 WEEKLY PERFORMANCE GRAPHS.................................................................................................. 425.2.1 Creating Graphs (Figure 5.3) ............................................................................................... 445.2.2 Side-by-Side Display of Graphs............................................................................................ 455.2.3 Reformatting Graphs............................................................................................................. 455.2.4 Saving and Printing Graphs ................................................................................................. 46

5.3 BREAK-EVEN ANALYSIS ................................................................................................................ 465.3.1 Inputs .................................................................................................................................... 475.3.2 Results ................................................................................................................................... 47

6 CREATING AND PLOTTING DECISION SURFACES ............................................................... 49

6.1 CREATING A NEW DECISION SURFACE MODEL .............................................................................. 496.2 PLOTTING A DECISION SURFACE .................................................................................................... 516.3 SIDE-BY-SIDE DISPLAY OF DECISION SURFACES ............................................................................ 56

7 SAMPLE SCENARIOS...................................................................................................................... 57

7.1 TUTORIAL SCENARIOS ................................................................................................................... 577.2 SCENARIO 1 - QUICK RESPONSE SOURCING ................................................................................... 57

7.2.1 Data Entry............................................................................................................................. 577.2.2 Running the Simulation......................................................................................................... 597.2.3 Interpreting the Simulation Results Table............................................................................. 607.2.4 Viewing the Inputs................................................................................................................. 607.2.5 Printing the Results Table..................................................................................................... 617.2.6 Saving the Results Table ....................................................................................................... 617.2.7 Viewing Graphs of Results .................................................................................................... 61

ii

7.2.8 Printing the Graphs .............................................................................................................. 637.2.9 Saving the Graphs................................................................................................................. 637.2.10 Running with Several Volume Errors ................................................................................... 63

7.3 SCENARIO 2 - TRADITIONAL SOURCING STRATEGY ....................................................................... 647.3.1 Data Entry............................................................................................................................. 647.3.2 Saving the Input Data ........................................................................................................... 667.3.3 Running the Simulation......................................................................................................... 667.3.4 Interpreting the Tabled Results............................................................................................. 667.3.5 Break-Even Analysis ............................................................................................................. 677.3.6 Printing the Results Table..................................................................................................... 677.3.7 Viewing Graphs of Results .................................................................................................... 687.3.8 Printing the Graphs .............................................................................................................. 697.3.9 Saving the Graphs................................................................................................................. 697.3.10 Saving (Deleting) Tabled Results.......................................................................................... 69

7.4 SCENARIO 3 - VENDOR MANAGED INVENTORY SOURCING............................................................ 697.4.1 Data Entry............................................................................................................................. 707.4.2 Running the Simulation......................................................................................................... 707.4.3 Looking at the Tabled Results ............................................................................................... 707.4.4 Editing the Results Table ...................................................................................................... 717.4.5 Viewing Graphs of Results .................................................................................................... 71

7.5 SCENARIO 4 - NEWSBOY SOURCING.............................................................................................. 727.5.1 Data Entry............................................................................................................................. 727.5.2 Running the Simulation......................................................................................................... 727.5.3 Looking at the Results Table................................................................................................. 727.5.4 Viewing Graphs of Results .................................................................................................... 73

7.6 SCENARIO 5 A BASICS SCENARIO USING THE MODEL STOCK STRATEGY....................................... 737.6.1 Data Entry............................................................................................................................. 747.6.2 Running the Simulation......................................................................................................... 767.6.3 Interpreting the Simulation Results Table............................................................................. 767.6.4 Viewing the Inputs................................................................................................................. 777.6.5 Printing the Results Table..................................................................................................... 777.6.6 Saving the Results Table ....................................................................................................... 777.6.7 Viewing Graphs of Results .................................................................................................... 777.6.8 Printing the Graphs .............................................................................................................. 787.6.9 Saving the Graphs................................................................................................................. 78

7.7 SCENARIO 6 - COMPARING SOURCING STRATEGIES FOR BASICS .................................................... 787.7.1 Data Entry and Execution..................................................................................................... 787.7.2 Looking at the Tabled Results ............................................................................................... 797.7.3 Viewing Graphs of Results .................................................................................................... 80

7.8 SCENARIO 7 - USING THE ‘PRESENTATION STOCK’ FEATURE ........................................................ 807.8.1 Data Entry and Execution..................................................................................................... 807.8.2 Looking at the Tabled Results ............................................................................................... 81

7.9 SCENARIO 8 - CREATING A SMALL DECISION SURFACE MODEL FOR A BASICS SCENARIO ............. 817.9.1 Data Entry and Execution..................................................................................................... 817.9.2 Specifying the Decision Surface Model................................................................................. 837.9.3 Saving the Decision Surface Model Data ............................................................................. 837.9.4 Plotting a Surface ................................................................................................................. 847.9.5 Adjusting the Scales on the Axes........................................................................................... 847.9.6 Switching the Displayed Measure......................................................................................... 847.9.7 Printing and Saving Plots ..................................................................................................... 857.9.8 Starting from Different Points in the Process ....................................................................... 85

7.10 SCENARIO 9 - CREATING A BIGGER DECISION SURFACE MODEL FOR A SEASONAL SCENARIO....... 867.10.1 Data Entry and Execution..................................................................................................... 867.10.2 Specifying the Decision Surface Model................................................................................. 867.10.3 Plotting Surfaces................................................................................................................... 87

iii

7.10.4 Interpreting and Manipulating 3-D Plots ............................................................................. 877.10.5 Displaying Different Plots from the Same Decision Surface ................................................ 88

7.11 SCENARIO 10 USING THE DETAILED VENDOR SPECIFICATION........................................................ 90

1

1 Introduction

1.1 Overview of the Sourcing Simulator

The Sourcing Simulator is an interactive software package that provides the capability to:

1. Simulate retailing and retailing/manufacturing scenarios for a line of product (SourcingAnalysis Model).

2. Model the relationship between selected operational parameters and/or environmentalfactors and retail performance measures (Decision Surface Model).

The package is designed to:

1. Allow the user to evaluate the impact of various retail and manufacturing operatingpractices under different operating conditions.

2. Provide a training tool for retail buyers and managers.

It is designed to run on Windows-based systems without requiring extraneous proprietarysoftware.

This document provides information on content, installation, and use of the Sourcing Simulator.Installation instructions are provided in Section 1.2. Chapter 2 gives overviews of the SourcingAnalysis and Decision Surface Models. Chapter 3 gives instructions on running the models.Chapters 4 and 5 provide detailed descriptions of inputs and outputs, for the Sourcing AnalysisModel. Chapter 6 gives instructions on creating and plotting decision surfaces. Chapter 7contains tutorial examples for the models.

Most of the information in this document is also available via on-line “Help” within the packageitself.

1.2 Installing the Simulator on your ComputerTo install the Simulator on your PC from diskettes, place Disk 1 in the floppy drive. Then selectthe Start button and click on Run. In the Run window type a:\setup and press return or click onthe OK button.

To install the Simulator on your PC from CD, place the Install Disk in the CD drive. If you havethe “auto run” feature turned on, the installation will start automatically. If not, select the Startbutton and click on Run. In the Run window select the drive associated with your CD player andthen click on the file setup.exe and press return or click on the OK button.

In either case, during installation you will be asked to select a directory name to install the files in.Then, a message will appear indicating when the installation has been successfully completed.You are then ready to run the Simulator.

NOTE: If a conflict occurs with an application (MS Office or MS Word are likely candidates) closethe application and try to run setup again. In the event that a conflict still exists, close all runningapplications and restart Windows.

2

2 Overview of the Sourcing Analysis andDecision Surface Models

2.1 Overview of the Sourcing Analysis ModelThe Sourcing Analysis Model simulates the retailing process for a line of product, which is offeredover one or several consecutive Selling Cycles of fixed duration, with any leftovers beingliquidated at the end of the final cycle. The simulated activity includes execution of an ordering(sourcing) strategy, interaction of a stream of customers with the shelf stock resulting in sales (orlost sales) and, depending on the scenario, price promotions, and/or markdowns. Manyperformance measures are evaluated.

The model tracks inventory by SKU throughout the Selling Cycle. SKUs are differentiated bystyle, color, and size.

The vendor of product can be modeled in any one of three modes: Perfect Supplier, Modeled byFill Rate, or Detailed Model. In the “Perfect Supplier” mode, any product ordered from the vendoris assumed to arrive on-time (in the lead-time specified) and complete (no shortages). In the“Modeled by Fill Rate” mode, a user-specified percentage of orders arrive “short,” i.e., in-complete. The amount of shortage is user-specified. In the “Detailed Model” mode, the vendormodel includes things like: manufacturing lead time, capacity, quality, shipping time, raw materialinventory and supply, collaboration with the retailer, make-to-order or make-to-stock inventorycontrol, and production plans are specified by the user. This mode allows a more thoroughexamination of the vendor’s ability to respond to the orders placed on them by the retailer.

The program generates a random stream of customers, week by week throughout the SellingCycle. The arrival rate can vary from week to week to reflect seasonality.

When a customer arrives, the program randomly assigns an associated style, color and size fromdiscrete probability distributions that describe consumer preferences. If the desired SKU is instock, one unit is purchased. If the desired SKU is out of stock, the customer may chooseanother specific SKU or leave.

In order to use the simulation model, the user must first define the scenario to be executed. Thisis done by interactively entering (or accepting default values for) a number of data items.

One of the key inputs is the buyer's plan. This consists of the length of the Selling Cycle (up to260 weeks), the anticipated total demand (volume), the number of styles, colors, and sizes in theline, the anticipated percentage of demand (mix) by style, color, and size and the anticipateddemand seasonality (i.e., percent of demand in each week of the season).

An important concept embodied in the model is that of buyer error. This permits the execution ofscenarios in which the buyer's plan is not an accurate estimate of the actual demand for theseason. Error comes in three forms: volume error, mix error, and seasonality error. Volume errorrepresents a difference between the actual demand volume and the plan volume. Mix errorrepresents differences in the anticipated and actual percentages of demand by style, by color, andby size. Seasonality error represents a difference between the planned (i.e., anticipated) andactual percentages of demand in each week. Mix error can change depending on the week withinthe Selling Cycle.

The stocking (and restocking) of the product on the shelves is governed by a sourcing strategy.The model allows the user to select from six strategies. Two of the strategies rely totally on the

3

buyer's plan figures; one permits limited use of point-of-sale (POS) data; while the other threepermit extensive use of POS data to re-estimate demand and place within season reorders.Depending on vendor reliability, the content of deliveries may or may not match the content of theorders and there may be minimum order size requirements. The model also provides break-evenanalysis comparisons between alternative strategies.

Promotions, if any, reduce the selling price for a specified string of weeks while markdownsreduce the selling price until the end of the Selling Cycle (or a subsequent markdown).Promotions are initiated at specified starting weeks; markdowns may be initiated either atspecified starting weeks or by lower than expected sales. Both promotions and markdowns canimpact customer behavior in two ways: by increasing the customer arrival rate and by increasingthe proportion of customers who look for another item when encountering a stockout of theircurrent choice.

While simulating the season, the model calculates a large number of performance measuresrelated to revenues, costs, inventory, and customer service on either a weekly or completescenario basis. These can be viewed in tabular and, for some measures, in graphical format.

2.2 Overview of Decision Surface ModelDecision surface modeling is a powerful tool that allows you to capture and graph the relationshipbetween selected inputs to the Sourcing Analysis model (e.g., lead time, SKU mix error.wholesale cost) and selected performance measures (e.g., service level, gross margin, GMROI).Once a decision surface model is built you can graph any of the selected performance measuresversus any one or two of the selected inputs to see and better understand how the performancechanges with changes in the value(s) of the input(s) without having to actually run the simulation.

In the Sourcing Simulator decision surfaces are modeled using neural networks. Neural networkspermit the modeling of a wide range of complex surfaces without having to guess beforehandwhat the analytic shape of the surface is.

In order to create a decision surface model you first have to make a “batch run” with the SourcingAnalysis Model in order to specify the inputs and performance measures you are interested in andgenerate some data points. In a batch run simulations are executed for all combinations of valuesin ranges that you specify for one or more inputs. Once the batch run is made, you can createdecision surface models relating any of a set of performance measures to any subset of theinputs that were specified in the batch run. Once a decision surface model is created, you mayspecify the range of interest for any one or two of the inputs in the model and graph theperformance measure against the input(s) over the specified range. Up to six graphs, from one orseveral decision surface models, may be displayed side-by-side for easy comparison.

Output files from batch runs and decision surface models may be saved for future use. Graphscan be printed or saved to a file that can be imported into a word processor.

4

3 How to Run the Sourcing Simulator

3.1 Running the ModelsTo start the Sourcing Simulator, double click on the Sourcing Simulator icon. The “MAIN” menu,which is a small window labeled Sourcing Simulator, will open.(Figure 3.1).

Figure 3.1. Sourcing Simulator's Main Menu.

Sourcing Analysis Model

In order to run the Sourcing Analysis Model, click on the Sourcing Analysis Model button. Thisleads you to an interactive Sourcing Analysis (or Data entry) window with six tabs, one for eachtype of data needed to specify a specific scenario (see Figure 4.1).

At this point, with either model, you may:

1. Open a previously saved data file and run.2. Run the simulation using the default data.3. Edit one or more pieces of data and run.

Runs may be for a single scenario or for a batch of related scenarios. In particular,“batch runs” allow you to collect data on how performance measures change as one or moreinputs is (are) varied over a range of values. The results of batch runs may be simply displayed ina table or also used to create a decision surface model to provide graphs of the relationshipsbetween selected inputs and performance measures..After data is entered on the six input tabs, select ‘Run’ or ‘Batch Run ’ under the Simulate drop-down menu.

5

Single Scenario Run

If you wish to simulate only the single scenario specified on the six input tabs, select Run.

After a few moments a small ‘Run Name and Volume Error’ window will open asking you tospecify a name for the simulation run and providing the option to simulate the scenario with morethan one Volume Error level.

The Run Name will be used as an identifier for both the data set and results from the run.

If you accept the default 'Single Value' option for Consumer Demand Volume Error, the scenariowill be simulated with only the default value (same as you entered on the Consumer Demand tab).

Alternatively, to reflect uncertainty as to what the Volume Error actually is, you may specify thatthe scenario be simulated for each Volume Error value within a range and the performancestatistics averaged.

For a 'Uniform Average' specify the Minimum and Maximum values along with thenumber of values (# Runs) within the range for which you want the simulation executed toprovide the average performance statistics. The values associated with each run will begiven equal weight in computing the average.

For a 'Weighted Average' you must also specify the weight (Weight %) that each run'sresults will have in the average. The values associated with each run will be given thespecified weight in computing the average. Be sure that weights total 100.

You may also choose whether to display the statistics for each Volume Error value, alongwith the average statistics (Show All Runs) or only the average statistics (ShowAverage Only)

After entering a name and Volume Error information (or accepting the defaults), click on ‘OK’.

In a few moments a ‘Simulation Results’ window will appear giving a table of performance resultsfor the scenario, labeled with the 'Run Name' you specified. Depending on the Volume Erroroption you chose, the table will contain a single column (Single Value or Average Only) or acolumn for each Volume Error value followed by a column for the average.

You may also view a plot of the weekly values for any one of a number of performance measuresby selecting ‘Graph’ from the Window drop-down menu.

Batch Run

If, after entering the data on the tabs, you wish to continue specifying and execute a batch run,select ‘Batch Run’.

After a few moments a window labeled “Select Inputs to Range” will open displaying a listof inputs which can be varied over a range of values to create the scenarios for the batchrun, along with an initially blank list of the inputs selected (See Figure 3.2).

6

Figure 3.2. Select Inputs to Range Window

In the upper left there is a dialog box labeled ‘Run Name Format.’ The character string inthis box is used as an identifier for each scenario in the batch.

Select and Add the inputs to be included in the batch run, one at a time, specifying theassociated ranges in the small window which opens with each selection.

Then, if you are not planning to use the batch run results to create a decision surfacemodel (DSM), just click on the Run button at the bottom of the window.

If you are planning to use the batch run results to create a decision surface model, clickon the 'Create DSM File' check box and then click on the 'Select Performance Measuresfor DSM' button. A new window will open displaying a list of those measures that may beincluded in a decision surface model. Use the list and Add button to select measures.Selected measures will appear in the table. Once your list is complete, click on OK toreturn to the previous window and then click on the Run button at the bottom of thewindow.

In a few moments after clicking on the Run button, a ‘Simulation Results’ window willopen and the results for each scenario in the batch will appear, column-by-column, asthey become available. The final column will contain the average statistics for the batch.So you can see the progress, the number of runs remaining is displayed at the top of thetable.

If you checked ‘Create DSM File’, the input data for the ranged inputs for each scenarioalong with the values for the selected performance measures is written to a “.dsm” file.When the last run is completed, you will be asked if you would like to go directly to theDecision Surface Model to create a decision surface using the data from the batch run. Ifyou respond ‘yes’, you will be given the chance to save the Simulation Results Table in a“.out” file after which the ‘Model Specification’ window of the Decision Surface Model willopen (see Chapter 6). If you respond ‘no’ you will be able to work with the results table asusual. For example, you may view a plot of the weekly values for any one of a number ofperformance measures for any scenario by selecting ‘New Graph’ from the Tools drop-down menu.

For a detailed description of the required input data along with the data entry procedures, seeChapter 4 on Sourcing Analysis Model Inputs. For a detailed description of the outputperformance results, see Chapter 5 on Sourcing Analysis Model Outputs.

7

Decision Surface Model

In order to create and/or graph a decision surface model, click on the Decision Surface Modelbutton. This leads you to a window labeled Decision Surface Model - Model Specification in whichyou can choose to either open a “.dsm” file from a previous batch run and create a new decisionsurface model or display graphs from a stored decision surface model.

For a detailed description of the process for creating and plotting decision surfaces, see Chapter6.

3.2 Saving and Printing DataAs suggested by options 1-3 above, for the Sourcing Analysis Model you may save andsubsequently access any number of data files (for individual scenarios or batch runs). At any time,to save the data currently displayed in the tabs, pull down the File menu and select theappropriate saving mode. A file ‘Saved As Default’ will become the file automatically opened thenext time the Sourcing Analysis Model is entered from the main menu. Selection of ‘Save As’ willcause the file manager to prompt for a file name. Sourcing Analysis Model data files areautomatically given the extension “.saf” and saved in a folder named “Inputs.” Selection of ‘Save’will prompt for file name if file does not already exist; otherwise the old version will be overwritten.

To open a file other than the default file, select ‘Open’ under the File drop-down menu andrespond to the file manager. The new data will replace the previous in the text boxes.

To obtain a formatted printout of the currently displayed data by select ‘Print’ on the File drop-down menu.

8

4 Sourcing Analysis Model InputsWhen you click on the Sourcing Analysis Model button on the Main menu, you are presented withan interactive Data Entry window (labeled "Sourcing Analysis") with six tabs, one for each type ofdata needed to specify a specific scenario, and text boxes in which to specify the number ofcycles and number of replications to be run. The fundamental format of each tab is a collection ofdata descriptions followed by a text box for data entry. You will note that each text box alreadyhas an entry in it. This is because whenever you enter one of the models in the SourcingSimulator the program opens a default data file.

At this point you may: 1. Edit one or more pieces of data and run. 2. Run the simulation using the default data. 3. Open a previously saved alternative data file and run. Six types (tabs) of data are required:

Buyer’s Plan Cost (and price) Data Markdowns / Promotions Sourcing Strategy Consumer Demand Vendor Specification

YOU SHOULD BEGIN DATA ENTRY/EDIT WITH BUYERS PLAN TAB (Figure 4.1) AND DOTHE VENDOR SPECIFICATION TAB (Figure 4.6) LAST. Other than that, you can move throughthe tabs in any sequence.

Runs may be “single-scenario” or “batch”. Batch runs allow you to collect data on howperformance measures change as one or more inputs is (are) varied over a range of values.Section 4.1 describes data entry for single-scenario runs for the Sourcing Analysis Model. Section4.2 describes data entry for batch runs.

9

Figure 4.1: Buyer's Plan Tab

4.1 Sourcing Analysis Model InputsBefore running the chosen scenario, it is necessary to specify the number of SimulationReplications, which is done in the text box at the top of the Data Entry window.

The number of simulation replications is the number of times the simulation of the of the currentscenario will be repeated.

For example, If you specify the number of replications to be 10, then the SourcingAnalysis Model will execute the simulation 10 times, each with a different set of randomcustomers, and average the performance measures for display in the Results Table. Youcan think of replication as replaying the scenario but with a different customer stream.

Ten replications is the default, and is recommended unless the selling cycle is very long (e.g., ayear), in which case 3-4 replications is sufficient.

4.1.1 Buyer's Plan Tab (Figure 4.1)

This tab allows you to input data on the product line, the duration (in weeks) of the scenario, andthe retail buyer's demand expectation (forecast). Specifically,

Product Name is the name of the product line (e.g., women's cotton blouses).

Number of Weeks in the Selling Cycle is the length of the Selling Cycle in weeks. [Integer]

10

Number of Cycles to Repeat (at the top of the window) is number of consecutive Selling Cyclesin the scenario. The length of the scenario is then Weeks in Selling Cycle times Number ofCycles to Repeat.

For example, if you enter 52 weeks and 1 cycle blank, you have specified a 1-yearscenario. If you enter 52 weeks and 3 cycles, you have specified a 3-year scenario.Similarly, if you enter 20 weeks and 2 cycles, you have specified a scenario with twoback-to-back 20-week Selling Cycles for the same item.

Note: In multi-cycle scenarios the initial inventory levels (See Sourcing Strategy tab) are reset atthe at the beginning of each cycle. Where possible, leftover stock from the previous cycle isused to do this. The total quantity and mix are set by applying the initial inventory option tothe expected demand for the Selling Cycle as estimated at the end of the current SellingCycle. If the leftover stock level for a SKU exceeds the calculated figure, the leftover stocklevel is used. Stock left at the end of the final cycle is liquidated.

The multi-cycle option is intended for scenarios involving basic goods in which case one mightwish to include several consecutive years. One would generally use the multi-cycle option withseasonal or fashion goods.

The Display Floor Space (sq. ft.) is the amount of space allocated to display the product on theselling floor. This number is only used to calculate the output ‘GM/square feet’ and does notaffect the Selling Cycle.

The Planned Number of Units to Sell is the total (over all SKUs) number of units of the productthat the buyer anticipates will sell across (one cycle of) the Selling Cycle (anticipated Selling Cycledemand volume, in other words) [integer]. You may enter either select 'Total' and enter the totaldirectly, or select 'Weekly Average' and enter the anticipated average demand per week.

Individual stock keeping units (SKUs) within the line are distinguished by three characteristics:Style, Color, and Size. Number of Styles, Number of Colors and Number of Sizes,respectively, indicate the number of different styles, colors, and sizes that there are in the productline. [Integers]

NOTE that the total number of SKUs in the line is the product of these three numbers.

The "plan mix" can be specified in either of two ways. The easiest way is to enter the Buyer'sPlanned Percent for Style, Color, and Size. The respective entries represent the buyer'santicipated percentages of total demand associated with each style, color, and size, respectively.

Be careful that each column sums to 100%.

NOTE: The buyer's anticipated percentage of demand for a particular SKU is the productof the percent figures for the corresponding style, color, and size.

Alternatively, you can enter the Buyer's planned percent for each individual SKU. To do so click onIndividual SKU Plan Mix and enter the number of SKUs and associated percentages in thewindow which opens.

Be careful that the entered percents total 100%.

NOTE that the numbered list of SKUs has all of (style1, color1) first, followed by (style1,color2), etc.

11

Selecting Presentation Stock allows you to specify the minimum inventory level of each SKU thatmust be kept for display purposes. There are five options for entering required stock levels:

1. By StyleThis option should be selected if the presentation stock level differs from onestyle to another but is not dependent on either color or size.

2. By ColorThis option should be selected if the presentation stock level differs from onecolor to another but is not dependent on either style or size.

3. By SizeThis option should be selected if the presentation stock level differs from one sizeto another but is not dependent on either style or color.

4. IndividuallyThis option should be selected if the presentation stock level differs from oneSKU to another.

5. Same for All (default)This option should be selected if the presentation stock level is the same for allSKUs.

The Presentation Quantity text box should be used to enter common level foroption 5.For the other options enter the levels in the lists which appear upon selection ofthe option.

The numbered list of SKUs for the 'Individually' option has all of (style1, color1)first, followed by (style1, color2), etc.

NOTE: Options 1-3 can only be chosen if 'Individual SKU Plan Mix' is NOT chosen. Also,if the Model Stock sourcing strategy is chosen, the 'Presentation Stock' and 'ModelStock' levels are the same thing.

The Presumed Seasonality specifies the percent of demand that the buyer anticipates will occurin each week of the Selling Cycle. If you specify more than one cycle, this pattern is repeated ineach.

You may choose to use one of 5 Pre-Defined patterns stored in the Model or indicate that youwant to specify your own user-defined pattern. If you select User-Defined, you will be given theoption of either selecting a previously saved user-defined pattern or defining a new one.

Selected pre-defined patterns are displayed in both graphical and tabular format.

To define a new pattern type, the name of the new pattern in the drop-down box and hit return.You will be asked whether the currently tabled pattern should be cleared. If you wish to entire anentirely different pattern respond 'Yes'. Then enter the percent demand for each week in theSelling Cycle in the cell beside the week number. The total should be 100! As you enter thedata a graph of the pattern will appear in the upper right of the window. Alternatively, you canrespond 'No' and simply replace the values for selected weeks in the currently displayed patternData can also be entered by copying and pasting a column of values from a spreadsheet.

12

User-Defined patterns are automatically saved and can be used later. To delete a user-definedpattern, select the pattern from the pull down menu, and press the delete key or select 'Delete'from the Edit pull down menu at the top of the screen.

4.1.2 Cost Data Tab (Figure 4.2)

This tab allows you to input data on product cost/selling price and operational costs. The data areorganized into three categories: Unit costs/prices, Procurement costs, and Miscellaneous costs.

Figure 4.2: Cost Data Tab

Unit Costs/Price Data

The Sourcing Simulator allows the user to enter product costs and prices in one of five ways.

• Same For All - All SKUs have the same product costs and prices.• Individually – Costs and prices differ by SKU.• By Style - Costs and prices differ by style (only).• By Color -. Costs and prices differ by color (only).• By Size -. Costs and prices differ by size (only).

The type of cost and pricing used is selected in the drop-down list box located to the right of theUnit Costs/Price Data ($/Unit) label.

Initial Wholesale Cost is the amount per unit paid to the vendor for the first shipment[Dollars/unit].

13

Replenishment Wholesale Cost allows the buyer to enter a different amount per unit for unitsthat are reordered within the season. In most cases this will be the same as the initial wholesale[Dollars/unit].

Retail is the base-selling price per unit at which the retailer offers the product [Dollars/unit].

This figure may be adjusted upward (to reflect premiums) or downward (to reflectmarkdowns) during the Selling Cycle. See Section 4.2, Markdowns/Promotions Tab.

Liquidation is the price per unit at which a leftover product can be disposed of (sold to a jobber orsold by drastic markdown) at the end of the scenario. This figure is applied to the total units onthe shelf at the end of the last week of the final Selling Cycle; the result constitutes part of therevenue for the scenario. [Dollars/unit].

The Retail price can be entered either as a dollar figure or as markup percent (of Initial Wholesalecost). Similarly, the Liquidation price can be entered either as a dollar figure or as a markdownpercent (of Retail price).

Procurement Costs

Ordering costs are any additional costs, if any, associated with ordering that have not beenaccounted for anywhere else, such as EDI costs or costs to place and follow up on an order(Buyer specific). [Dollars/Order]

A figure can be entered directly in the text box or computed as the sum of a number ofconstituent costs. To build up the cost from constituent costs, double click on the text boxand enter the constituent costs in the window which opens. If you then save the currentdata as the default, the information will be there for future use.

Initial Fixed Shipping cost is a cost incurred when the initial (beginning of scenario) delivery ofproduct is received by the store. It is represents a cost incurred by the retailer to get product tothe store, and does not depend on the number of units in the shipment, such as shipping via UPS.[Dollars/Order]

Initial Shipping Cost per Unit is a cost incurred, such as import duties, for each unit received inthe initial delivery. [Dollars/Unit/Order]

The two initial shipping costs can be entered directly in the text box or selected from apre-stored list. Double clicking on the associated text boxes will open a window in whichyou can enter a variety of shipping alternatives and their associated costs. If you thensave the current data as the default, the information will be there for future use. You willonly have to click on the shipping alternative in order to enter the costs.

Repl. Fixed Shipping cost is a cost incurred whenever a replenishment order of product isreceived by the store. It is represents a cost incurred by the retailer to get product to the store,and does not depend on the number of units in the shipment, such as shipping via UPS.[Dollars/Order]

Repl. Shipping Cost per Unit is a cost incurred, such as import duties, for each unit received aspart of a replenishment order. [Dollars/Unit/Order]

These two figures can be entered directly in the text box or computed as the sum of anumber of constituent costs. To build up the cost from constituent costs, double click on

14

the text box and enter the constituent costs in the window which opens. If you then savethe current data as the default, the information will be there for future use.

Program Overhead refers to any additional cost incurred in managing special accounts, such ascosts that would be incurred in dealing with overseas vendors. [Dollars]

Miscellaneous Costs

These data should be entered as a percent of the unit Wholesale cost.

Inventory Carrying is the annual charge that the retailer applies to a dollar tied up in inventory.

Handling represents any cost, other than shipping and ordering incurred by the retailer to prepareand place product on the shelf, e.g., cost due to tagging, stocking, etc.

4.1.3 Consumer Demand Tab (Figure 4.3)

This tab allows you to enter data needed to specify the consumer demand, i.e., the random streamof customers and their SKU preferences. You need to enter data on demand volume, demand mix,demand seasonality, and the behavior of customers upon encountering a stockout.

Figure 4.3: Consumer Demand Tab

The Sourcing Simulator was designed to simulate situations where the actual customer demanddiffers from what the Buyer has planned for. Information on what the Buyer planned for wasentered on the Buyer's Plan Tab. The Consumer Demand Tab is where you enter actual demandinformation and thereby create "Buyer plan error" (if any).

15

Consumer Demand (volume) is the "expected" number of customers who actually "enter thestore" during a simulated Selling Cycle.

(The term "expected" is used in the statistical sense. The actual number of customers that enterthe store is based upon a statistical distribution (Poisson) and will vary between runs of thesimulation. The "expected" number refers to the mean of this statistical distribution.)

You have two options for entering this figure:

1. Click on Actual Value and enter the actual expected number of customers. Thisnumber should be entered as either a total value for the Selling Cycle, or as a weeklyaverage, depending on how the planned number of units to sell was specified on theBuyer's Plan Tab. [integer]

2. Click on Percent Error from Plan and enter a positive or negative percent [integer]. Ifyou enter a positive percent, the demand volume will be that percent larger than thebuyer's plan volume. If you enter a negative percent, the demand will be that percentsmaller than the buyer's plan volume.

If a customer encounters an out of stock situation for the selected SKU, the customer will eitheralter their choice or leave the store. The % Who Look for Alternative After Stockout is thepercent who will look to buy a different SKU [integer].

NOTE: The Sourcing Analysis Model makes the assumption that each customer will buy at mostone item. Allowing customers to make more than one purchase must be modeled as two or moreseparate customers by increasing the demand volume.

SKU mix refers to the break up of the overall demand volume into individual SKU preferences.

Based on your input, the Sourcing Analysis Model will calculate the actual (statistical sense again)percent of the customers who will choose each individual SKU.

You have two options for entering the required data:

1. If you click on Actual Values, a button appears that, when clicked, will jump you backto the Buyer's Plan Tab. Here, a new column will appear next to the one containingthe buyer's anticipated percentages for each style, color and size. In each new cell,enter the actual percent of demand you want to simulate. [Integer]

2. If you click on Percent Error from Plan, you will be presented with one text box eachfor style, color and size. Enter the total percent by which styles, colors, and sizes,respectively, differ from the figures in the plan [integer].

The Model will then compute error percents for each individual style, color, and size.For example, if you enter 40, 40, and 20 in the three boxes, you are indicating that thetotal error for styles is 40%, that for colors is 40%, and that for sizes is 20%. If thereare three styles in the line, the Model will randomly create errors for individual styles,such as 10% on style 1, 20 % on style 2, and 10% on style 3, which total 40%.

Selecting Demand Twist allows the actual SKU mix to change over time

Demand Twist Factor is the average amount that the mix percentages can changefrom Demand Twist Period to Demand Twist Period [non-negative number]. TheSourcing Simulator will randomly 'twist' the mix every Demand Twist Period such thatthe average change in mix percentage will be equal to the Demand Twist Factor. The

16

Simulator will also randomly assign a positive (for higher percentage) or negative (forlower percentage) direction associated with each individual adjustment.

The actual adjustment associated with each SKU may be greater or less than theaverage Demand Twist.

Range Of Twist Deviation limits how far the actual adjustments may deviate (+/-)from the average entered.

Demand Twist Period is how often the mix changes (in weeks).

The Actual Seasonality specifies the expected (statistical sense) percent of actual demand ineach week of the Selling Cycle. The Model uses this pattern in generating the random customerstreams.

You may choose to make this pattern identical to that in the buyer's plan by clicking on Same asPresumed Seasonality. Alternatively you may create a "plan seasonality error" by specifying apattern which differs from that in the buyer's plan. In the later case, you may choose to use one ofthe 5 Pre-Defined patterns stored in the Model or indicate that you want to specify your own user-defined pattern. If you select User-Defined, you will be presented with a drop-down list box inwhich you can either select a previously saved user-defined pattern or define a new one.

To select a previously defined pattern choose one on the drop-down box. A graph of the patternwill appear in the upper right of the window and the weekly percents will appear below.

To create and store a new pattern, first replace the current pattern name in the drop-down boxand hit return. You will be asked whether you wish to edit the previous user-defined pattern orclear the entries and enter an entirely new seasonality. After responding, edit or enter the weeklypercents. The total should be 100! A graph of the pattern will appear in the upper right of thewindow. The new pattern will automatically be saved and can be used later.

To delete a saved user-defined pattern, select the pattern from the drop-down box, and thenselect Delete from the edit pull down menu at the top of the screen.

4.1.4 Markdowns / Promotions Tab (Figure 4.4)

This tab allows you to enter data on the extent and timing of promotions, and markdowns within aSelling Cycle, along with information on the impact of price reductions on customer behavior.

17

Figure 4.4: Markdowns/Promotions Tab

The Retail price that you enter on the Cost Data tab is the price that will apply throughout theSelling Cycle if there are no premiums, promotions or markdowns. The extent of premiums,promotions, and markdowns are entered as percentage adjustments in this figure.

Promotions (promotional markdowns) are prescheduled. The number along with timing andextent of each are specified before the Selling Cycle starts.

Number of Planned Promotions is the number of promotions scheduled during a Selling Cycle

If the planned number is greater than zero, then in the Markdown Schedule window below therewill be a row for each promotion for you to enter the following:

Week Occurs is the week of the season when this promotion begins.

Duration is the number of weeks this promotion will run.

Promotions may not overlap. Be careful to make sure that one promotion endsbefore the next one begins. The end of the last promotion should precede thestart of the first Markdown (below)

Markdown % is the percent reduction in Retail price associated with this promotion.

% Who Look for Alternative is the percent of customers who, upon not finding theirselected SKU in stock, switch their choice to another specific SKU.

18

NOTE: This figure may be reset relative to the value originally specified on the ConsumerDemand Tab to reflect the impact of the reduced Retail price in attracting customers (whowould normally just leave the store) to look for an alternative SKU.

Markdowns (for purpose of clearing stock) are also prescheduled. The number along with timingand extent of each markdown are specified before the Selling Cycle starts. However, you maychoose to have the first scheduled markdown begin early if actual sales drop below thoseprojected from the buyer's plan (See Performance Based Markdowns, below).

Number of Planned Markdowns is the number of different markdowns scheduled during theSelling Cycle.

If the planned Number of Markdowns is greater than zero, then in the Markdown Schedulewindow below there will be a row for each markdown for you to enter the following:

Week Occurs is the week of the season when this markdown takes effect.

Once a markdown takes effect, it remains in effect until the next markdown. The numberwhich appears under Duration (weeks) is calculated by the Model. The first markdownshould be scheduled to start after the end of the last promotion.

Markdown % is the percent reduction in Retail price associated with this markdown.

The figure entered should reflect the full adjustment in Retail Price, not the incrementaladjustment relative to earlier markdowns.

% Who Look for Alternative is the percent of customers who, upon not finding their selectedSKU in stock, switch their choice to another specific SKU.

NOTE: This figure may be reset relative to the value originally specified on the ConsumerDemand Tab to reflect the impact of the reduced Retail price in attracting customers (whowould normally just leave the store) to look for an alternative SKU.

If you click on Performance Based Markdowns, you will be asked to enter a ratio(Actual/Projected Sales Ratio) of actual cumulative sales to projected cumulative sales. If,during the simulation in any week beyond the middle of the Selling Cycle but before the beginningweek of the first scheduled markdown, the actual ratio falls below this number, the first scheduledmarkdown will be advanced to the beginning of the next week.

NOTE: This ratio should be entered as a percent between 0 and 100.

Price Elasticity of Demand determines the increase in expected customer arrival rate (demand)following a reduction in the Retail Price of the product [real].

For example, if the price elasticity of demand is specified at 0.7, then for every 1.0 percentdecrease in price, there will be a 0.7 percent increase in the number of customers thatarrive.

Notes: In multi-cycle scenarios, promotion and markdown schedules are repeated each cycle.

Intervals with price premiums may be specified by entering them as a markdown with a negativemarkdown percent. For example, entering a "markdown" starting in week 1 and lasting 3 weekswith a markdown percent of -25 will result in a 25% price markup during weeks 1-3. The rate ofcustomer arrivals will be reduced consistent with the specified elasticity.

19

4.1.5 Sourcing Strategy (Figure 4.5)

This Tab allows you to select the replenishment strategy to be used and enter information aboutthe ordering and delivery of product.

Figure 4.5: Sourcing Strategy Tab

This Tab allows you to select the replenishment strategy to be used and enter information aboutthe ordering and delivery of product.

The Replenishment Strategy drop-down menu allows you to select from six alternativestrategies for determining the content and timing of stock replenishment orders/deliveries after theSelling Cycle has begun. They are Traditional, Vendor Managed Inventory, Quick Response,NewsBoy, Model Stock, and Target Weeks Supply.

a. Traditional

The Traditional strategy involves no re-estimation of demand and no reordering. That part of theBuyer's Plan that is not placed in stock at the beginning of the Selling Cycle is prescheduled fordelivery within season. There is no mechanism to correct for Plan error.

Replenishment Schedule

Replenishment deliveries may be prescheduled to arrive at regular fixed intervals (Fixed Interval)or the user may specify a detailed schedule (User-Defined).

20

For Fixed Interval deliveries, the Number of In-Season Deliveries, the Week of the FirstDelivery, and the number of Weeks between Deliveries [integers] need to be entered.

For example, entering in 6, 1, and 2, respectively, indicates that 6 deliveries will be made,one every other week, beginning at the end of week 1. The nominal size of each deliverywill be the same, equal to one sixth of the buyer's original planned volume less the size ofthe initial shelf stock. The SKU mix in each delivery will be that of the Buyer's Plan.

For User-Defined, the Number of In-Season Deliveries [integer], the Delivery Week [integer],and % to Deliver [real] for each need to be entered.

Initially the Total box contains the percent of the Buyer's Plan volume that is included inthe initial inventory. As the % to deliver is specified in the list, the total is updated. The %figures indicate percent of Buyer's Plan volume. In the end they should total 100%.

b. Vendor Managed Inventory (VMI)

The Vendor Managed Inventory (VMI) sourcing strategy allows a limited amount of demand re-estimation. Initially, the vendor is aware of the buyer's entire forecast (plan), but only ships thespecified initial inventory. The buyer is then allowed to use POS information in placing reorders.The reorders may or may not conform to the original plan. The vendor bases the bulk ofproduction on the original plan (but may accommodate some increase in volume). When thebuyer's modified reorders are placed with the vendor, the vendor will ship according to thesereorders as long as its stock allows. With this strategy, the vendor may not be able to ship all thatis ordered for some SKUs and takes a loss if the buyer has over-estimated the demand for otherSKUs.

Additional In-Season Capacity (see Vendor Requirements) is the percent increase in volumeover that in the original plan that the vendor is willing/able to provide during the Selling Cycle.

Entering 0 means that the vendor will ship at most the number of units for a given SKUthat were specified in the Buyer's Plan. Entering 10, for example, means that the vendoris able to ship up to 10% more of that SKU.


Reorders may be placed at regular intervals over some interval of the Selling Cycle with deliveriesoccurring after a fixed lead time (Fixed Lead Times), or the user may specify a detailed scheduleof orders and deliveries (User-Defined).

For Fixed Lead Times, the Number of Reorders, the Week of the First Reorder, WeeksBetween Reorders, and the length of the fixed Reorder Lead Time need to be specified.[Integers]

For example, if 10, 3, 1, and 2, respectively, are entered, orders will be placed weekly for10 weeks, beginning at the end of the third week of the Selling Cycle, and delivery willoccur at the end of the week, two weeks later.

The nominal size of each order will be the same, equal to one tenth of the buyer's originalplanned volume less the size of the initial shelf stock. The (re-estimated) SKU mix foreach order will be the mix of actual sales up through the week in which the order isplaced.

For User-Defined, the Number of Reorders [integer], followed by the Week to Place Reorder[integer], the associated Delivery Week [integer], and the associated % to Reorder [real] all needto be specified.

21

Note that initially the Total box contains. As the % to reorder is specified for the orders inthe list, the % the total is updated. In the end the total should read 100%.

The size of each order is equal to the specified % of the buyer's original planned volume.The (re-estimated) SKU mix for each order will be the mix of actual sales up through theassociated "week to place reorder".

c. Quick Response

Under the Quick Response (QR) strategy, the demand estimate (forecast) for each SKU isperiodically updated, based on POS data. Reorders are then placed on the vendor. This re-estimation/reordering adjusts for Buyer Plan errors in both demand volume and mix.


Under QR, reorders may be placed at regular intervals over some part of the Selling Cycle withdeliveries occurring after a fixed lead time (Fixed Lead Times), or the user may specify a detailedschedule of reorders and deliveries (User-Defined).

NOTE: If the Initial Inventory Order Method is specified as a Service Target, then theReorder Schedule must be defined by Fixed Lead Times.

For Fixed Lead Times, the Number of Reorders, the Week of the First Reorder, WeeksBetween Reorders, and the length of the fixed Reorder Lead Time all need to be specified.[Integers]


For User-Defined, the Number of Reorders [integer], the Week to Place Reorder [integer], theassociated Delivery Week [integer], and the associated % to Reorder [real], all need to bespecified.

Note: Initially the Total box contains the percent of the Buyer's Plan volume that isincluded in the initial inventory. As the % to reorder is specified for the orders in the list,the total is updated. In the end the total should read 100%.

d. NewsBoy

Under the NewsBoy strategy, the demand estimate (forecast) for each SKU is periodicallyupdated based on POS data and reorders are placed on the vendor. As with the QR strategy, there-estimation/reordering adjusts for Buyer Plan errors in both demand volume and mix. However,in contrast to QR, the order quantity for a SKU is based on providing a (target) customer servicelevel over the interval from the time the order is placed through the week prior to that in which thenext reorder (if any) will be delivered (or the end of the Selling Cycle if no additional orders will beplaced within the cycle).

In Season Service Target (see Vendor Requirements) is used to determine the amount to orderof each SKU at the time of each reorder.

For example, suppose that a reorder for a SKU is being made at the end of week 3 to bedelivered at the end of week 5, the next reorder will be placed at the end of week 4 to bedelivered at the end of week 6, and the target service level is set at 95%. Then, theamount of the SKU ordered at the end of week 3 will be set (taking into account current

22

inventory and outstanding orders) with the goal of meeting at least 95% of the demand forthat SKU in weeks 4-6.

NOTE: If the Initial Inventory Order Method is specified as a Service Target, the twoservice targets need not be the same.


Reorders may be placed at regular intervals over some part of the season with deliveriesoccurring after a fixed lead time (Fixed Lead Times), or the user may specify a detailed scheduleof orders and deliveries (User-Defined).

For Fixed Lead Times, the Number of Reorders, the Week of the First Reorder, WeeksBetween Reorders, and the length of the fixed Reorder Lead Time need specified [integers].

For example, if 10, 3, 1, and 2 are respectively entered, orders will be placed weekly for10 weeks, beginning at the end of the third week of the Selling Cycle, and delivery willoccur at the end of the week, two weeks later.

For User-Defined, the Number of Reorders [integer], followed by the Week to PlaceReorder [integer], and the associated Delivery Week [integer] need to be specified.

e. Model Stock

The Model Stock strategy does not allow for demand re-estimation and is based solely on theBuyer's Plan information. This strategy attempts to have a target level (number of units) for eachSKU on hand after each delivery. This target level is referred to as the Model Stock Quantity, andis entered on the Buyer's Plan Tab in place of the presentation stock quantity. Selection of thisstrategy will produce a button that will jump you back to the Buyer's Plan Tab.

Model Stock quantities may be entered in one of five ways:

1. By StyleThis option should be selected if the model stock level differs from one style toanother but is not dependent on either color or size.

2. By ColorThis option should be selected if the model stock level differs from one color toanother but is not dependent on either style or size.

3. By SizeThis option should be selected if the model stock level differs from one size toanother but is not dependent on either style or color.

4. IndividuallyThis option should be selected if the model stock level differs from one SKU toanother.

5. Same for All (default)This option should be selected if the model stock level is the same for all SKUs.

NOTE: Above options 1-3 can only be chosen if Individual SKU Plan Mix is NOT selected.

The Model Stock Quantity text box should be used to enter common level for option 5.For the other options enter the levels in the lists which appear upon selection of the option.

23

The numbered list of SKUs for the 'Individually' option has all of (style1, color1) first,followed by (style1, color2), etc.


Under Model Stock, reorders are placed at regular intervals over some part of the Selling Cyclewith deliveries occurring after a fixed lead time.

The Number of Reorders, the Week of the First Reorder, Weeks between Reorders, and thelength of the fixed Reorder Lead Time all need to be specified. [Integers]


f. Target Weeks Supply

In the Target Weeks Supply strategy, the demand estimate (forecast) for each SKU is periodicallyupdated based on POS data. This re-estimation/reordering adjusts for Buyer Plan errors in bothdemand volume and mix.

The Target Weeks Supply strategy attempts to order so as to have enough inventory on hand tosatisfy the demand for a fixed number (Target Weeks Supply) of weeks after the shipment is to bereceived. At the time of reorder, the demand for the next 'Target Weeks Supply' plus 'ReorderLead Time' is calculated. The current amount of inventory on hand and any inventory on order(that has not yet been received) is then subtracted and the remaining amount is ordered.

Target Weeks Supply (see Vendor Requirements) is the number of weeks of demand thatshould be in inventory upon receipt of a replenishment order.


Under Target Weeks Supply, replenishment orders are placed at regular intervals over some partof the Selling Cycle with deliveries occurring after a fixed lead time

The Number of Reorders, the Week of the First Reorder, Weeks between Reorders, and thelength of the fixed Reorder Lead Time all need to be specified. [Integers]


Initial Inventory

Order Method specifies the volume of stock that is to be delivered to be on the shelf at thebeginning of the upcoming Selling Cycle It can be entered in a number of ways depending on theReplenishment Strategy chosen. The mix of the initial stocking is based on the current estimateof demand mix. For a single-cycle scenario or the first Selling Cycle of a multi-cycle scenario thisis the Buyer’s Plan.

The five primary options for initial inventory are:

24

A Percentage of the estimated volume for the upcoming Selling Cycle.

A specific Quantity (i.e., number of units of product).

A Safety Stock Percentage. When you choose this option the Model computes theexpected demand until the first replenishment order in the upcoming Selling Cycle isreceived and adds to this the specified (safety stock) percentage.

For example, if the first replenishment is due at the beginning of week 4 and theyou expect to sell 100 units in the first three weeks, then a safety stockpercentage of 25% will result in an initial inventory of 125 units.

A Service Target. When you choose this option the Model uses the expected demandvolume, mix, and seasonality to compute the initial stock level with the objective ofproviding the specified service level until the first replenishment order is delivered in theupcoming season. The service level obtained from the simulation may be higher or lowerthan the Service Target entered due to differences between the estimated and actualconsumer demand.

A number of Weeks Supply. (based on the expected demand volume, mix, andseasonality).

Service Target can be used only with the Quick Response, Newsboy, and Target Weeks Supplystrategies. The Model Stock strategy uses none of these options; the initial stock is the modelstock.

If you specified a 'Presentation Stock' on the Buyer's Plan tab, Presentation Stock becomes anoption for Order Method for Initial Inventory too. If you select it, the inventory level at thebeginning of the Selling Cycle will be the presentation stock you specified on the Buyer's Plan tab.The total number of units and associated percentage of Plan volume will be shown (in gray).

Weeks Prior to Season Placed and Initial Order Lead Time specify the timing (number ofweeks in advance of the first Selling Cycle and delivery lead time for the initial (inventory) order,respectively.

For example, specifying 10 and 6, respectively, results in the initial order being placed 10weeks before the beginning of the selling cycle with delivery 4 weeks before the beginningof the cycle

These two inputs are only important if you are using the Detailed Manufacturer Model (VendorSpecification tab).

Vendor Requirements

Minimum Order Quantity per SKU is the smallest number of units of each SKU that the vendoris willing to ship. For example, the vendor may ship a single unit of each SKU, or the vendor mayonly ship SKUs in packages of five.

Minimum Order Quantity per Order is the smallest number of units that the vendor is willing toship.

4.1.6 Vendor Specification (Figure 4.6)

In the Sourcing Analysis Model the vendor of product can be modeled either as a "black box"which simply responds to orders (more or less as placed) or a manufacturer which orders raw

25

material, creates and executes a production schedule, and responds to the retailer's orders undervarious "partnering" arrangements.

Note: The terms Vendor and Manufacturer are used interchangeably.

This tab is used to define the characteristics of the vendor. The inputs on this tab depend uponthe Manufacturer Model selected. There are three choices.

Figure 4.6: Vendor Specification Tab

Manufacturer Model

In the "black box" mode, with each of the retail replenishment strategies, the amount actuallydelivered may be the same or less than the amount ordered or prescheduled for delivery. Thereare two options:

Selecting Perfect Supplier means that the vendor will always ship exactly what is orderedor prescheduled in the lead time specified on the Sourcing Strategy tab.

In contrast, Fill Rate Model is used to model a vendor that sometimes short ships.

The % of Orders Short indicates the percentage of replenishment orders that areshipped incomplete.The Range of Shortage indicates the typical range of the amount by which shortorders are short [percent].

26

For example, if you specify 20 for '% of orders short' and 5-10 for the 'range ofshortage', then the Simulator will randomly select about 20% of the season'sreplenishment orders to be short and, for each of these, it will randomly select theamount of shortage to be between 5% and 10%.

The units that are short are generated randomly with more popular SKUs having agreater potential for being short than less popular SKUs.

With either of these options the only other input is the way in which the raw MaterialSupply is differentiated. The choices are no differentiation ('Same For All'), different foreach SKU ('By individual SKU'), or different for each style, color, or size of the product. Inthe "black box" mode this will be used only in determining what is displayed in the RawMaterial Mix graph.

Selecting Detailed Model opens a number of dialog boxes in which you specify detailedcharacteristics of the vendor's manufacturing operation. The inputs associated with this choiceare given below.

Raw Material Supply

Sourcing Simulator allows you to model the supply of a primary raw material. The inputs in thisgroup are used to determine the characteristics of this raw material and its supply.

Material Supply defines the number of different raw material types for this line of products. Adrop down list of options include:

• Same for All – all SKUs in this product line use the same raw material type• by Individual SKU – each SKU in the product line has a unique raw material type• by Style – each Style has a unique raw material type, i.e., all SKUs in a given Style require the

same raw material type• by Color - each Color has a unique raw material type, i.e., all SKUs in a given Color require

the same raw material type• by Size - each Size has a unique raw material type, i.e., all SKUs in a given Size require the

same raw material type

Min Total Order Quantity determines the minimum order size of a raw material order over all rawmaterial types. If an order is generated for less than this minimum it will be increased up to theminimum.

Min Order Quantity is the minimum order size for a raw material type. Depending upon theMaterial Supply above, this may be by SKU, by Style, by Color or by Size. If “Same for All” isselected for Material Supply then this input is not used.

Lead Time (Weeks) is the number of weeks required from order to receipt of the raw material atthe manufacturer.

Cost Data (per unit)

Here you to enter various cost data for the manufacturer. In particular,

Material Wholesale Cost is the cost per unit for the manufacturer to purchase raw material fromits supplier.

Inventory Carrying Cost is the annualized cost for carrying inventory as a percentage of thewholesale cost. This is applied to both raw material and finished goods inventories at themanufacturer.

27

FG Liquidation Price is the value per unit of residual finished goods inventory at the end of thefinal cycle of the scenario. For example, this could be the amount per unit that could be obtainedby selling any residual units on the open market.

Material Liquidation Price is the value per unit of residual raw material inventory at the end ofthe final cycle of the scenario. For example, this could be the amount per unit that could beobtained by selling any residual units on the open market.

Production Cost is the manufacturer’s cost to produce a finished good unit exclusive of rawmaterial. This can include labor, equipment, etc.

Inventory Policy

With this group of inputs you define how both raw material and finished inventories are managed.

Ship Backorders is used to indicate if the retailer allows backorders, i.e., if the manufacturer isnot able to meet an order, does the manufacturer allow the manufacturer to ship the missingitems in some subsequent shipment. The check box can be clicked to toggle between allowingbackorders (check mark is visible) and not.

Ship Residual FG Units at Season End is used to indicate if residual finished goods units at themanufacturer at the end of the final selling cycle in the scenario are shipped to in the lastshipment to retail. If not, then these residual units are liquidated at the FG Liquidation Price. Thecheck box can be clicked to toggle between allowing the residual units to be shipped to the retailer(check mark is visible) and not.

Collaboration Policy defines the degree of information sharing between the retailer and themanufacturer. There are two options in the drop-down list' Plan and Orders Only' and 'Plan andPOS Data'.

Plan and Orders Only means that the retailer will communicate the preseason plandetails, i.e., Planned Volume, Assortment SKU Mix, Seasonality, etc. from the Buyer’sPlan tab and the replenishment plan from the Sourcing Strategy tab to the manufacturer.The only other information shared is the orders that are made during the season (if any).

With Plan and POS Data, the retailer shares the same information as with 'Plan andOrders Only' plus the retailer’s latest re-estimation of the remaining needs for the rest ofthe season. This allows the manufacturer to have a much more clear idea of what theyneed to be producing so that they can better anticipate the reorders from the retailer.

FG Inventory Policy is a drop down list of two choices: 'Make to Order' and 'Make to Stock'.

Make to Order means that the manufacturer will not carry finished goods inventory and willtry to produce exactly what is ordered.

Make to Stock means that the manufacturer will carry finished goods inventory. If thisoption is selected then the following inputs will pop up to allow you to specify the target levelfor finished goods inventory, FG Supply. FG Supply can either be expressed in Units or inOrders as defined by the FG Units.

If Units is checked then the FG Supply is the number of units desired.

28

If Orders is checked then the FG Supply is the number of average-sized ordersdesired. In this case, the number of units will be FG Supply times the average ordersize from the retailer.

Manufacturing Process Data

You use this group of inputs to specify key parameters about the manufacturing process thatproduces the finished goods.

Process Time (Weeks) is the number of weeks required to produce a shop order, i.e., theamount of time required to make product.

Shipping Time (Weeks) is the time in weeks to ship product from the manufacturer to the retailerand get it onto the retail shelves.

Weekly Capacity (Units) is the maximum number of units that can be produced in one week.

Quality Dropout Rate (%) is the percent of product in a shop order that will be deemed ofunacceptable quality and thus not be available to ship to the retailer.

Raw Material per FG Unit is the number of raw material units required to produce one finishedgood unit.

Plan Production (Button)

This button is used to define when and how much the manufacturer plans to produce each weekin terms of finished goods and order in terms of raw material. It is important to understand thatthis only used as a plan and the actual production may differ. The purpose is to determine anypre-season material orders and production as well as when material orders and shop orderreleases are planned during the season.

When the Plan Production button is clicked a Manufacturing Plan Spreadsheet (see Figure 4.7)will pop up.

29

Figure 4.7 Manufacturing Plan Spreadsheet

The spreadsheet has the following 10 columns:

Week – This represents the week number during the selling cycle where 1 represents the firstweek of the cycle. Negative numbers represents weeks prior to the selling cycle.

Expected Demand – This column shows the retailer’s plan in terms of the number of units theybelieve will sell during each week of the selling cycle.

Required Shipments – This represents the number of units required to be shipped in each weekaccording to the retailers Buyer’s Plan and Sourcing Strategy.

Finished Goods Inventory – This represents the number of finished goods units in inventoryeach week.

Completed WIP – This represents the number of units each week that are planned to beconverted from Work In Process (WIP) into finished goods.

Shop Order Release – This is the number of units to start production each week, i.e., the amountreleased to the shop floor for conversion from raw material into finished goods.

Cumulative Material Reqm’ts – This is the cumulative number of units of raw material unitsrequired to be able to produce the number of units in the Shop Order Release column.

Material Inventory – This is the number of units of raw material in inventory each week.

30

Material Delivery – This is the number of units of raw material to be delivered from the suppliereach week.

Material Order – This is the number of units of raw material ordered from the supplier each week.

The values in the first three columns (Week, Expected Demand and Required Shipments) aredetermined by the values in the Buyer’s Plan and Sourcing Strategy tabs. You, the user, mustinput the Shop Order Release and Material Order values. The other 5 columns (Finished GoodsInventory, Completed WIP, Cumulative Material Reqm’ts, and Material Inventory, MaterialDelivery) are determined by the values you input in the Shop Order Release and Material Ordercolumns and the Required Shipments columns. The relationships are described below.

The idea is that you look at the required shipments and decide when to produce product in orderto meet the requirement. If you want to release a shop order in some week then the number ofunits must be least the Minimum Order Quantity per Order specified on the Sourcing Strategy Taband at most the Weekly Capacity value specified in the Vendor Requirements Tab. If you enter avalue outside this range, the program will automatically decrease (or increase) the figure to be inrange. Notice also that some of the rows of these columns have a white background while othershave blue. The white cells indicate the weeks when Shop Order Releases or Material Orders canbe entered. For example, if the last shipment to retail is week 10 and the Processing Time is 2weeks, then any shop order releases after week 8 could not possibly be shipped to the retailer.Also, raw material cannot be ordered until the manufacturer knows the retailer’s first order. Theweek that this is made is specified on the Sourcing Strategy Tab in the Initial Inventory section.Similarly, a shop order cannot be made before the earliest raw material could be available.

The units associated with a Shop Order Release value entered for some week will completeproduction (appear as Completed WIP) 'Processing Time' weeks later where Processing Time isas specified on the Vendor Specification Tab. The number of units will be reduced by the QualityDropout Rate (%). For example, assume 500 units is entered in the Shop Order Release columnfor week 1. Assume also that the Processing Time is 2 weeks, Quality Dropout Rate is 10% andthe Raw Material per FG Unit is 2. In this case 90%x500=450 units will appear in the CompletedWIP column in week 1+2=3. In addition, the Finished Goods Inventory value for week 3 will beincreased by 450 units. Also, the Cumulative Material Reqm’ts value will be increased by500*2=1000 reflecting that 1000 raw material units would be needed in order to produce 500units. Finally, the Material Inventory value for week 1 will be reduced by 1000 units.

Negative Finished Inventory values are shown in red to indicate a deficiency between the requiredshipments and amount produce thus far. This is a visual cue to indicate that shop order releasesneed to made (or adjusted) to eliminate the deficit.

Once all the Shop Orders are specified, you can insert values in the Order Material column to besure that enough raw material is ordered to meet the Shop Order values. For example, if theLead Time (Weeks) value on the Vendor Specification Tab is 3, then a value entered in theMaterial Order column in week 1 will be show up as the value in the Material Delivery column inweek 1+3=4. This amount will also be added to the Material Inventory value in week 4.

As with the Finished Goods Inventory, negative Material Inventory values are shown in red toindicate a deficiency between the required shipments and amount produce thus far. This is avisual cue to indicate that shop order releases need to made (or adjusted) to eliminate the deficit.

When the Manufacturing Plan spreadsheet is active, there are two menubar options that can aidthe process of specifying Shop Order Releases and Raw Material Orders. First, on the Edit drop-down menu there is a Clear option that provides three choices: Shop Order Releases; MaterialOrders, and Entire Plan. These are used to clear out the values in the Shop Order Releasecolumn, Material Order column, or both, respectively. There is also a Refresh option that willrecalculate all values in the table based upon current parameters

31

In addition, on the Tools drop-down menu there is an Order Planner option. This tool is availableto facilitate filling in the Shop Order Releases and Material Orders. If Order Planner is selected,an Order Planner window will open as shown in Figure 4.8. Under Options, there are two choices:Pre- and In-Season and Entire Season. With the first choice you can separately plan Pre-Seasonproduction and material orders from those In-season (as shown in the Figure). Entire Seasonallows you to plan these together. (Since these are very similar, only the former will be described.Use of the Entire Season option is very similar.)

Figure 4.8 Order Planner Window

This Order Planner window has four sections:

1) Pre-Season Shop Order Releases2) In-Season Shop Order Releases3) Pre-Season Raw Material Orders4) In-Season Raw Material Orders

Each of these sections has a drop-down list with four options:

1) Manual2) Fixed Number of Orders3) Fixed Order Size4) Backward Load based on Capacity

The use of these options is as follows:

Manual – This is used if you intend to enter orders directly into the spreadsheet.

Fixed Number of Orders – This is used if you want to specify the number orders during theperiod of interest (Pre-Season or In-Season). When you select this choice, dialog boxes for thefollowing three inputs will appear:

32

1) Number of Releases (Number of Orders, if a Material Order) – The number of orders tobe released during the period of interest.

2) Start Week – The week (prior to or within the first selling cycle) for the first order release.3) Weeks Between Releases (Weeks Between Orders, if a Material Order) – The number

of weeks between releases.

For example, if you enter 3 releases starting in week –7 with 2 weeks between releases then arelease will be made in weeks –7, -5, -3, and –1.

Note: with this option the size of Shop Order Releases is allowed to exceed the weeklymanufacturing capacity.

Fixed Order Size – This is used if you want the system to schedule enough releases of a givensize to fulfill the requirements. When you select this choice dialog boxes for the following inputswill appear:

1) Order Size – The size in units of the releases during the period of interest.2) Start Week – The week (prior to or within the first selling cycle) for the first order release.3) Weeks Between Releases (Weeks Between Orders, if a Material Order) – The number

of weeks between releases.

Backward Load based on Capacity – This option will delay the release of the (shop or material)order as late as possible to meet the requirements. With this option the size Shop Order Releaseswill be within the specified weekly manufacturing capacity.

For example, consider the Pre-Season Shop Order Releases where the required pre-seasonshipment to the retailer are 2000 units in week -1, the Weekly Capacity is 1500 units, and theProcessing Time is 2 weeks. In this case, the latest week to releases a shop order and have itavailable for shipping by week –1 is week –3 since the lead time is 2 weeks. However, thisrelease could not be for 2000 units since the Weekly Capacity is 1500 units. Therefore, theremaining 500 units would have to be released in week –4.

Note that at the bottom of the Shop Order Release and Material Order columns a total is given asa guide to indicate the number of units ordered based the values above. This can help you decideif the shop orders and/or raw material orders are appropriate.

4.2 Batch Run InputsThe data entry process for a batch run is the same as that for a single-scenario run except thatafter entering (accepting defaults) for a single scenario, you select one or more of the inputs for“ranging” and for each of these specify a set of values within a range.

The set of inputs which can be ranged is restricted to those the list which appears in ‘Select Inputsto Range’ window which opens when you select Batch Run on the Simulate drop-down menu(See Figure 4.9).

33

Figure 4.9. Select Inputs to Range Window

Inputs for the batch run are specified one-at-a-time. For each, first select the input on the list andclick on the Add button. A small 'Input Information' window will open asking for the Minimum andMaximum values of the input to include along with the Increment, or step-size, in between.

For example, entering (30, 60, 10) for ‘Initial Inventory Percentage’ will specify initial stocking percentages of 30, 40, 50, and 60.

The value of the input originally specified on the corresponding input tab will initially appear in the'Min.' box. You may change this if it is not the smallest value you want to consider. After enteringthe three values click on OK. The selected input along with the range information will appear inthe list of selected inputs.

During the batch run a scenario for every combination specified input values will be simulated. Asyou add inputs to the list, the size of the batch run is computed and displayed in the upper right inthe 'Number of Runs' box.

If the run size seems to big or you change your mind about including some of the selected inputsyou may remove inputs from the list by selecting them and clicking on the Delete button.

The Run Name Format character string is used to provide an identifier for each scenario in thebatch. The string format is 2 letters to indicate the sourcing strategy followed by, for each rangedinput (in the sequence in which the inputs were checked), one letter to indicate the input variablefollowed by the value of the input. For example, “QR V10 I40 L2” indicates that in this run the QRsourcing strategy was used, the Volume error was 10%, the Initial inventory percentage was 40%,and the reorder Lead time was 2 weeks.

After specifying the inputs and ranges, if you are not planning to use the batch run results tocreate a decision surface model (DSM), just click on the Run button at the bottom of the window.

If you are planning to use the batch run results to create a decision surface model,click on the Create DSM File check box in the lower left and then click on the 'Select PerformanceMeasures for DSM' button. A new window will open displaying a list of those measures which may

34

be included in a decision surface model. Use the list and Add button to select measures one-at-a-time. To include all measures, click on the Add All button. Selected measures will appear in thetable. As with the input list, unwanted entries may be deleted. Once your list is complete, click onOK to return to the previous window and then click on the Run button at the bottom of the window.

Clicking on the 'Create DSM File' checkbox will cause the input data for the ranged inputs for eachscenario along with the associated output values for the selected performance measures to bewritten to a “.dsm” file (in a directory named ‘Batch Run Data’) for use in creating a DecisionSurface Model.

NOTES: You should design batch runs carefully:

1. With a batch run the Simulator will execute the scenarios, one after another, for everycombination of values specified for the ranged inputs. Thus, for example, if you select 3inputs for ranging and specify 5 values for each, the Simulator will execute 125 (5x5x5)scenarios. The maximum number of scenarios that can be accommodated in a single batchrun is 399. As you select inputs and specify ranges and increments the implied total is talliedin the Total Number of Runs box. If you exceed 399, it will turn red.

2. In specifying ranges for inputs one at a time, it is possible to create scenarios that are

infeasible. For example, if you select ‘Number of Reorders’ and ‘Reorder Lead Time’ forranging with a Quick Response sourcing strategy for a 10-week season and specify (3,8,1)and (1,3,1) as the data, the set of implied scenarios includes one with 8 reorders and a 3week lead time, which can’t work.

3. In determining the values of inputs to include within the range, the Simulator adds the

‘Increment’ value that you entered to the ‘Min’ value once, twice, three times, etc. until the‘Max’ is reached. Depending on the values entered, it is possible that the ‘Max’ will not beincluded. For example, if you enter (1,4,2) for ‘Reorder Lead Time’, the Simulator will executescenarios with lead times of 1 and 3, but not 4.

35

5 Sourcing Analysis Model OutputsThe Sourcing Analysis Model provides the results from simulated scenarios in two formats,

1. A table entitled Simulation Results.2. Graphs of weekly values of certain performance measures.

5.1 Simulation Results Table (Figure 5.1)After completing a simulation run (single scenario or batch), the Sourcing Simulator automaticallyopens a window and displays a table of performance results. If the number of replicationsspecified is more than one, then the values in the table are the averages over all replications.

Figure 5.1: Simulation Results Table

Alternatively, from the data entry window you can select 'View Results' on the Simulate drop-downto open an empty Simulation Results Table and then open a stored file of previous simulationresults by selecting 'Open' on the File drop-down menu.

You may tailor the contents of the table to focus on particular measures of interest to you. Untilyou do so, a full set of measures will be displayed.

5.1.1 Output Statistics

36

The Simulation Results Table has two sections. The upper section contains output statisticspertaining to the retailer. If you used the detailed manufacturer model, the lower part of the tablecontains statistics pertaining to the manufacturing operation.

The two sets of output statistics for the Sourcing Analysis Model are as follows:

5.1.1.1 Retail Output Statistics

Customers is the average actual number of customers per Selling Cycle who entered the store.(Recall that the Sourcing Simulator randomly generates customers based on input information youprovide the simulation.)

Inventory Measures

Initial Inventory is the average total number of units that were in the store at the beginning of thefirst week of a Selling Cycle.

Replenishment Inventory is the average total number of units that were delivered to the storeafter the beginning of a Selling Cycle (based on reorders or prescheduled drops).

Total Offering is the average total number of units that were delivered to the store for a SellingCycle, i.e., the sum of Initial Inventory and Replenishment Inventory.

Average Inventory is the average number of units in the store each week.

Inventory Turns is the average number of times the stock "turned over" during a Selling Cycle.

It is calculated as Total Offering divided by Average Inventory.

Sales Measures

% of Offering Sold is the percent of all units offered (i.e., delivered to the store) that were sold.

% Liquidated is the percent of all units offered that remained on the shelf at the end of the finalselling cycle of the scenario (and were then sold off at the liquidation price you entered).

% Liquidated plus % of Offering Sold should total 100%

% Sell Thru is the percent of units offered that were sold at retail price (as opposed to a markeddown/up price).

In Stock Measures

In Stock % is the average percent of SKUs that were in stock each week.

IS % before 1st RO Receipt is the average percent of SKUs that were in stock until the firstreorder was received.

IS % before 1st Markdown is the average percent of SKUs that were in stock before thebeginning of the week when the first markdown began.

Lost Sales Measures

37

Lost Sales (Overall) is the percent of customers who left the store after encountering an out-of-stock situation.

LS before 1st RO Receipt is the Lost Sales before the first reorder was received [percent].

LS before 1st Markdown is the Lost Sales before the beginning of the week when the firstmarkdown began.

Service Level

Service Level (Overall) is the percent of customers who found (and thus purchased) their firstchoice of SKU. (Customers who purchased an alternative SKU after encountering a stockout arenot included).

SL before 1st RO Receipt is the Service Level before the first reorder was received.

SL before 1st Markdown is the Service Level before the beginning of the week when the firstmarkdown began.

Revenues

Sales Revenue is the total revenue generated by sales to customers during the average sellingcycle.

Liquidation Revenue is the revenue generated from all units that were sold off (at the liquidationprice) at the end of the final selling cycle.

Total Revenue is the sum of the Sales Revenue and Liquidation Revenue.

Revenue/Unit Offered is the Total Revenue divided by the Total Offering.

Costs

Cost of Goods is the average total cost (for all goods offered) per Selling Cycle.

It is calculated as Initial Wholesale cost (weighted average of figures entered by user)times the Initial Inventory plus Replenishment Wholesale cost (weighted average offigures entered by user) times Replenishment Inventory.

Ordering Cost is the average cost per Selling Cycle to the retailer for having placed and followedup on all orders.

It is calculated as Ordering Cost (entered by user) times the average number of ordersplaced per Selling Cycle.

Handling Cost is the average cost per Selling Cycle to the retailer for having product preparedand displayed once they were in the store.

It is calculated as the Handling Cost percent (entered by the user) times Cost of Goods.

Shipping cost is the average cost per Selling Cycle to the retailer for having shipped to the store,e.g., from a distribution center.

It is calculated as Initial Fixed Shipping Cost (entered by user) plus the Repl. FixedShipping Cost (entered by user) times the number of replenishment deliveries received

38

plus Initial Shipping Cost/Unit (entered by user) times Initial Inventory plus Repl. ShippingCost/Unit (entered by user) times Replenishment Inventory.

Inv. Carrying Cost is the average cost per Selling Cycle of holding inventory.

It is calculated Average Inventory times (weighted average) Wholesale cost times theInventory Carrying charge (entered by user).

Program Overhead (entered by user) is the average cost per Selling Cycle to manage theprogram.

Margins

Gross Margin is the Total Revenue less the Cost of Goods.

GM / Sq. Foot is the ratio of the Gross Margin to Display Floor Space (entered by user).

GM % is the ratio of the Gross Margin to the Total Revenue.

% of GM Potential is a ratio of the actual Gross Margin to the potential gross margin that wouldhave accrued had all units offered sold at Retail Price (entered by user).

Adjusted Gross Margin is the Gross Margin less the Inventory Carrying Cost, Handling Cost,Shipping Cost, Ordering Cost, and Program Overhead.

GMROI is the Gross Margin divided by the average inventory investment during the season.

Average inventory investment is calculated as Average Inventory times the (weightedaverage) Wholesale Cost of the product.

GMROISL is calculated as GMROI times the In-Stock %.

5.1.1.2 Mfg. Output Statistics

Raw Material Inventory Measures

Residual Units is the number of units of raw material not consumed, i.e., the number of unitsleftover after the last shipment to retail.

Units Ordered is the number of units of raw material ordered from the supplier.

Average Inventory (Raw Material) is the average number of units of raw material held by themanufacturer from the time of the first receipt of raw material until the last shipment of finishedgoods to retail.

Inventory Turns (Raw Material) is the number of times the raw material inventory turns overfrom the time of the first receipt of raw material until the last shipment of finished goods to retail.It is computed as the total number of units received divided by the Average Inventory (RawMaterial) divided by the number of selling cycles.

Finished Goods Inventory Measures

Units Demanded is the total number of units of product ordered by the retailer.

39

Units Produced is the total number of units of product actually produced by the manufacturer.

Units Shipped is the total number of units of product actually shipped from the manufacturer tothe retailer.

Units Backordered is the total number of units of product ordered by the retailer but not shippedby the manufacturer on time.

Residual Units is the number of units of product produced and left over after the last shipment toretail.

Average Inventory (Finished goods) is the average number of units of product carried infinished goods inventory at the manufacturer.

Inventory Turns (Finished goods) is the number of times the finished goods (product) inventory“turned over” from the time of the first receipt of raw material until the last shipment of finishedgoods to retail. It is calculated as the Units Produced divided by the Average Inventory (FinishedGoods) divided by the number of selling cycles.

Service Measures

% Shipped On-Time is the percentage of product that was shipped to the retailer on-time, i.e.,within the lead-time specified.

% Backordered is the percentage of all product ordered by the retailer that was not shipped on-time.

% of Backorders Filled is the percentage of all backordered units that was eventually shipped tothe retailer.

Costs

Raw Material Costs is the total cost of raw material ordered by the manufacturer. It is calculatedas the Units Ordered times the Raw Material Cost.

Inventory Carrying Costs is the sum of the raw material and finished goods inventory carryingcosts. Each is calculated as the average inventory (raw material or finished good) times theannual Inventory Carrying Cost Rate divided by the number of weeks from the first receipt of rawmaterial until the last shipment of finished goods to retail.

Production Cost is the cost of producing product. It is calculated as the Units Produced timesthe Production Cost.

Revenues

Revenue is the revenue received from the sale of product to the retailer. It is calculated as thenumber of units shipped for initial inventory times the Initial Wholesale Cost plus the number ofunits of replenishment shipped times the Replenishment Wholesale Cost.

Liquidation Revenue is the revenue generated by liquidating both residual raw material andresidual product. It is calculated as the Residual Units of raw material times the Raw MaterialLiquidation Price plus the Residual Units of Finished Goods times the Finished Goods LiquidationPrice.

Total Revenue is the sum of the Revenue and Liquidation Revenue.

40

Revenue/Unit Produced is Total Revenue divided by the number of Units Produced.

Margins

Gross Margin is Total Revenue minus Raw Material Costs and Production Cost.

Adjusted Gross Margin is the Gross Margin minus Inventory Carrying Costs.

5.1.2 Displaying Results from Several Scenarios

The table allows side-by-side display (in different columns) of the results from a number ofdifferent single scenarios or from a batch run (see Figure 5.1).

With batch run results you may choose to display the results from each individual scenario withinthe batch, the average scenario results, or both. To select the desired option, click on Batch Runon the View drop-down menu. The default is “Both”.

Selecting 'Sourcing Analysis' on the Window drop-down menu allows you to return to the SourcingAnalysis Model (Data entry) window without closing the Simulation Results window, create andrun a new scenario, and have the results appear in the first open column of the table. To return tothe inputs associated with a specific scenario (for which the results are currently displayed) selectthe corresponding column in the table and click on Reload Scenario on the Edit drop-down menu.

Opening a file of stored simulation results while the results of one or more scenarios are currentlydisplayed in the Simulation Results Table will cause the stored results to appear in first emptycolumn(s) of the table.

5.1.3 Editing and Tailoring the Results Table

The look and contents of the Simulation Results Table may be altered by using the Edit and/orView drop-down menus.

When the Simulation Results Table opens, both the (upper) retail and (lower) manufacturingstatistics sections are displayed. To view the manufacturing statistics you may raise the dividerbetween the two sections. Alternatively, you may hide the retail statistics (and display themanufacturing statistics) section by selecting 'Retail Outputs/Hide Grid' on the View drop-downmenu. To redisplay the retail statistics section (along with the manufacturing results section)either de-select 'Retail Outputs/Hide Grid' or select 'Retail Outputs/Reset Outputs'' on the Viewdrop-down menu. Corresponding actions can be initiated by selecting 'Mfg Outputs' on the Viewdrop-down menu.

The name you specified before each simulation run is used to label the corresponding column ofresults. If you wish, you can re-label a scenario by either double clicking on the cell displaying thecurrent label, or by selecting 'Change Column Label' on the Edit drop-down menu. Scenariolabels may not be longer than 20 characters.

You may also switch pairs of columns in the table by double clicking on the label of one of the pairand then selecting the other from the drop-down menu that appears.

Individual scenario results may be deleted from the table by selecting (highlighting) the column inthe table and then choosing Delete on the Edit drop-down menu

If the table contains measures which are not of interest to you, individual rows (includingsubsection headers such as INVENTORY) may be hidden. To hide rows in the retail(manufacturing) statistics section select 'Retail (Mfg.) Outputs' then 'Customize' on the View drop-down menu.

41

Selecting 'Retail Outputs/Customize' opens a Customize Retail Outputs View window with all therow labels from the retail section listed as "Visible Outputs." Similarly, electing 'Mfg.Outputs/Customize' opens a Customize Mfg Outputs View window with all of the row labels fromthe manufacturing section listed as "Visible Outputs".

To hide an individual row, select its label and then click on "<". The label will be moved to the"Hidden Outputs" list and the row will no longer appear in the simulation Results Table. To restorean individual row, select its label and click on ">". The label will return to the "Visible Outputs" listand the row will reappear in the Simulation Results Table.

Alternatively, rows may be hidden by selecting (marking) them in the Simulation ResultsTable itself and clicking on 'Hide Rows' on the Edit pull-down menu.

All visible rows in the section may be simultaneously hidden by clicking on "<<". Similarly, allhidden rows in a section may be simultaneously restored by clicking on ">>".

Alternatively, all hidden rows in a section may be restored by selecting 'Reset Outputs' onthe respective View submenu.

You may create your own table by clearing all rows and then restoring the individual rows that youwant.

NOTE: the edited set of rows (reduced or expanded) becomes the default display set for theModel that you are currently running for this and future sessions (until you make additional edits).

5.1.4 Saving and Printing (Tabular) Simulation Results

The results for any or all of the scenarios currently displayed in the Results Table may be savedas either a system file which can be reopened in Simulation Results Window at a later time or asa text file which can be imported into a word processor.

Selecting 'Save As' on the File drop-down opens a window displaying the labels of all columns(scenarios) in the table and asking you to specify the name for a '.out' file. Note that all columnlabels are "selected" (highlighted in blue). To save the full table of results, provide a file name andclick on 'OK'. The saved '.out' file will contain the tabular results, the associated inputs, and theassociated graph data. It can be opened at a later time from the File drop-down menu when theSimulation Results Window is open.

To save only chosen scenario results, deselect the scenarios which are not wanted by clicking onthem, provide the file name, and click on 'OK'.

To save tabular results (only) in a text file, click on Save As on the File drop-down menu. Thenchoose 'Text File' on the Save File as Type drop-down menu and proceed as above to specify thescenarios to be included and name the '.txt' file.

To print the Results Table, select 'Print Output' on the File drop-down menu at the top of the table.If you have hidden either Retail or Mfg results, the hidden results section will not be printed. Tocopy the results of one or several scenarios for “pasting” into another file, select thecorresponding columns in the table and click on Copy on the Edit drop-down menu.

5.1.5 Viewing Inputs

By selecting 'Inputs' on the Tools drop-down menu you may review the input data associated withany scenario for which the results are currently displayed in the Simulation Results Table.

42

A window (Figure 5.2) will open displaying the six tabs for the data associated with the scenariowhose results appear in the first column of the Results table. The Data Set drop-down menuallows you to select the scenario of interest.

Figure 5.2: Inputs viewed from the Simulation Results Table

5.2 Weekly Performance Graphs

By selecting 'New Graph' on the Tools drop-down menu, you may view a week-by-week plot ofany one of the following performance measures for any scenario(s) for which the results arecurrently displayed in the Simulation Results Table.

Retail Measures

Actual Cumulative ($) Sales displays the cumulative (from the first through the current week)actual dollar sales

Projected Cumulative ($) Sales displays the cumulative (from the first through the current week)projected' (based on Buyer's Plan) dollar sales.

Both Actual and Projected may be displayed in the same graph.

Purchases displays either the actual number of purchases each week.

43

Customers displays either the actual number of 'Customers' who entered the store in each week.The number of customers in any week can be thought of as the potential number of sales in thatweek.

Both Purchases and Customers can be displayed in the same graph.

Weekly Service Level displays the Service Level achieved in each week.

Cumulative Service Level displays Service Level the cumulative (from the first through thecurrent week).

Both Weekly and Cumulative can be displayed in the same graph.

GMROI displays the cumulative (from the first through the current week) GMROI.

Gross Margin displays the cumulative (from the first through the current week) Gross Margin.

Stockouts displays the number of stockouts that occurred in each week. A stockout occurs timea new customer doesn't find his/her first choice of SKU.

Lost Sales displays the number of sales that were lost in each week. A lost sale occurs eachtime a customer leaves the store without buying anything (1st, 2nd, or any choice).

Both Stockouts and Lost Sales can be displayed in the same graph.

No. SKUs Out of Stock displays the number of SKUs that were out of stock at the end of eachweek. An empty shelf doesn't necessarily imply a stockout. A stockout occurs when a customerwants an SKU that is out of stock

Average Inventory displays either the average number of units in stock in each week.

Inventory Receipts displays the receipts (number of units delivered) in each week.

Both Average Inventory and Receipts can be displayed in the same graph.

Percent Volume Error displays the error in the current estimate of the demand volume. It iscalculated as the actual demand volume minus the estimated demand volume divided by theactual demand volume.

Estimated Raw Material Mix displays the estimated (based on Buyer's Plan) demand mix as ofthe current week (in terms of the raw material differentiation you chose on the VendorSpecification tab).

Raw Material Mix displays the actual demand mix as of the current week (in terms of the rawmaterial differentiation you chose on the Vendor Specification tab).

Both Estimated and Actual Mix can be displayed in the same graph.

Manufacturing Measures

Raw Material Inventory displays the number of units of raw material inventory for each week.

Raw Material Receipts displays the number of units of raw material received by the manufacturerfor each week.

44

Both Inventory and Receipts can be displayed in the same graph.

Retail Orders displays the number of units of product ordered by the retailer for each week.

Retail Shipments displays the number of units of product shipped to the retailer for each week.

Both Orders and Shipments can be displayed in the same graph.

Work In Process displays the number of units of product in production for each week.

Shop Order Releases displays the number of units of product releases for production for eachweek, i.e., it is the number of units scheduled for production each week.

Both Work In Process Shop Order Releases can be displayed in the same graph.

Retail Backorders displays the number of units backordered for each week, i.e., it is the numberof units that the manufacturer was unable to ship to the retail per the Retail Orders.

Finished Goods Inventory displays the number of units of the finished goods, i.e., product, ininventory each week at the manufacturer.

5.2.1 Creating Graphs (Figure 5.3)

When you select 'New Graph', a Graph window will open displaying an "empty" graph and themenu bar will change. The Graph window and the menu bar will both contain Data Set, RetailGraph, and Mfg Graph menus. Use either Retail Graph (Mfg Graph) menu to select the retail(manufacturing) measure to plot and use either Data Set menu to select the scenario(s) to use.On the Data Set menus there will be one entry for each column displayed in the Results Table,labeled the same as the column in the table. If there are no results currently displayed in theSimulation Results Table, the Data Set menus will be empty.

45

Figure 5.3: Graphical Simulation Results

Pairs of measures that may be displayed on the same graph may be selected at the same time.

The graph can also be used to visually compare the performance measures from two or morescenarios. Just select each of the scenarios of interest from the Data Set menu.

Note: If Raw Material Mix is selected, the data for only one scenario can be graphed at atime.

5.2.2 Side-by-Side Display of Graphs

Two or more graphs that cannot be displayed in the same graph may be displayed simultaneouslyby opening additional Graph windows. To open an additional Graph window, simply select 'NewGraph' on the Tools drop-down menu.

You will need to slide the Graph windows around for best viewing.

5.2.3 Reformatting Graphs

Using the Format drop-down menu, you can reformat the graph in any open Graph window foreasier viewing by altering the type of line displayed in the graph and/or introducing grid lines. Youmay also choose between line and bar graphs.

Line type may be changed to help distinguish between plots corresponding to differentscenarios or between different measures simultaneously displayed. The line type can bechanged to either solid, solid marked, or patterned using the Line Type menu.

46

The default display has no grid lines. The Grid Lines menu allows you to add (remove)horizontal grid lines, vertical grid lines, or both.

The default display uses line graphs. The Graph Type menu allows you to switchbetween line and bar graphs.

5.2.4 Saving and Printing Graphs

By selecting 'Save Graph Data' on the File drop-down menu you may save the full set of graphsfor each scenario in the current Simulation Results table in a comma delimited text file which maybe imported into a spread sheet. A window will automatically open asking for a file name (with a".txt" extension).

The ".txt" files cannot be opened in the Simulator. However, whenever you save a SimulationResults table in a ".out" file, the data for all the associated graphs is also saved. To view thegraphs, you first open the ".out" file in the Simulation Results window and select Tools, then 'NewGraph.'

To save the graph (as a graph) into a file which may be imported into a word processor documentselect Print Graph on the File drop-down menu and click on 'Print to File' and 'OK' in the smallwindow that opens.

To print the currently displayed graph, select 'Print Graph' on the File drop-down menu and clickon 'OK' in the small window that opens.

5.3 Break-Even AnalysisBreak-even wholesale cost analysis allows you to determine the average wholesale cost at whicha unit of product could be purchased in a particular scenario to achieve a target Gross Margin orGMROI value. The target Gross Margin or GMROI may be a user-specified numerical value or avalue from another currently active run.

To perform break-even analysis, select 'Break-Even' from the Analysis drop-down menu when theSimulation Results window is open. After all of the inputs are entered, select Analyze - Run. Theresults of the analysis will appear below the inputs (Figure 5.4).

47

Figure 5.4: Break-Even Analysis

5.3.1 Inputs

Output Measure is the measure you would like to calculate a break-even cost for (either GrossMargin or GMROI).

Calculate Break-Even Cost for is the scenario for which you want the wholesale cost calculatedSelect a run from the drop-down menu.

Target Value to Match: You may choose either "Use Value from Run" or "Enter Specific Value."If you specify "Use Value from Run", a pull down menu of all active simulation runs will appear.Select the run that you want the target Gross Margin or GMROI value to match. If you select"Specific Value", a text box will appear for you to enter a numerical value.

For Example, suppose that you have made two runs, one using the Weeks Supplystrategy and one using the Newsboy strategy, and you have both of their results showingin the Simulation Results window. You would like to know what wholesale cost you couldhave paid while operating with the Newsboy strategy and attain the same Gross Marginas with the Weeks Supply strategy. The Output Measure would be Gross Margin, the"Calculate Break-even Cost for" run would be the Newsboy run, and the Weeks Supplyrun would be selected as the run to use for the Target Value.

5.3.2 Results

After you click the Analyze button, a table of results and a summary statement will appear. Theresults in the table are as follows:

48

Measure is the measure (Gross Margin or GMROI) for which the analysis is conducted.

Target Value is the value of Gross Margin or GMROI used as the Target value in the break-evenwholesale cost calculation.

Average Wholesale Cost is the (weighted average) wholesale cost that was used in thesimulation run of the run that you calculated the wholesale cost for.

Average Break-Even Wholesale Cost is the average cost that could be paid in the run that thebreak-even cost was calculated for to achieve the target value of the output measure selected.

Average Target Wholesale Cost is the (weighted average) wholesale cost used if an active runwas chosen as the source of the Target Value. If a specific numerical value was entered, thisoutput is not shown.

49

6 Creating and Plotting Decision SurfacesWhen you click on the Decision Surface Model button on the Main menu, you are presented witha Model Specification window. You may either use this window to create a new decision surfacemodel or, by clicking on Plot Specification on the menu bar, switch to a Plot Specification windowand proceed to make plots from a previously created decision surface model, make side-by-sideplots of several graphs, or simply display a previously stored graph.

For quick reference during execution, the Decision Surface Model has a Wizard that lists andbriefly explains your options in each window. Unless you disable it (using the check-box in thewizard’s main menu) the wizard windows will open automatically. If the automatic feature isdisabled, you may access the wizard by clicking on the Wizard button at the bottom of either ofthe Decision Surface Model windows.

6.1 Creating a New Decision Surface ModelThe following is an outline of the procedure that you should follow to create a new decisionsurface model:

Step 1: If there is data on the Inputs tab, go to Step 2. Otherwise, open a(“.dsm”) file containing batch simulation run results (inputs andperformance measures). Use Open under the File menu.

Step 2: Select the desired inputs and ranges on the Inputs tab and select thedesired performance measures on the Performance Measures tab.

Step 3: Click on the Create Surface button and respond to the questions.Step 4: To save the model for future reference, select Save As on the File drop-

down menu in the Plot Specification window (Optional).

The following provides more detail on each step:

Step 1Decision surface models are created by training a neural network using stored data from a batchrun of the Sourcing Analysis Model. If you came directly from making a batch run in the SourcingAnalysis Model, the associated ".dsm" file is automatically opened in the Model specificationwindow. Go to Step 2. Otherwise, in order to create the surface model, you initially have to openthe stored data file by selecting Open on the File drop-down menu and responding to the filemanager to select the correct “.dsm” file from the Batch Run Data directory.

Step 2The labels of inputs and performance measures for the batch run are displayed on theirrespective tabs. You may create decision surface models relating all or any subset of the listedinputs to each or any subset of the listed performance measures. To specify which, simply clickon the selection box to the left of the label (See Figure 6.1.).

50

Figure 6.1. Model Specification Window - Inputs Tab

Each input label is followed by 2 pairs of boxes. The first pair (Data Range) gives the ‘Minimum’and ‘Maximum’ values of the input that appear in the file. The second pair (Selected Range)allows you to reduce the Data Range. When you select an input, the second pair of boxesdisappears. The program will use the full range of data for this input in creating the surfacemodel. For the unselected inputs you may restrict the range of data used to create the model byraising the ‘Lowest’ and/or reducing the ‘Highest’ values. Doing so will cause all data points withvalues of this input outside the modified range to be ignored.

When you are finished with input selection, click on the Performance Measure tab and selectthose measures for which you want decision surface models. (A model will be created for eachmeasure selected.)

51

Figure 6.2. Model Specification Window - Performance Measure Tab

Based on your choices, the program selects the associated records from the data file. Theserecords constitute the set of “training patterns” for the neural network.

Step 3When you click on Create Surface, a progress bar appears. Depending on the number of inputsand performance measures and the speed of your computer, the training process may takeanywhere from a few seconds to several minutes to execute.

Once the process is finished, the Plot Specification window will open with the previously selectedinputs (and their data ranges) and performance measures listed.

Step 4Saving a decision surface model will allow you to view graphs from it later without going throughthe model creation process again (which can save you considerable time).

To save the model, just click on Save As on the File pull-menu in the Plot Specification windowand follow the instructions in the dialog window which opens. Unless you override the defaults,the file will be given a “.wts” extension and saved in a subdirectory called ‘DSM Data’.

6.2 Plotting a Decision SurfaceA specific decision surface model relates each of the performance measures that you selected inthe Model Specification window to all of the inputs that you selected. The associated graphswhich you can specify and display will plot one performance measure against one or two of the

52

inputs with all other inputs fixed at specified (single) values. Thus with any model there are anumber of possible graphs.

The Plot Specification window is used specify and plot graphs. This window is openedautomatically upon completion of surface model creation. Alternatively, in order to plot graphsfrom a previously stored surface model, you may open the window directly from the ModelSpecification window by clicking on Plot Specification on the menu bar (See Figure 6.3).

Figure 6.3. Plot Specification Window

The following procedure summarizes the flow of operations that should be executed to obtain adecision surface graph:

Step 0: If there is data displayed in the window, skip to step 1. Otherwise, clickon Open on the File drop-down menu and select the desired model(“.wts”) file .

Step 1: Click on ‘2-Dimension’ (‘3-Dimension’) and select one (two) input(s) fromthe displayed list.

Step 2: Edit the range for each of the selected inputs (Optional). Edit the step-size for each of the selected inputs (Optional). Edit the fixed value for theunselected inputs (Optional).

Step 3 Select one or more performance measures from the displayed list.Step 4: Click on Plot, enter a title, and click on OK to display the graph.Step 5 To save the graph, select Save on the File drop-down menu in the Plot

window. To print the graph, select Print on the File drop-down menu inthe Plot window. (Optional). A previously saved graph may be openedeither by selecting Open Graph on the File drop-down menu of the Plotwindow or by selecting View Graph on the menu bar in the PlotSpecification window. A side-by-side display of up to six stored graphsmay be created by selecting Compare on the menu bar of the PlotSpecification window.

The following provides more detail on each step:

53

Step 0If you have just created a new surface model the Plot Specification window is openedautomatically displaying the list of inputs and performance measures for that model. Theminimum and maximum values of the data for each input to the model will also be shown (DataRange) along with the midpoint of the Data Range (”Fixed” column under Plotting Range).

If you switch to the Plot Specification window without creating a new surface model, you will needto open a stored model (“.wts” file) in order to display the corresponding input data andperformance measures list.

Step 1:While the decision surface model may have more than 2 inputs, a graph, of course, can displayperformance measures against only one or 2 inputs, with the other inputs set at some fixedvalues.

If you want to plot performance against a single input, click on ‘2-Dimension’ and click on thecheckbox in front of the input. If you want to plot performance against two inputs, click on ‘3-Dimension’ and click on the checkboxes for the two inputs.

For the selected inputs, the ‘Fixed’ value displayed under Plotting Range will be replaced by a pairof ‘Minimum’ and ‘Maximum’ values which match those of the Data Range plus a ‘Step-size’value. Unless you edit these values (Step 2) they will determine the range and step-size for thecorresponding input in the graph.

Unless you edit it (Step 2), each unselected input will be set at the ‘Fixed’ value displayed underPlotting Range.

Step 2:If you wish, for the graph you may shrink the range of any selected input by increasing the‘Lowest’ and/or decreasing the ‘Highest’ value shown in the dialog box under Plotting Range(Make sure the chosen range falls within the Data Range.).

You may replace the default step-size value(s) for labeling the input axis (axes) in the graph. Tryto select a value that divides the corresponding range evenly. You may want several step-sizesand see how each looks.

You may also replace the default ‘Fixed’ value for any unselected input with any value in its DataRange. (Make sure you don’t specify a value outside of this range.).

Step 3:In a 3-dimensional graph you may display only one performance measure against the selectedinputs, whereas in a 2-Dimensinal graph you may choose to display two or more of the listedperformance measures simultaneously. Select the measure(s) to be displayed by clicking on thecorresponding checkboxes. In order for the graphs to be meaningful, when two or more measuresare selected they should have the same units (e.g., dollars or percent).

Step 4:When you click on Plot, a Plot Title window is opened. Enter a title for the graph and click on OK.The specified graph is created and displayed in a Plot window (See Figures 6.4a and 6.4b).

54

Figure 6.4a. Plot Window - Sample 3D Plot

55

Figure 6.4b. Plot Window - Sample 2D Plot

You may alter the scale for the vertical (performance measure) axis of the graph by Scale Z-Axis(followed by Manual) on the Format drop-down and modifying the minimum and maximumcoordinates to be displayed. The step-size within the new range will be set automatically. Thisfeature is useful for changing perspective and/or creating side-by-side displays.

You may modify the appearance of 3-Dimensional graphs by adding/removing horizontal andvertical grid lines on the axes (using the Format drop-down menu) and/or by changing theorientation of the graph (using the Rotate drop-down menu). To rotate the graph you may selecteither Manual or Automatic rotation. Selection of Manual, provides a scrollbar for you tomanipulate. Selection of Automatic causes the graph to slowly rotate until stopped by selectingStop on the Rotate drop-down menu.

If you click on a corner of one of the rectangles on a plotted 3-Dimensional surface, a window willopen displaying the coordinates (associated values of the two inputs and performance measure)of the point.

With 2-Dimensional graphs you may switch performance measures in the plot using the Categorydrop-down menu. This menu contains each of the performance measures in the decision surfacemodel.

56

Step 5:You may wish to save a graph, in order to be able to redisplay it at a later time or to import it to adocument. This is done by choosing Save Graph on the File drop-down menu and following theinstructions in the dialog window that opens. Unless you specify otherwise the graph will be savedin a Windows Metafile (“.wmf”) format in a subdirectory named ‘Plots’.

An individual stored graph may be opened in either of two ways, by selecting View Graph on themenu bar in the Plot Specification window or by selecting Open Graph on the File drop-downmenu in the Plot window. In either case make your selection from the list which appears. Whendisplayed, stored graphs cannot be reformatted or rotated. Up to six stored graphs may bedisplayed side-by-side using Compare on the menu bar of the Plot Specification window (SeeSection 6.3 below.).

To print the displayed graph, select Print Graph on the File drop-down menu.

6.3 Side-by-Side Display of Decision SurfacesFor comparison purposes it is often useful to see several decision surface graphs simultaneously.In the Plot Specification window you can create (save and/or print) side-by-side displays of up tosix previously stored graphs.

To do so select Compare on the menu bar in the Plot Specification window. A smaller GraphComparison window will open. Click on Graph followed by Add. The contents of the Plotssubdirectory will be displayed. Select the first graph that you want to include and click on Open.The graph will be displayed in the Graph Comparison window. Continue to add (up to a total ofsix) graphs by clicking on Add and selecting graphs. Graphs may be removed from the display byselecting Graph followed by Delete and selecting the graph from the Plots subdirectory.

As graphs are added or removed you may need to resize the window. The plotted graphs cannotbe re-scaled or rotated. Thus in anticipation of side-by-side plotting, for best comparisonperspective, you should make sure, before saving, that the views and scaling on the vertical(performance measure) axes of individual graphs of the same or similar measures are the same.

To print or save a side-by-side graph, use the File drop-down menu as in Step 5 above. Similarly,you may open a stored side-by-side graph either by selecting either View Graph or Compare onthe menu bar in the Plot Specification window or by selecting Open Graph on the File drop-downmenu in the Plot window.

57

7 Sample Scenarios

7.1 Tutorial Scenarios

Following are sample scenarios intended to introduce you to using the Sourcing Simulator Models.The scenarios illustrate the alternative operational modes, different strategies and other featuresof the Sourcing Simulator including data input, execution, and examination of results. Werecommend that you go through the scenarios in order.

Scenario 1 uses a Quick Response sourcing strategy.Scenario 2 uses a Traditional strategy (to illustrate offshore sourcing).Scenario 3 uses a Vendor Managed Inventory strategy.Scenario 4 uses a NewsBoy strategy.Scenario 5 uses a Model Stock strategy for a Basic good.Scenario 6 compares the Target Weeks and Newsboy strategies for a Basic good.Scenario 7 demonstrates the Presentation Stock Feature in the Basics Model.Scenario 8 demonstrates the Creation and Plotting of a Small Decision Surface Model for a

Basics Scenario.Scenario 9 demonstrates the Creation and Plotting of a Bigger Decision Surface Model for a

Seasonal Scenario.Scenario 10 demonstrates the use of the Detailed Vendor Specification.

The presentation assumes that you are running the Sourcing Simulator package and followingalong with the steps as they are described.

Now run the Sourcing Simulator and click on the Sourcing Analysis Model button to openthe Sourcing Analysis Model (data entry) window.

When you have done this, proceed with reading.

7.2 Scenario 1 - Quick Response Sourcing

(Default Data)

With this initial scenario, we shall execute the default-input data set (after carefully examining itscontents) and examine the results both in tabular and graphical format. The product is seasonalgarment and the operational sourcing strategy is Quick Response (QR).

7.2.1 Data Entry

Since we are using the default data, you will not need to edit any of the data displayed on the 5tabs in the window that is open in front of you. However, we will go through each default entry tobe sure that you understand exactly what inputs are required.

Recall that data entry/edit should begin with the Buyer’s Plan data. So, if it is notshowing, click on the Buyer’s Plan tab.

7.2.1.1 Number of Simulation Replications

58

At the top of the data entry window, you see that the simulation is to be replicated 10 times (i.e.,repeated with 10 different random customer streams) and the results averaged.

7.2.1.2 Buyer’s Plan Data

In this example, the Selling Cycle is 20 weeks long and is not repeated. The product line offeredis “garment” and consists of two styles each in four colors and six sizes, making a total of 48SKUs. A fixed presentation stock is not used.

Allocated display floor space data is not used in this scenario.

The buyer anticipates selling 4800 units. Of these, the buyer anticipates that 60% will be style 1,leaving 40% style 2. The color and size mix figures should be interpreted in a similar manner.

The buyer anticipates a ‘MidPeak’ (i.e., pattern of demand) with the weekly percents rising untilaround week 10 and then tapering off, as pictured in the small graph and shown in the table.

You may now proceed through the tabs in any sequence.

7.2.1.3 Consumer Demand Data

Recall that the actual number of customers that the Simulator randomly generates and their SKUpreferences need not match the volume and mix in the Buyer’s Plan.

In this example, the actual volume is specified as a 10 percent error (deviation) from the figureentered on the Buyer’s Plan tab, meaning that the buyer’s figure is a little low. (Note: the samevolume could have been specified by selecting “actual value” and entering 5040 (i.e., 1.10 x4800)).

Similarly, in this example, the actual mix is specified on the basis of percent error from thecorresponding figures in the Buyer’s Plan. Specifically, the Simulator will randomly generate abreakout of demand by style, by color, and by size such that the demand and plan figures for stylediffer by a total of 40, while those for color and size will differ by totals of 40 and 20, respectively.(Alternatively, the actual demand mix could have been specified by indicating “actual values” onthis tab and filling in the percents for each style, color, and size in the reopened Buyer’s Planwindow.) The specified mix will not change as the season progresses (i.e., Demand Twist is notselected).

The default selection for actual demand seasonality is “MidPeak”, which happens to be “same aspresumed seasonality” (though that it is not selected) meaning that the buyer’s presumedseasonality is accurate in this case. (Note that the price elasticity associated with any markdownsmay dampen the drop off in demand beyond week 10.)

Looking again at the Consumer Demand block, 24% of those customers who encounter astockout (will be randomly selected to) look for an alternative choice.

7.2.1.4 Cost Data

The wholesale price paid to the vendor(s) in this example is $12.50 per unit for both the initialstock and replenishment stock. In turn, the product is offered in the retailer’s store at a base (or“first”) price of $25.50. Units left on the shelf at the end of the season (i.e., end of week 20) will beliquidated for $8.00 each. All SKUs cost the same and are priced the same.

59

(Although not used in the default scenario, just to see how to build up these figures fromconstituent costs/prices, double click on the Ordering Cost dialogue box and look at thewindow that opens. When you are finished click on Cancel.)

Associated with both the initial delivery each subsequent replenishment delivery to the store thereis a fixed shipping cost of $100.00 plus $1.00 per unit delivered. In addition there is a $25.00 costassociated with each order. There is no overhead associated with managing this particularprogram.

The cost of carrying inventory is computed as 20 cents on the dollar (wholesale unit cost) peryear. Over a 20 week season this amounts to about 7.7 cents (20/52 x 20) on the dollar.

The handling charge to prepare and place units on the shelf is computed as 8 cents on the dollar(wholesale unit price).

7.2.1.5 Markdowns/Promotions

In this example, there will be one markdown of 25% in the base (Retail) price effective at thebeginning of week 18. It will remain in effect until the end of the season, i.e., for 3 weeks.

The price elasticity figure that will be used to adjust the expected demand (customer arrival rate)in weeks 18, 19, and 20 is 0.7 which means each 1% reduction in the base price will result in0.7% increase in expected demand.

Also, once the markdown is in effect, the percent of customers who will look for an alternativeSKU when they encounter a stockout will be 50% (up from 24%).

There are no promotions.

7.2.1.6 Sourcing Strategy

The illustrated sourcing strategy is Quick Response.

40% of the Buyer’s Plan volume will be on the shelf at the beginning of the season. This order willbe placed 10 weeks in advance of the start of the season and delivered just before the seasonstarts. The retailer will re-estimate demand for each SKU and place a reorder at the end of eachof the first 14 weeks of the season. Individual SKUs can be ordered 1 unit at a time and an ordercan be placed for a single unit (i.e., there are no minimum order restrictions in this example).Each reorder will be received after a fixed lead-time of 2 weeks.

7.2.1.7 Vendor Specification

The vendor is a 'perfect supplier', meaning that the content of each shipment will exactly matchthe order. The Material Supply is 'by color', which means that the Raw Material Mix graphs willdisplay the weekly estimated and/or actual product demand mix by color.

Since in this example scenario you are using the default values, “data entry” is nowcomplete and you are ready to run the simulation.

7.2.2 Running the Simulation

60

To run the simulation select ‘Run’ on the Simulate drop-down menu. Accept the default name,QR1 and the default single demand volume error option by clicking on ‘OK’. The scenario willsimulated only for the case in which the buyer's estimate of demand volume is 10% low.

In a few seconds a new window labeled Simulation Results will open showing the results for thisscenario. Note that the column of results is labeled with ‘QR1’.

7.2.3 Interpreting the Simulation Results Table

The Simulation Results Table that appears displays the average (over the 10 replications) valuesfor the full set of measures of performance for the season. Specifically,

5413 customers entered the store and interacted with the stock.

At the beginning of the season there were 1918 units on the shelf. Within the season (weeks 3-17, in particular) an additional 3450 units were received, based on reorders, making a totaloffering of 5368 units. On average the shelf stock ran at 1178 units, with 4.56 turns.

98.1% of the units offered were sold during the season, leaving 1.9% to be liquidated at season’send. 86.6% of the units offered were sold at first (base) price.

On average over the entire season, 94.6% of the SKUs were in stock. Prior to the delivery of thefirst reorder (in week 3), the in-stock percentage ran at 98.4%, while prior to the markdown inweek 18, it ran at 98.3%. Over the entire season, 2.7% of the customers left empty-handed (lostsales). However, up until the markdown in week 18 all but 0.4%customers were able to find aSKU to purchase. For the entire season the service level (percent of customers who found theirfirst choice SKU in stock) was 96.0%, whereas prior to the receipt of the first reorder, it ran at98.4% and prior to the markdown, it ran at 99.3%.

The total revenue for the season was $131,165), of which $130,355 came from sales tocustomers and $810 from end-of-season liquidation. Dividing by the total offering gives a revenueper unit of $24.43.

The wholesale cost of the units offered was $64.060. In addition, the store incurred $6868 inshipping costs, $375 in additional ordering costs, $5368 in handling costs, and inventory carryingcosts of $1133.

The gross margin for the season was $60,064 which is 91.8% of the maximum ($65,429) thatcould have been obtained from the offering if all units had sold at first (base) price. (Since you didnot enter a Display Floor Space figure on the Buyer's Plan tab, the gross margin per square footfigure simply repeats the gross margin). The associated adjusted gross margin (subtracting thecosts of shipping, handling, and carrying) was $50,316. The associated GM% (dividing the grossmargin by the total revenue) was 48.8%. The GMROI (dividing by the wholesale cost of theaverage shelf stock level) was 4.35 while the associated GMROISL (multiplying GMROI by InStock %) was 4.12

7.2.4 Viewing the Inputs

If you wish to recall the inputs leading to the results in the table, you can see them by using theTools drop-down menu.

Let’s suppose you can’t remember the retail price of the product. Selecting ‘Inputs’ on the Toolsmenu opens a small Inputs window with six data tabs showing. Selecting the ‘Cost Data’ tabdisplays the data from the original Cost Data input tab and you see that the retail price was$25.50.

61

Note that clicking on the Data Set drop-down menu shows one entry, ‘QR1’, which matches thelabel of the Inputs window.

7.2.5 Printing the Results Table

If you wish to print the Results table, select ‘Print Output’ on the File drop-down menu at the top ofthe Simulation Results window. You will get a table with a format similar to that in the window.You can also select one or more (contiguous) columns in table and then click on Copy on the Editmenu to make a copy which can be “pasted” into another file.

7.2.6 Saving the Results Table

You can save the tabled results in a system file, which allows you to view them at a later time inthe Simulation Results window or as a text file which you can import to a spreadsheet.

To save the results, select ‘Save As’ on the File drop-down menu. If you wish the results to besaved for future viewing, your filename should have the extension “.out”. To save them in a textfile, your filename should have the extension “.txt”. The default extension is “.out”. To switch to“.txt”, select it on the ‘Save File As Type’ drop-down menu.

In order to make it easier to recall the scenario associated with the results, you may change thelabel at the top of the column of results by double clicking on the label (to mark it), deleting thecurrent label, and entering the new label. Alternatively you may mark the label for deletion byselecting the column and then choosing ‘Change Column Label’ on the Edit drop-down menu.

For use in the next example, save the results in a file named QR1.out.

7.2.7 Viewing Graphs of Results

In addition to the above table of results, the Sourcing Analysis Model also provides graphs of theweek-by-week values of several performance measures. (The plotted values are those from oneof the 10 replications of the season. This means that there may not be a perfect match withcorresponding figures in the previous table.)

To see the graphs, click on Graph on the Tools drop-down menu on the Simulation Resultswindow (or on the original input data window). A graph window will open with an empty graph.

Select ‘QR1‘ on the Data Set drop-down menu. The Actual Cumulative Sales graph will bedisplayed. To display another, select the graph you wish to see on the Retail Graphs menu.

Let’s examine each category of retail graph so that you understand what is plotted.

7.2.7.1 Actual and Projected Cumulative($) Sales

Actual Cumulative Sales are sales-to-date as generated by the Simulator in the scenario, whileProjected Cumulative Sales are figures based on the buyer’s plan. You can select either or both.

Selecting both of them lets you see that beyond the 2nd or 3rd week of the season the actualdollar sales to date exceed that which was expected from the buyer’s plan (i.e., projected). Thedemand re-estimation and reordering process in QR was able to correct for the fact that theBuyer’s Plan volume was 10% low.

62

The Format drop-down menu allows you to change the appearance of the graphs. Adding gridlines will aid in reading values off the graph.

7.2.7.2 Purchases and Customers

Purchases is the actual sales each week while Customers is the number of customers thatentered the store each week. You can select either or both.

Selecting both of them reveals that there is a good match (almost every customer buys) until nearthe end of the season after the markdown (beginning of week 18) and after the last replenishmentarrives (end of week 16).

7.2.7.3 Weekly and Cumulative Service Level

Service Level is the percent of customers who find their first choice SKU in stock. You can chooseto plot the service level for each week, the cumulative to date, or both.

Consistent with the ‘Purchases and Customers’ graphs, the service level is quite high until the lastcouple of weeks of the season. This demonstrates how the demand re-estimation and reorderingprocess in QR is able to produce a good match between supply and demand in spite of a largeinitial plan (forecast) error.

7.2.7.4 Gross Margin and GMROI

These two graphs show the Gross Margin and GMROI to date, respectively, for each week in theseason. (They cannot be plotted together.)

With this example both graphs run at a negative value for a few weeks until sales revenues buildup enough to cover the cost of the initial stock and then continue to increase as total sales buildup.

7.2.7.5 Stockouts and Lost Sales

The first of these two graphs shows the number of customers who found their selection to be outof stock during the week, while the second shows the number of customers who actually leave thestore each week without making a purchase. Recall that a customer who encounters a stockoutmay end up buying a different SKU. The two measures may be selected individually or together.

The patterns are consistent with those for Purchases, Customers and Service Level. Plotting bothon the same graph lets you see not all stockouts result in lost sales.

7.2.7.6 No. of SKUs Out of Stock

This graph shows the number of SKUs that are out of stock at the end of each week,

7.2.7.7 Average Inventory and Inventory Receipts

You can choose to plot average number of units in stock during each week, the weekly receipts ofnew stock from the vendor, or both.

63

Looking at the receipts, you can see the initial stock of around 40% of the Buyer’s Plan volumefollowed by no receipts (for two weeks) until the first reorder placed at the end of week one comesin at the beginning of week 4. This first reorder receipt is followed 13 others, for a total of 14 asspecified on the Sourcing Strategy input tab. The Weekly inventory level drops quickly during thefirst three weeks when there are no receipts and then more slowly as reorders based on Point-of-Sale reshape the stock until week 17 when the final reorder is in. Then the stock drops off quicklysince there is no replenishment and the stock has been shaped to pretty well match the actualdemand mix.

7.2.7.8 % Volume Error

This graph shows how quickly (or slowly) the demand re-estimation procedure used with the QRstrategy hones in on the actual (versus Buyer’s plan) demand volume..

You can see that the % Volume Error moves from around 9% (low) in the plan (you entered 10%;the random demand generator actually generated 9%) to around 3% (high) after one week ofsales data and to less than 2% after 2 weeks of sales data.

Estimated and Actual Raw Material Mix

The Actual Raw Material Mix graph shows the actual percent of demand (week-by-week) for each(n this case) color while the Estimated Raw Material Mix graph shows corresponding estimatedpercent (which doesn’t change) for each or the 4 colors in the line. They may be selectedindividually or together.

By selecting both, you can see that, as with volume, the re-estimation procedure is quick tocorrect for the initial Buyer plan mix error.

7.2.8 Printing the Graphs

Selecting ‘Print Graph’ on the File drop-down menu and clicking on ‘OK’ in the small windowwhich opens will print the currently displayed graph.

7.2.9 Saving the Graphs

Selecting ‘Save Graph Data’ on the File drop-down menu will save the full set of graphs for thescenario in a comma delimited, text file which may be imported into a spread sheet. A window willautomatically open asking for a file name (with a “.txt” extension).

To save the graph (as a graph) into a file which may be imported into a word processor document,select ‘Print Graph’ on the File drop-down menu and click on ‘Print to file’ in the small window thatopens.

The “.txt” files cannot be opened in the Sourcing Analysis Model. However, whenever you save aSimulation Results table in a “.out” file, the data for all the associated graphs is also saved. Toview the graphs, you first open the “.out” file in the Simulation Results window and select ‘Tools’then ‘Graph.’

7.2.10 Running with Several Volume Errors

Before looking at other sourcing strategies and comparing results, let's illustrate the use of therun-time option to obtain average results for several different volume error values.

Suppose you were really not sure what volume error value to enter on the Consumer Demand tab.Say, you think that "it most likely it is 10% but it could be 5% or 15% and maybe even 20%."

64

Leaving the Simulation Results Table open with the "QR1" results showing, select 'SourcingAnalysis Data Entry' on the Window drop-down menu to return to the Data Entry window. Nowselect 'Run' on the Simulate menu. The Run Name and Volume Error window will open.

Change the default name to "QR Avg. Error" to clearly label the results.

Your uncertainty about volume error can be reflected as follows. Enter 5 for Min, 20 for Max and4 for # Runs. The associated four Error % values will appear in the box to the right. EnterWeights%s of 10 for the 5% and 15% errors, 5 for the 20% error, and 75 for the 10% error. (Notethat the weight %s total 100%).

Keeping the 'Show Average Only' default, click on OK. The Simulation Results Table willreappear with weighted average results (weighted as you specified) appearing in the 2nd column.

You can see that the average results are pretty close to those for the single 10% error value(QR1).

To get a feel for how much difference the volume error could make (within the 5% - 20% range),repeat what you just did but select 'Show All Runs'. The Simulation Results Table will get 5 newcolumns, one for each of the four error levels (as labeled), plus the average again.

Note how the demand re-estimation used with QR was able to compensate for the increasingerror level, maintaining the high service levels and capturing the unanticipated demand.

7.3 Scenario 2 - Traditional Sourcing Strategy(Editing Default Data/Comparing Scenario Results)

For this second scenario we shall edit (some of) the default data to create an alternative whichmodels a 20-week season offshore sourcing scenario in which wholesale costs are lower offshoreand differ with the product style. We’ll display the results along with those from Scenario 1 toallow easy comparison. To do this we will make use of the Traditional sourcing strategy.

7.3.1 Data Entry

We’ll begin from the point where we are looking Buyer’s Plan tab in the data entry window and theopen data file is the default file (default.saf) used in Scenario 1.

7.3.1.1 Number of Simulation Replications

We will use the default value of 10 replications.

7.3.1.2 Buyer’s Plan Data

Again we will use the default data. Recall that the buyer anticipates selling 4800 units over the 20-week Selling Cycle with a Mid Peak seasonality pattern. The line consists of 48 SKUs with 2styles in 4 colors and 6 sizes with expected sales mix as shown. The allocated floor space is notspecified.

7.3.1.3 Consumer Demand

65

We will use the default data on this tab as well. The volume error is 10 percent, while the mixerror is 40-40-20 and ‘Error Twist’ is unselected. The actual seasonality pattern is Mid Peak. 26%of customers who encounter a stockout look for an alternative SKU.

Having the same Buyer’ Plan and Consumer Demand data as with Scenario 1, will result in thesame 10 (replicates) customer streams as in Scenario 1 and thereby permit a meaningfulcomparison of results.


On this tab we’ll make our first data edit.

In simulating offshore sourcing, we’ll assume that the entire offering (volume and mix from theBuyer’s Plan) is produced and delivered to the retailer by the beginning of the Selling Cycle.There is no chance reorder within the season.

To achieve this, specify the Traditional sourcing strategy with an initial inventory of 100% of Plan.Notice that the replenishment schedule data is automatically adjusted.

Leave the rest of the data at the default settings.

7.3.1.5 Cost Data

We’ll keep the $25.50 retail base price and the $8.00 liquidation price. However, to reflect areduction in labor cost associated with offshore manufacturing, we’ll specify lower wholesale costswith offshore manufacturing. In particular, select ‘By Style’ on the Unit Cost/Price Data drop-downmenu. Note that table now contains 2 rows, one for each of the 2 styles in the Buyer’s Plan. Forstyle 1 change the $12.50 wholesale costs to $10.80 each; for style 2 enter $11.30 for eachwholesale cost (and enter $25.50 Retail Price and $8.00 for Liquidation Price). Since the Buyer’sPlan mix is 60% style 1 and 40% style 2, and all garments are being bought according to the plan,the weighted average wholesale cost will be $11.00 (.60 X $10.80 + .40 X $11.30) per unit

To reflect the need to bring the product from offshore, lets say that there is an initial fixed shippingcost of $1000 along with an initial variable shipping cost of $1.50/unit. (With no replenishmentorders, the associated shipping costs will be ignored and can be left as they are. Actually thereplenishment wholesale cost figures won't be used either.) To reflect the extra effort to managean offshore contact, enter a $1000 program overhead. Finally, let’s assume that all costs ofordering are included in these figures and set the ordering cost to $0.0.

(Alternatively, we could have kept the default shipping and ordering cost figures as costsassociated with product once in the US and incorporated the cost to get the product to the US inthe wholesale cost and overhead figures.)

Leave the carrying and handling charges of 0.20 and 0.08, respectively.

7.3.1.6 Markdowns/Promotions

We’ll use the default data − no promotions; one scheduled markdown of 25% at the beginning ofweek 18; a price elasticity of 0.7; an increase to 50% in percent of customers who look for asecond choice when encountering a stockout.

66

7.3.1.7 Vendor Specification

We'll assume that the vendor is a perfect supplier and continue to distinguish raw material bycolor.

Data entry is now complete and you are ready to run the simulation for Scenario 2.

7.3.2 Saving the Input Data

Before running the simulation, we will see how to save input data for a particular scenario.

Select ‘Save As’ on the File drop-down menu and specify ‘offshore.saf’ as the file name inthe small window that opens. We’ll use this file in the next example.

To see that you have been successful, now select ‘Open’ on the File menu. You should see‘offshore.saf’ in the list of files in the small window that opens.


Recall that to run the simulation all you need to do is to select ‘Run’ on the Simulate drop-downmenu at the top of the window. Accept the default single volume error value. Recalling that youare really using this strategy to simulate an offshore sourcing strategy, try to change the run nameto 'Offshore' but misspell it, say 'ofshor'. Click on 'OK'.

In a few seconds the results window will open showing the results for this scenario in the firstcolumn, labeled ‘ofshor’

Noting the spelling error change the label at the top of the column by double clicking onthe label at the top of the column, deleting the misspelling, typing “Offshore”, and hittingenter.

7.3.4 Interpreting the Tabled Results

Rather than looking at the results from this offshore scenario by itself, you can view them side-by-side with those from the QR scenario (Scenario 1) for comparison.

To do this select ‘Open’ on the File drop-down menu on the Simulation Results window and thenselect the file “QR1.out” (the filename you used to save the results from Scenario 1). The resultsfrom Scenario 1 will appear in the second column of the Simulation Results table.

Since with the "Offshore" strategy there were no reorders within season the "Before 1st ROReceipt" rows under IN STOCK, SERVICE LEVEL, and LOST SALES are not meaningful.Similarly, since you did not specify Display Floor Space on the Buyer's Plan tab, the GM/Sq. Footfigures are not meaningful. Hide the corresponding rows using Retail Outputs/Customize on theView drop-down menu.

Now turning to the results, first note that the number of customers entering the store was thesame (5413) in both scenarios. In fact, the Simulator generated exactly the same streams ofcustomers for both scenarios.

Whereas under QR, where most of the stock was ordered within the season, in the offshorescenario the entire offering (4800 units) was in the store at the beginning of the season. As aresult of demand re-estimation and reordering (recall that the Buyer Plan errors were 10% lowvolume and 40-40-20 mix), the total offering was nearly 600 larger under QR. In spite of this,

67

since product came in small quantities throughout the season, the average inventory under QRwas much lower (1178vs. 2446) and the turns much higher (4.56 vs. 1.96).

Even though the total offering under QR was much larger, there was a much better matchbetween offering and demand. This resulted in a much lower percent liquidated (1.9 vs.15.4),larger average percent of SKUs in stock (98.3% vs. 88.3% before markdown and 94.6% vs.82.7% overall), higher service level (99.3% vs. 70.9% before markdown and 96.0% vs. 63.7%,overall), fewer lost sales (0.4% vs. 18.3% before markdown and 2.7% vs. 25% overall), andhigher dollar sales ($131,165) vs. $107,861).

The cost advantage of the offshore strategy is seen in the cost of the goods ($52,800 vs. $67,105)and associated onshore handling cost ($4224 vs. $5368). However, this advantage is partiallyoffset by the higher costs of shipping plus program overhead ($8400 vs. $6868). In addition, withthe much lower average stock level, the inventory carrying cost was lower under QR ($1133 vs.$2082).

In spite of the 14% higher wholesale cost, the QR strategy resulted in about a $9000 larger grossand adjusted gross margins. While with QR, the store had a lower GM% (48.8 vs. 51.0),it hadmore than twice the GMROI (4.35 vs. 2.03), and GMROISL (4.12 vs. 1.68).

7.3.5 Break-Even Analysis

Let’s use the Break-Even Analysis feature to answer the following question: “How high of awholesale cost per unit could we pay with Quick Response and still expect to obtain a GrossMargin as large as the one we got with the offshore sourcing strategy.

To answer this question, first choose ‘Break-Even Analysis’ on the Tools drop-down menu. In theBreak-Even Analysis window that opens, select ‘Gross Margin’ as the output measure and ‘QR1’as the strategy to calculate the break-even cost for. Then click on the ‘Use Value From Run’button and select ‘Offshore’ as the run to provide the target Gross Margin value. Finally, select‘Run’ on the Analyze drop-down menu.

From the results that now appear in the Break-Even Analysis window, you can see that in order toget the “target” Gross Margin of $55,061(obtained with the offshore strategy at an averagewholesale cost of $11.00), with the QR1 strategy, the wholesale cost could rise to $14.18), orabout 29% higher than that with the offshore strategy.

To get a print out (of the window), select ‘Print’ on the File drop-down menu.

Repeat the analysis for GMROI instead of Gross Margin. Go back to the drop-down menu foroutput measure and select ‘GMROI.’ Selecting Run on the Analyze drop-down menu shows thatin order to match the offshore strategy GMROI of about 2.0 with QR1, the wholesale cost couldrise even further to $16.89)

To see what the wholesale cost with QR1 would need to be in order to obtain a Gross Margin of$68,000 (about $3000 more than that obtained with QR1 with the $12.50 wholesale cost), first re-select ‘Gross Margin’ as the output measure. Then click on ‘Enter Specific Value’ and enter“68000” in the text window which appears. Finally, select Run on the Analyze drop-down menu.

You should find that the wholesale cost would have to drop to $11.77.

When you are finished with testing the Break-Even Analysis feature, close the window.


68

To print the Results table, select ‘Print Output’ on the File drop-down menu at the top of theSimulation Results window.


You may view any of the 11 retail graphs for either scenario individually, or, more importantly, youcan plot results from both on the same graph for visual comparison. Let’s look at a few examples.

7.3.7.1 Cumulative ($) Sales

Select 'New Graph’ on the Tools drop-down menu. When the Graph window opens, select Both'Actual' and 'Projected Cumulative ($) Sales’ on the Retail Graphs menu and choose ‘QR1’ on theData Set menu. You will then see the plot of actual versus projected dollar sales from Scenario 1.Now choose ‘Offshore’ on the Data Set drop-down menu. This will add the graphs from Scenario2.

From the graph you can see that from week 9 (adding vertical lines using Format may help) onthe sales from resulting from the QR strategy begin to exceed those from the offshore strategyand, by week 14, the actual sales with offshore fall below the projected, while those with QRcontinue to exceed projected.

7.3.7.2 Service Level

Now go back to the Retail Graphs menu and select ‘Weekly Service Level’. The graph switchesto provide a comparison of weekly service levels for the two sourcing strategies.

You see that while offshore sourcing provides perfect service in the first few weeks (recall that allstock was put on the shelf at the beginning of the season), the level drops off rapidly from week 7.On the other hand, with the exception of the two weeks prior to the delivery of the first reorder,with QR sourcing the service level remains near perfect until the markdown begins in week 18.

7.3.7.3 Inventory

Next select ‘Average Inventory’ on the Retail Graphs menu. The graph switches to display theaverage inventory level in each week for the two sourcing strategies. Here you can clearly seewhy the average inventory figure for the offshore strategy in the Simulation Results table is abouttwice that for the QR strategy.

Click on ‘offshore’ on the Data Set menu to deselect it and note how inventory plot for offshoredisappears from the graph.

Service Level and Inventory, Side-by-Side

Let's display the two previous graphs side-by-side so that you can see at the same time how QRand Offshore compare with respect to both Average Inventory and Service Level.

In the currently open Graph window, display Average Inventory for both strategies. Then select'New Graph' on the Tools drop-down menu to open a second Graph window. In this window,display Service Level for both strategies. Slide the windows around so that both graphs areclearly visible.

69

You can see that, by having the right mix on the shelf, QR is able to maintain a very high level ofservice with much less inventory.


Recall that selecting Print Graph on the File drop-down menu at the top of the (currently selected)Graph window and clicking on ‘OK’ in the small window which opens will print the currentlydisplayed graph.


Selecting ‘Save Graph Data’ on the File drop-down menu at the top of the graph window will savethe full set of graphs for the scenario in a comma delimited, text file which may be imported into aspread sheet. A window will automatically open asking for a file name (with a “.txt” extension).

To save the graph (as a graph) into a file which may be imported into a word processor documentselect ‘Print Graph’ on the File drop-down menu at the top of the graph window and click on ‘Printto File’ in the small window which opens.

The “.txt” files cannot be opened in the Sourcing Analysis Model. However, whenever you save aSimulations Results table in a “.out” file, the data for all the associated graphs is also saved. Toview the graphs, you first open the “.out” file in the Simulations Results window and select Tools,then ‘Graphs.’

7.3.10 Saving (Deleting) Tabled Results

Recall that if you wish to save the results table to a text file which can be imported into wordprocessor you should select ‘Save As’ on the File drop-down menu and specify “.txt” as thefilename extension. The saved file will contain both columns of results.

To save the results in file that can be reopened in the Simulation Results window, you shouldselect ‘Save As’ on the File drop-down menu and give the filename a “.out” extension. The savedfile will contain both columns of results along with all associated input and graph data.

NOTE: to see side-by-side the results from two (or more) scenarios in the Simulation ResultsTable, it is not necessary to save and open results files as we have done so far. As long as youdon’t close the Simulation Results Table window, you can return to the Data Entry window (byselecting 'Sourcing Analysis' on the Window drop-down menu), modify the most recent scenario’sdata, run the simulation with the new data, and view the results along with those for all scenariosrun since the last time you closed the Simulation Results window.

It is also possible to delete columns of results. To do so, select the column in the table and thenselect ‘Delete’ on the Edit drop-down menu on the Simulation Results window.

To illustrate, delete the “Offshore” column from the current table and leave the SimulationResults window open. We’ll add to it in Scenario 3.

7.4 Scenario 3 - Vendor Managed Inventory Sourcing

(More on Data Entry and Scenario Comparison)

For the third scenario we shall edit the sourcing strategy data for Scenario 1 (QR1) to create aVMI scenario and display the results with those from QR1 (and Offshore).

70

7.4.1 Data Entry

We’ll pick up from where we left off with Scenario 2. In particular the Simulation Results windowshould be open and display the column of results labeled ‘QR1’ and the data entry window shouldcontain the data from Scenario 2.

If it is not showing, click on the Data Entry window (or select 'Sourcing Analysis' on theWindows menu) so it comes to the front.

Again, editing the sourcing strategy data for Scenario 1 will create the new alternative. This timewe won’t describe the other tabs.

First we need the data for Scenario 1 to be displayed in the Data entry window. Select ‘Open’ onthe File drop-down menu and choose ‘default.saf' in the small window that opens.

Now we will proceed with editing the data.


In this example the season’s offering will be delivered in two deliveries, one prior to the beginningof the 20-week Selling Cycle and the other at the end of week 10. The first (initial Inventory)delivery will consist of 60% percent of the Buyer’s Plan volume in the Buyer’s Plan mix. Thesecond delivery will contain the remaining 40% of the Buyer’s Plan volume but in a mix thatmatches that observed in actual sales through week 8.

To specify this strategy, first select the 'Vendor Managed Inventory' sourcing strategy. Then editthe Initial Inventory to be 60 ('Percent of Plan') and enter ‘5’ as the Additional In-Season Capacity(%). Recall that the buyer’s plan volume is 10% low. The additional 5% capacity will permit thestore to order and receive 5% more of those SKUs that are selling well.

Next, choose the ‘User-defined’ reorder schedule specification mode. In the Delivery Schedulewindow which appears, enter 1 for ‘Number of Reorders’, followed by 8 for ‘Week to PlaceReorder’, 10 for’ Delivery Week’, and 40 for ‘% to Reorder’.

Date entry for Scenario 3 is now complete and you are ready to run.


Select ‘Run’ on the Simulate drop-down menu. Accept the default run name, VMI1,and defaultsingle value for volume error by clicking on ‘OK’ in the ‘Run Name and Volume Error’ window. Ina few seconds the results from this scenario should appear in the Simulation Results window nextto those (you left) from the QR scenario.

7.4.3 Looking at the Tabled Results

You should now feel pretty comfortable interpreting results in the table, so we won’t go into detail.

However, a quick glance reveals that the specified VMI strategy, like the previous offshorestrategy, did not match the level of performance obtained with QR. But, how much better/worsewas VMI than Offshore?

71

In order to answer this we need to recall the Offshore results. If you had saved them in a “.out”file we could open it and have the results added to the current table. Unfortunately, you didn’t.However, recall that you did save the data file for Offshore. So, you can regenerate the results.

Click on the Data Entry window to bring it into the foreground. Then select ‘Open” on the Filedrop-down menu and choose ‘offshore.saf’ in the window which opens. Select ‘Run’ on theSimulate drop-down menu. Edit the Run Name to read “Offshore”. In a few seconds the resultsfor the offshore scenario will appear in the third column of the Simulation Results table.

7.4.4 Editing the Results Table

Let’s examine a number of the ways you can modify the results table.

To illustrate the switching of columns in the table, suppose that you wish the order to be Offshorethen VMI then QR. First double-click on the Offshore label and then choose VMI 1 on the drop-down menu which appears. Then double-click on the QR 1 label and choose VMI 1.

To illustrate the elimination of a scenario, suppose you are no longer interested in displaying theVMI scenario results. First click on the VMI 1 column and then choose Delete Column(s) on theEdit drop-down menu.

To illustrate how to change the set of measures which are displayed, suppose that you do notwant to see ‘Inventory’ measures other than ‘Average’ and ‘Turns’, or the ‘In Stock’ measures. Toeliminate them, select 'Retail Outputs/Customize' on the View menu. Simultaneously select (sincethey are adjacent in the display) the three unwanted ‘Inventory’ measures and then click on "<".Then simultaneously select (including the row with the ‘IN STOCK’ label) and click on "<".Finally click on OK. The table display will be refreshed without the selected measures.

Recall that the reduced set becomes the default set of displayed measures until you again editthe set of displayed measures or close and reopen the Simulation Results window.

To restore one of the deleted measures, say Total Offering, select 'Retail Outputs/Customize' onthe View menu. Then select 'Total Offering' on the Hidden Outputs list and click on ">" followedby OK. To simultaneously restore the rest of the deleted measures, repeat the process but clickon ">>" followed by OK.

Finally let’s create a small “tailor made” results table containing only five measures by first clearingthe entire table and then restoring the five measures of interest. First clear the table by selecting'Retail Outputs/Customize' on the View drop-down menu and clicking on "<<". You will see that allmeasures and Labels are moved to the Hidden Outputs list. Now select and move, one-at-a-time,the five measures you want in your table (say, Inventory Turns, Service Level %, % Liquidated,Revenue/Unit Offered, and GMROI). Then click on OK to see the result.

Note: Regardless of the sequence in which they are selected, the measures will appear inthe same order as in the original table.


You should also be pretty comfortable with displaying and interpreting graphical results. Note thatyou can choose to plot results from all three scenarios (not just two at a time) in the same graph.

For example, after opening the Graph window, select the ‘Inventory Receipts’ and select each ofthe three data sets.

The graph that appears shows the marked difference in the inventory receipts among the threestrategies.

72

7.5 Scenario 4 - NewsBoy SourcingIn this scenario we’ll edit the sourcing strategy data used in Scenario 3 to create a NewsBoystrategy and compare (some) results with straight QR.

7.5.1 Data Entry

We’ll pick up from where we left off with Scenario 3. Without closing the Simulation Resultswindow, select 'Sourcing Analysis Data Entry' on the Windows menu to return to the Data Entrywindow that should contain the data from Scenario 3.

(If it doesn’t, use the File drop-down menu to open ‘default.saf.’)


In this example the initial inventory level and the replenishment order quantities will be specifiedwith the goal of achieving customer service levels that you enter. Otherwise the strategy willoperate like QR with regular reorders taking into account POS information to date.

On the Replenishment Strategy drop-down menu select ‘NewsBoy’. In the Vendor Requirementsdialog box to the right you will see ‘In-Season Service Target (%)’ appearing along with ‘thedefault minimum order quantities.

First, you (may) need to reset the Reorder Schedule to be ‘Fixed Lead Times’ with 14 reorders,the first in week 1, 1 week between, and a lead time of 2 weeks.

Then select ‘Service Target’ for ‘Order Method’ in the Initial Inventory box and enter “98” in thetext box. You have indicated that you want the beginning-of-season stock level for each SKU setwith the objective of providing a 98% service level in the first (in this example) 3 weeks until thefirst reorder will be delivered.

Now enter “95” in the ‘In-Season Service Target’ text box. In this case, you are indicating that youwant the quantity of each reorder for each SKU set with the objective of providing a 95% servicelevel over the (3-week in this example) interval from the time the order is placed through the weekprior to that, in which the next reorder is received (In the case of the 14th reorder the interval forservice level coverage is from week 15 through week 20).

Leave the rest of the entries as they are.

Data entry for Scenario 4 is now complete and you are ready to run.


Select ‘Run’ on the Simulate drop-down menu and accept the default name. Make sure that'Single Value' is selected for volume error. The results will appear in the Simulation Resultswindow.

7.5.3 Looking at the Results Table

73

Since we just want to compare results with those of the QR1 run (Scenario 1), first select andclear all columns in your Simulation Results table which correspond to strategies other than QR1and NewsBoy1.

(If you had already cleared the table after running Scenario 3, open the file "QR1.out" thatyou saved for Scenario 1 to get the QR1 results in the table again.

Looking at the table, you see that by most measures the performance of the two strategies is veryclose, with neither being best on all measures.

The most noticeable differences occur in the inventory and service measures.

Specifying the Initial Inventory on the basis of a target service level prior to receipt of the firstreplenishment order resulted in far fewer units being placed in stock at the beginning of theseason, and far more ordered as replenishment stock. This, in turn, resulted in a much lowerAverage Inventory (higher turns) with the NewsBoy strategy and a correspondingly higher GMROI.

Although the Initial Inventory for NewsBoy was nominally set to provide a Service Level of 98% upto the time of the first replenishment delivery, because the erroneous Buyer’s plan demandestimates were used to calculate the Initial Inventory levels, the actual Service Level over thisinterval was only 83.5%. However, once sales data became available the Newsboy strategy wasable to achieve the target 95% Service Level over the season. The straight QR strategy with amuch higher Initial Inventory level and use of POS data achieved a higher Service Levelthroughout most of the season.


Look at the inventory and service level graphs to get a better grasp on the above results.

First select ‘New Graph’ on the ‘Tools’ drop-down menu. Then select ’Average Inventory’ on theRetail Graphs menu and click on both data sets on the Data Set menu.

You will see plots of the Average Inventory level for each week for each of the two strategies. Thelower initial and average inventory levels with the NewsBoy strategy are quite apparent. You canalso see how NewsBoy maintains a rather constant stock level from the time of receipt of the firstreorder until the receipt of the last, whereas with the QR strategy the stock level is allowed to dropoff gradually from the much larger initial inventory level. The final order with NewsBoy is verylarge since it trying to adjust the stock level at the end of week 16 to cover the last 4 weeks of theseason (ignoring the potential impact of markdown).

Now switch category to ‘Weekly Service Level’ to see a comparison of the Service Levelsprovided in each week by the two strategies. (Try adding grid lines using the Format menu.)

From this graph you can see how buyer's underestimate of demand volume causes the servicelevel for NewsBoy strategy to fall way off until the first reorder arrives, even though the overalllevel for the season is near 96%. In contrast, with the higher initial inventory the QR strategyprovides a high Service Level throughout the season until into the markdown period.

Opening a second Graph window and again displaying the Average Inventory for both strategies,allows you to see the comparative impact of inventory level on service level.

7.6 Scenario 5 A Basics Scenario using the Model StockStrategy

74

In this example we illustrate the use of the Sourcing Analysis Model on the retailing of a basicgarment over a 3-year period. The operational sourcing strategy in the example is Model Stock.

Open your Sourcing Simulator file and click on the Sourcing Analysis button to open theSourcing Analysis Data Entry window.

When you have done this, proceed with reading.

7.6.1 Data Entry

Although we will use mainly the default data, we will go through the entries on the six tabs is somedetail to make clear the proper interpretation in the multiyear context.

Recall that data entry/edit should begin with the Buyer’s Plan data. If it is not showing, click onthe Buyer’s Plan tab.

Number of Simulation Replications

At the top of the data entry window, reset the number of Simulation Replications to be 3. Thismeans that the 3-year scenario will be replicated 3 times (i.e., repeated with 3 different randomcustomer streams) and the results averaged.

Buyer’s Plan Data

In this example, as with the previous, the product line offered is ‘garment’ and consists of twostyles each in four colors and six sizes, making a total of 48 SKUs

Set ‘Weeks in Selling Cycle’ to 52 and ‘Number of Cycles to Repeat’ to 3 to specify 3-yearduration of the scenario.

Set ‘Display Floor Space’ to 75.

The buyer anticipates selling 4800 units per year. Of these, the buyer anticipates that 60% will bestyle 1, leaving 40% style 2. The color and size mix figures should be interpreted in a similarmanner. (The ‘Individual SKU Plan Mix‘ option would be selected if you wished to enter thefigures for each of the 48 SKUs rather than have them computed from the attribute figures.)

There is no ‘Presentation Stock‘ requirement.

The buyer anticipates a ‘MidPeak’ seasonality with the weekly percents rising until around week26 and then tapering off in each year, as pictured in the small graph.

You may now proceed through the tabs in any sequence.

Consumer Demand

The actual number of customers per year that the Simulator randomly generates and their SKUpreferences need not match the volume, mix, or seasonality in the Buyer’s Plan.

In this example, the actual volume is specified as a 10 percent error (deviation) from the figureentered on the Buyer’s Plan tab, meaning that the buyer’s figure is low.

75

The initial actual mix is specified on the basis of percent error from the corresponding figures inthe Buyer’s Plan, specifically, 40% for Style, 40% for Color and 20% for Size.

Select Demand Twist and enter 5, 5, and 10 for ‘Demand Twist Factor’, ‘Range of TwistDeviation’, and ‘Demand Twist Period’, respectively. This will result in the actual mix percentsbeing randomly adjusted (by small amounts based on the specified ‘Demand Twist Factor’ and‘Range of Twist Deviation’) every 10 weeks.

The default selection for actual annual demand seasonality is ‘Mid Peak’ which is the same as the‘Presumed Seasonality’ in the Buyer’s Plan (even though ‘Same as Presumed Seasonality’ is notchecked).

Recall that, the ‘Customer Behavior after Stockout’ data govern the (random) behavior ofcustomers who have not found their desired SKU in stock. In particular, throughout the 3 years,24% of those who encounter a stockout look for another choice.

Cost Data

The wholesale price paid to the vendor(s) in this example is $12.50 per unit for both the initialstock and replenishment stock for all SKUs. In turn, all SKUs are offered in the retailer’s store atthe same base (or “first”) price of $25.50. Selecting 'Markup %' on the drop-down menu under"Retail" shows that $25.50 represents a 104% markup over wholesale cost. Unsold units will beliquidated at the end of the third year for $8.00 each.

Associated with each delivery to the store (i.e., order received, including the initial drop) there is afixed shipping cost of $100.00 plus $1.00 per unit delivered. In addition there is a $25.00 costassociated with each order. There is no overhead associated with managing this particularprogram.

The cost of carrying inventory is computed as 20 cents on the dollar (wholesale unit cost) peryear.

The handling charge to prepare and place each unit on the shelf is computed as 8 cents on thedollar (wholesale unit price).

Markdowns/Promotions

In this example, there will be one promotion and one end-of-year markdown to clear out stock ineach year.

The promotion is a “back-to-school” promotion, scheduled to begin in week 30 and last for 2weeks. During this period, the base (‘Retail’) price will be marked down 25%. Once the reductionis in effect, the percent of customers who will look for an alternative SKU when they encounter astockout will be 50% (up from 24%). Set Number of Promotions to 1 and enter the data (asneeded) in the "Promo 1" row in the Markdown Schedule box.

The markdown is scheduled to start in week 47 and last through the end of the year (6 weeks). Inthis case the base (‘Retail’) price will be marked down 40%. Once the reduction is in effect, thepercent of customers who will look for an alternative SKU when they encounter a stockout will be65% (up from 24%). Enter the data (as needed) in the MD1 row.

The price elasticity figure that will be used to adjust the expected demand (customer arrival rate)in weeks when the price is reduced is 0.7 which means each 1% reduction in the base price willresult in 0.7% increase in expected demand.

Sourcing Strategy

76

Select ‘Model Stock’ as the Replenishment Strategy.

Click on the Buyer's Plan button that appears in the Initial Inventory window. This jumps you backto the Buyer’s Plan tab to specify the model stock levels for the SKUs. Note that the originalwords, ‘Presentation Stock’, have been replaced by the words, ‘Model Stock’, and the text box ischecked.

We’ll specify the Model Stock, as is often the case, by specifying levels for each size. To do this,select ‘By Size’ in the drop-down window under the words, ‘Model Stock’, and enter the stock levelfigures in the boxes which appear next to the ‘Buyer’s Planned Percents’ for sizes. Enter4,8,12,16,12,8 for sizes 1-6, respectively.

The resulting Model Stock will consist of 480 units, 60 for each style-color combination.

Now click on the ‘Sourcing Strategy’ tab to return. Enter 49 as the number of reorders.

The retailer will place a reorder (to replace sales out of the model stock) at the end of each ofeach week, starting with the first week of the first simulated year. Individual SKUs can be ordered1 unit at a time and an order can be placed for a single unit (i.e., there are no minimum orderrestrictions in this example). Each reorder will be received after a fixed lead-time of 2 weeks.

Vendor Specification

The manufacturer is a "perfect supplier" and the raw material is differentiated by color.

Data entry is now complete and you are ready to run the simulation.


To run the simulation simply select ‘Run’ on the Simulate drop-down menu at the top of the top ofthe window. Accept the default name, ‘Model Stk 1’ and default 'Single Value' for volume error byjust clicking on ‘OK’.

In a few seconds (running 3 years versus one season takes more time) the Simulation Resultswindow will open showing the results for this scenario.

7.6.3 Interpreting the Simulation Results Table

The Simulation Results Table displays the average (over the 3 replications) values for the full setof measures of performance (mostly annualized) for the 3-year interval. Specifically,

5465 customers per year entered the store and interacted with the stock.

At the beginning of the first week of year 1 there were 480 units on the shelf. On average 4747additional units were received and placed on the shelf each year. On average the shelf stock ranat about 409 units, with 12.02 turns per year.

97.8% of units offered were sold during the average year.

On average over the 3-year period 91.4% of the SKUs were in stock. Prior to the delivery of thefirst reorder (in week 4 of year 1), the in stock percentage ran at 91.7%. Over the 3-year period12.2% of the customers left empty-handed (lost sales). Up until the delivery of the first reorder,lost sales ran at 7.2%. For the entire 3 years the service level was 82.6%; prior to the receipt ofthe first reorder it ran at 85.2%.

77

The average total revenue per year was $116,049. Dividing by total offering gives a revenue perunit of $23.65.

The wholesale cost of the units offered was $61.343 per year. In addition, the store incurred(annually) $9841 in shipping costs, $1233 in additional ordering costs, $4907 in handling costs,and inventory carrying costs of $1021

The gross margin was $54,706 per year, which is 85.7% of the maximum that could have beenobtained from the offering if all units had sold at first (base) price. The associated GM% (dividinggross margin by total revenue) was 47%. The associated adjusted gross margin (subtracting thecosts of shipping, handling, and carrying) was $37,703. The associated GMROI (dividing by thewholesale cost of the average shelf stock level) was 10.72, while the associated GMROISL(multiplying GMROI by In Stock %) was 9.79.

7.6.4 Viewing the Inputs

If you wish to recall the inputs leading to the results in the table, you can see them by using theTools drop-down menu.


If you wish to print the Results table, select ‘Print Output’ on the File drop-down menu. You willget a table with a format similar to that in the window.

7.6.6 Saving the Results Table

You can save the tabled results in a system file, which allows you to view them at a later time inthe Simulation Results window, or as a text file which you can import to a spreadsheet (spacedelimited).

To save the results, select ‘Save As’ on the File drop-down menu. If you wish the results to besaved for future viewing, your filename should have the extension “.out”. To save them in a textfile, your filename should have the extension “.txt”. The default extension is “.out”. To switch to“.txt”, select it on the ‘Save File As Type’ drop-down menu.

In order to make it easier to recall the scenario associated with the results, you may change thelabel at the top of the column of results by double clicking on the label (to highlight it), deleting thecurrent label, and entering the new label. Alternatively you may mark the label for deletion byselecting the column and then choosing ‘Change Column Label’ on the Edit drop-down menu.


The Sourcing Simulator provides graphs of a number of performance measures. These valuesare plotted week-by-week over the entire 3-year interval.

To see the graphs, click on the 'New Graph' on the Tools drop-down menu. Select ‘Model Stk 1‘as the Data Set and select the graph you wish to see on the Retail Graphs menu.

Since you already familiar with the graphs we’ll just look at a couple of examples.

Actual and Projected Cumulative($) Sales

Selecting both lets you see that beyond the first 6 to 9 months the actual dollar sales to date lagwhat was expected from the buyer’s plan (i.e., projected) and that the gap widens with time. This

78

is not too surprising given that the Model Stock levels were based on the Buyer’s Plan which hada 10% (low) volume error and 40-40-20% mix error.

Purchases and Customers

Choosing ‘Customers’ displays the number of customers that entered the store in each of the 156weeks. The ‘Mid Peak’ annual seasonality pattern is evident. This pattern shows up in the graphsof several other measures as well.

Adding ‘Purchases’ shows the significant mismatch between supply and demand which producedthe high ‘Lost Sales’ and low ‘Service Level’ figures in the table (and graphs, if you care to look).

% Volume Error and Raw Material Mix

Since there is no demand re-estimation with the Model Stock strategy, the ‘Volume Error’ graph isnot provided and ‘Estimated’ raw material mix graphs are flat. However, by looking at the graph of‘Actual’ raw material mix you can see how the ‘Twist’ data input on the Consumer Demand tabcaused the actual demand mix to shift throughout the 3-year interval.


Selecting ‘Print Graph’ on the File drop-down menu and clicking on ‘OK’ in the small windowwhich opens will print the currently displayed graph.


Selecting ‘Save Graph Data’ on the File drop-down menu will save the full set of graphs for thescenario in a comma delimited, text file which may be imported into a spread sheet. A window willautomatically open asking for a file name (with a “.txt” extension).

To save the graph (as a graph) into a file which may be imported into a word processor documentselect ‘Print Graph’ on the File drop-down menu and click on ‘Print to File’ in the small windowwhich opens.

The “.txt” files cannot be opened in the Sourcing Analysis Model. However, whenever you save aSimulation Results table in a “.out” file, the data for all the associated graphs is also saved. Toview the graphs, you first open the “.out” file in the Simulation Results window and select ‘NewGraph’ On the ‘Tools menu.

Scenario 6 will start from here, so leave the Results Table open (but close the Graphwindow).

7.7 Scenario 6 - Comparing Sourcing Strategies forBasics

In this example we’ll run simulations using the ‘Target Weeks Supply’ and ‘NewsBoy’ strategiesand display the results along with those obtained using the ‘Model Stock’ strategy to permitperformance comparisons.

7.7.1 Data Entry and Execution

79

We’ll pick up where we left off with Scenario 5. In particular, the Simulation Results windowshould be open and displaying the column of results labeled ‘Model Stk 1’ and the SourcingAnalysis Data Entry window should contain the data from Scenario 5.

If it is not showing, click on the Data Entry window (or select 'Sourcing Analysis DataEntry' On the Window drop-down menu) so it comes to the front.

We will be editing only the Sourcing Strategy data, so click on the Sourcing Strategy tab tobring it to the front.

Sourcing Strategy

Lets first specify and run a ‘Target Weeks Supply’ strategy.

Select ‘Target Weeks Supply’ on the Replenishment Strategy drop-down list. A box in which toenter the ‘Target Weeks Supply’ will appear in the Vendor Requirements window. Enter 3.

A ‘Target Weeks Supply’ of 3 means that weekly reorder quantities will be calculated with the goalof having a three weeks supply on the shelf once the order is delivered. Demand and SKU mixwill be re-estimated based on POS data prior to each reorder.

Enter 10 as the 'Percent of Plan' in the Initial Inventory box. This means that the shelf stock at thebeginning of the first week of year 1 will be approximately 480 (10% of the 4800 plan volume) withthe Buyer’s Plan mix.

Date entry for the ‘Target Weeks Supply’ run is now complete. Run the simulation andaccept the ‘Target Wks 1’ label and 'Single Value' volume error.

Now let’s specify and run the ‘NewsBoy’ strategy.

To do so, return to the Sourcing Strategy tab and select ‘NewsBoy’ on the ‘ReplenishmentStrategy’ drop-down list.

Enter 95% for the ‘In-Season Target Service Level’ in the Vendor Requirements box window.Recall that this means that the weekly reorder quantities will be calculated with the goal of beingable to supply at least 95% of the demand from the time the reorder is placed until the nextreorder is received. As with ‘Target Weeks Supply’, reorder quantities are based on the currentdemand estimates.

Select 'Service Target' as the Order Method in the Initial Inventory window and enter 95% toindicate that the initial inventory level should be set so as to try to provide 95% service until thearrival of the first replenishment order.

Date entry for the ‘NewsBoy’ run is now complete. Run the simulation and accept the‘NewsBoy 1’ label. (NewsBoy takes about twice as long to run as the other two.)

When the simulation is complete the results from the three strategies should be displayed in theSimulation Results Table.


You can study the results in as much detail as you wish. We’ll just make a couple ofobservations.

80

Note that the ‘Initial Inventory’ level was roughly the same with each strategy. However, since thedemand estimates used by the strategies to set this level were those in the Buyer’s Plan, whichwas low on volume and off on mix, the service level before the first reorder was received was low(lost sales high). The ‘Target Weeks Supply’ and ‘NewsBoy’ strategies, which re-estimatedemand based on POS data, were able to correct in large measure for the plan error (eventhough the demand mix jumped around somewhat) and provide service levels of better than 92%(with very few lost sales) over the 3-year period. In contrast, since the ‘Model Stock’ was neverupdated from the POS data, the service level remained low (lost sales high) throughout the 3-years.

The superior performance of the ‘Target Weeks Supply’ and ‘NewsBoy‘ strategies is also reflectedin the revenue and margin figures.

Comparing the average inventory and service level figures for the Model Stock' and ‘TargetWeeks Supply’ strategies again illustrates the value of demand re-estimation. 'Target Weeks' isable to provide better service with about 75% of the inventory because the inventory more closelymatched the actual demand.7.7.3 Viewing Graphs of Results

Recall that you can choose to plot results from all three scenarios (not just one at a time) in thesame graph to obtain graphic comparisons.

In particular, you might want to look at the graphs of ‘Service Level’ (both ‘Weekly’ and‘Cumulative’), ‘Lost Sales’ and ‘Inventory’ (both ‘Average’ and ‘Receipts’). Remember that thereare markdowns that draw more customers without additional stock at the end of each year andthe initial inventory levels are reestablished at the beginning of each year.

Scenario 7 will start from here, so leave the Results Table open (but close the Graphwindow).

7.8 Scenario 7 - Using the ‘Presentation Stock’ FeatureIn this example we’ll demonstrate the use and impact of the ‘Presentation Stock’ feature availablewith the Sourcing Analysis Model. The simulation will be rerun using the ‘Target Weeks Supply’strategy with a presentation stock requirement and the results will be displayed along side thoseobtained in the previous ‘Target Weeks Supply’ run.


Select 'Sourcing Analysis Data Entry' on the Window drop-down menu. Bring the ‘Buyer’s Plan’tab to the front and click on ‘Presentation Stock’. Select ‘By Size’ in the drop-down list, and enter6,12,18,24,18,12, respectively, for sizes 1-6 in the cells which appear beside the ‘Buyer’s PlannedPercent’ figures.

On the ‘Sourcing Strategy’ tab select the ‘Target Weeks Supply’ strategy. Enter '3' for number ofweeks supply in the Vendor Requirements box and select 'Presentation Stock' as the OrderMethod in the Initial Inventory box. The dialog box will display 720 units, i.e., the total number ofunits in the presentation stock. The associated percent (15%) of Plan volume (4800) will also bedisplayed.

The initial inventory will be the presentation stock and, if necessary, the presentation stockquantities will override the ‘Target Weeks Supply’ figures in specifying reorder quantities.

81

Run the simulation accepting the default run name and single value for volume error. Theresults should appear in the 4th column of the Results Table.


In order to focus on the impact of introducing the presentation stock into the ‘Target WeeksSupply’ scenario, get rid of the two columns of results corresponding to the other strategies byhighlighting the associated columns and selecting ‘Delete’ on the ‘Edit’ drop-down menu.

Looking at the Table you can see that the presentation stock requirement had a definite impact onthe results. The initial stock level is significantly larger than 10% of the expected sales. This hadthe positive impact of better service (fewer lost sales) prior to the receipt of the first reorder at theend of week 4. The higher service level early on (permitting an earlier accurate fix on the actualdemand) in combination with the requirement to maintain minimum stock levels (which are largerthan a 3-week supply for some SKUs) resulted in a larger total offering. The minimum stocklevels also resulted in a much higher average inventory level and correspondingly lower turns.

The lower ‘% Sell Thru’ and ‘Revenue/Unit’ indicate that minimum stock requirements forced thestore to carry more units of the poor sellers.

Over the 3-year span the service level was raised close to 99% with only 0.9% lost sales.

Higher sales resulted in higher revenue and gross margin. However, the much higher averageinventory caused reductions in ‘% of GM Potential’, GMROI, and GMROISL.

Viewing Graphs of Results

To get a clearer picture of these results you should explore the associated side-by-side graphs forthe two runs, especially the graphs for ‘Purchases’, ‘Cumulative Service Level’, ‘Lost Sales’, and‘Average Inventory’.

7.9 Scenario 8 - Creating a Small Decision SurfaceModel for a Basics Scenario

To introduce the use of the Decision Surface Model, in this example we'll make a small batch runfor a 3-year scenario with the 'Newsboy' sourcing strategy. In the batch run we'll range theminimum order quantity per SKU and use the Decision Surface Model to visualize the impact onseveral of the performance measures.


If it is not showing, open the Sourcing Analysis Data Entry window.

We'll will make data entries/edits on three of the tabs.

Buyers Plan Data

Go to the Buyer's Plan tab and set 'Weeks in the Selling Cycle' to 52 and 'Number of Cycles toRepeat' (above the tabs) to 3 to specify the 3-year scenario. Also set 'Simulation Replications' to3. With long selling cycles, fewer replications are needed. Beyond this we will use default plandata.

82

Sourcing Strategy Data

Go to the Sourcing Strategy tab and select the 'Newsboy' strategy. Change 'Number of Reordersto 49 so that the last reorder will be available at the beginning of the last week of the year.

Select "Service Target" for 'Order Method' in the Initial Inventory box and enter "95" as thepercent. Also " 95" for the 'In-Season Service Target'. This indicates that reorder size and mixshould be aimed at providing a 95% service level.

Beyond this we'll use default sourcing strategy data

Markdowns/Promotions

Go to the Markdowns/Promotions tab and enter 1 for 'Number of Planned Promotions' and 0 for'Number of Planned Markdowns'. Then, 'Week Occurs' to 30 and 'Duration' to 2 in the tablebelow. There will be one two-week promotion starting in week 30 of each year and no markdownsto sell off stock. Leave the other data as is.

The next step is to specify and execute the batch run to create the data for the Decision SurfaceModel.

Select 'Batch Run' on the Simulate drop-down menu.

A window labeled "Select Inputs to Range" will open providing a list of those inputs to the SourcingAnalysis Model that you may select as inputs for a decision surface model.Select ‘Min Order Qty/SKU’ and then click on the Add button to select it for ranging.

In the small "Input Information" window that opens the dialog boxes under Min contains thecurrent (single) value for 'Min Order Quantity/SKU' from the Sourcing Strategy tab. Enter 36 in thedialog box that appears under Max and enter 1 in the box under Increment. Then click on OK toenter the ranging data into the table.

You have now specified a batch run with 36 runs, one for each value of ‘Min Order Qty/SKU’ from1 through 36. Note that the total number of runs in the batch (36) appears in the upper rightcorner of the window.

The default label for the batch run is "Ne Ki" which is short for "Newsboy strategy with minimumorder quantity per SKU equal to i". “Ne K1” – “Ne K36” will be used as the column labels in theSimulation Results Table.

Next, click on the "Create DSM File" box to indicate that you are making this batch run to providedata for a decision surface model. (If you don't do this, the batch run results will only be displayedin the Simulation Results Table and will not be accessible to the Decision Surface Model)

Now click on the 'Select Performance Measures for DSM' button. A small window labeled 'SelectPerformance Measures for DSM' will open with a list of performance measures. Choose the fourmeasures 'Average Inventory', 'Inventory Turns', 'Service Level', and 'GMROI', by clicking first onthe measure and then on the Add button for each. The four selected measures should appear inthe ' Performance Measures Selected' list. (If you wanted to change your mind and delete one ormore you could do so by selecting the measure and then clicking on Delete. But don't do it now)

Data Entry is now complete. Click on 'OK'. In the 'Select Performance Measures for DSM'window and then click on 'Run'.

83

A window labeled "Save As" will open. You need to enter a name for the DSM file that the batchrun will generate. Enter the name, say, "minord", (The ".dsm" will be added automatically) andclick on 'Save' to save the file and start the execution of the batch run.

In a few seconds the Simulation Results Table will open and the results from the 36 runs in thebatch will appear, column by column, as the runs are completed. So that you can see theprogress, the number of runs remaining is displayed on the menu bar at the top. The columns arelabeled with the name “Ne K1” through “Ne K36”. The “i” in “Ki” indicates the value of 'Min OrderQty /SKU' for that run.

When the batch run is complete, a small window will open asking if you would like to move directlyto the Decision Surface Model. Click on 'Yes'.

Click on "no" in response to being asked if you would like save the Simulation Results Table.

At this point the Model Specification window of the Decision Surface Model will open.

7.9.2 Specifying the Decision Surface Model

The Model Specification window is displaying the Inputs tab containing the label 'Min OrderQty/SKU', its range from the batch run (Data Range), and a pair of dialog boxes displaying thesame range (as Selected Range).

Unless you have previously hidden it, a small "Decision Surface Model-Wizard" will appear. Makeas much use of the wizard windows as you wish. To hide them click on 'Always Show' on theWizard Main Menu. To restore them click on the Wizard button in the current window.

In the Model Specification window click on 'Min Order Qty/SKU' to select it as the (only) input foryour decision surface model and click on the Performance Measures tab to bring it to the front.

The Performance Measure tab displays the labels for the four measures you selected beforemaking the batch run, namely, 'Average Inventory', 'Inventory Turns', 'Service Level', and'GMROI'. These are the measures that may be selected as outputs, or "responses" for yourdecision surface model. In this case select all four by clicking on them.

You have now specified a decision surface model that will relate each of the four measures to 'MinOrder Qty/SKU' over the range from 1 to 36 units.

At this point you should click on the 'Create Surface' button at the bottom of the window to initiatethe fitting of the model to the data. A progress bar will appear.

When the model is complete, the Plot Specification window will open.

7.9.3 Saving the Decision Surface Model Data

If you think that you might wish to make plots from this model at a future time, you may save thedata that will allow you to do so without having the having to specify and create the model again.

To do so select 'Save As' on the File drop-down menu (in the Plot Specification window), enter afile name in the small window which opens, and click on 'Save'. The file will be saved as a ".wts"file in a subdirectory labeled "DSM Data".

For future reference, go ahead and save the current data in a file named " minord".

84

7.9.4 Plotting a Surface

Since we came to the Plot Specification window immediately after creating a new decision surfacemodel, the inputs and performance measures for that particular model are displayed.

Note that the 'Min Order Qty/SKU’ label is followed by the range of values used in creating themodel (Data Range) plus the four columns for specifying the values to be used in the plot (PlottingRange). The middle value of the range (18.5 for 1-36) is displayed under the column labeled"Fixed".

Since in this example we have only one input, we can only make 2-dimensional plots of the listedmeasures against 'Min Order Qty/SKU'. Therefore, click on "2-Dimension" and then on 'Min OrderQty/SKU'. Note that midrange value is replaced with the lowest and highest values, plus a step-size for plotting. If you wish you can specify a Plotting Range that is smaller than the Data Rangeby increasing the lowest value and/or reducing the highest value. You can also change thescaling in the plot by changing the step-size. At this point, let's stay with the original figures.

To complete the specification of a plot, we must select one of the performance measures.At this point select ‘Inventory Turns’ and click on the Plot button at the bottom of the window.Supply an identifying title to appear in the plot, e.g., "minord" and click on "OK" to display theactual plot.

The plot that appears shows that larger ‘Min Order Qty/SKU’ requirements result in fewer turns.In the particular scenario of this example, the impact of raising the minimum from 1 unit to 3dozen is to reduce the number of turns from around 10.4 to around 3.3 per year.

7.9.5 Adjusting the Scales on the Axes

Vertical (Inventory Turns) Axis

Note that the numbering on the vertical (Inventory Turns) axis runs from just below the smallestplotted turn value to a little above the largest. Let's change it to run form a low of 0 to a high of 12.To do so select 'Scale Y Axis' and 'Manual' on the Format drop-down menu in the Plot window. Inthe "Manual Axis Setup" window that opens enter 0 and 12 as the minimum and maximum values,respectively, and click on OK at the bottom of the window.

The display will be refreshed with the modified scaling.

Horizontal ( Minimum Order Qty/SKU) Axis

We can also adjust the scaling (increase the minimum, decrease the maximum, and/or changethe step-size) on the horizontal (Minimum Order Qty/SKU) axis. However to do so we must go tothe Plot Specification window. So click on it to bring it to the front.

Lets reduce the displayed range from 1 to 2 dozen and reduce the step-size to 2 units. Go aheadand edit the entries and then click on Plot to display the new graph.

Note that the minimum value on the inventory turns axis shifted from 0 to near the smallest valuein the plot. If you want to adjust the scaling on both axes, you should start with the horizontal axis.

7.9.6 Switching the Displayed Measure

To replace ‘Inventory Turns’ with one of the other three measures you can close the Plot window,change your selection on the list in the Plot Specification window, and click on the Plot button toreopen the Plot window with the new graph.

85

Alternatively, you can simply select the other measure on the Category drop-down menu in thePlot window itself. Try this to replace ‘Inventory Turns’ with ‘GMROI’. In the new plot you can seethe drop in GMROI corresponding to the reduced turns that accompany the larger minimum orderquantities.

7.9.7 Printing and Saving Plots

To get a printout of the current plot, simply select ‘Print Graph’ on the File drop-down menu in thePlot window. You might want to try it.

To save the current plot, select ‘Save Graph’ on the File drop-down menu and provide a file namein the Save As window which opens. The plot will be saved in a Plots subdirectory as a “.WMF”file (unless you select a different type). Go ahead and save the GMROI plot as, say,“gmroivsminord”.

7.9.8 Starting from Different Points in the Process

In this example we have illustrated the complete decision surface modeling and plotting process,beginning with specification of the batch run for the Sourcing Analysis Model to generate the data.Depending on the information you saved, you can get the same plots by:

1. Opening a “.dsm” file in the Model Specification window, specifying and creating themodel, specifying the plot(s) and plotting.

2. Opening a “.wts” file in the Plot Specification window, specifying the plot(s) andplotting.

3. Opening a saved plot.

Let’s look at each using the files we have saved in going through this example. Close eachwindow until you get back to the main menu for the sourcing simulator.

Starting with a “.dsm” File

Click on the Decision Surface Model button.

The Model Specification will open with no data displayed.

Select ‘Open’ on the File drop-down menu. Then select ‘minord.dsm’ and click on ‘Open’ in thesmall window which appears to display the input and performance measure lists that weredisplayed automatically before.

From here you can proceed as before to specify the same or an alternative model and plots.

Starting with a “.wts” File

Starting from the main menu, click on the Decision Surface Model button to open the (empty)Model Specification window.

Click on Plot Specification on the menu bar to open the (empty) Plot Specification window.

Select ‘Open’ on the File drop-down menu. Then select ‘minord.wts’ and click on ‘Open’ in thesmall window which appears to display the input and performance measure lists that weredisplayed automatically when you came to this window after having specifying the model before.

86

From here you can proceed as before to specify and display the same or additional plots.

Starting with a Stored Plot

Starting from the main menu, click on the Decision Surface Model button to open the (empty)Model Specification window. Then click on Plot Specification on the menu bar to open the(empty) Plot Specification window.

Select ‘View Graph’ on the menu bar. Then select the file named ‘gmroivsminord.’ and click on‘Open’ in the small window which appears to display the plot of ‘Inventory Turns’ against ‘MinOrder Qty/SKU’ that you saw before.

7.10 Scenario 9 - Creating a Bigger Decision SurfaceModel for a Seasonal Scenario

In this example we'll create a bigger Decision Surface Model --one having 3 inputs—for a 20-week seasonal scenario using the default data (QR sourcing strategy). This will permit us todisplay 3-D plots of performance measures against selected pairs of inputs.


Open the Sourcing Analysis Model to display its Data Entry window.

Since we are using default data for the underlying scenario, just select 'Batch Run' on theSimulate drop-down menu.

Let's examine how the retail performance is impacted, in combination, by 'Volume Error Percent','Initial Inventory Percentage', and 'Reorder Lead Time'. To do so select and add each of them,ranging 'Volume error Percent' from -20 to 20 in steps of 10; ranging 'Initial Inventory Percentage'from 30 to 60 in steps of 10; and ranging 'Reorder Lead Time' from 1 to 4 in steps of 1. Note thatthis specifies a batch run with 80 runs (5 x 4 x 4).

Click on "Create DSM File" and then on 'Select Performance Measures for DSM'.

To provide a wide range of possible plots which can be generated from the decision surfacemodel, select and add the measures 'Average Inventory', Inventory Turns', 'Service Level', 'TotalRevenue', 'Gross Margin', 'Adjusted Gross Margin', Adjusted Gross Margin', 'GMROI', and'GMROISL'.

Then click on 'OK' followed by 'Run'

Name the DSM file, say, “qrthreein”, and click on ‘Save’ to start the batch run. With 80 runs in thebatch, the execution will take more time than that for Scenario 8.

After the last run, select ‘Yes’ in the first window which opens and ‘No’ in the window which followsto go directly the Decision Surface Model.

7.10.2 Specifying the Decision Surface Model

The Model Specification window is displaying the Inputs tab containing the label of each of thethree inputs (‘Volume Error Percent’, ‘Initial Inventory Percentage’, and Reorder Lead Time’), theirranges from the batch run (Data Range), and pairs of dialog boxes displaying the same ranges(as Selected Range).

87

You may actually select one, two, or all three of these to be inputs for a decision surface model(and also reduce the Data Range for the unselected variables if you wish). Since we originallyspecified the batch run with the idea of examining the impact of all three on performance, selecteach of them. Note that as you select an input, the dialog boxes vanish, meaning that the fullData Range will be in creating the decision surface model.

Now switch tabs to select the performance measures to be included in the model. To provide thewidest range of possible plots which can be generated from the decision surface model, select allof them.

Click on the ‘Create Surface’ button. The progress bar will appear. Since this decision surfacemodel has two more inputs and four more outputs (measures) than the one in Scenario 8, youshould expect the model fitting process to take longer.

When the model fitting is complete, the Plot Specification window will open.

If you think that you may wish to use this model again, as soon as the Plot Specification windowopens, save the associated “.wts” file using ‘Save As’ on the File drop-down menu. For usebelow, save the model as "qrthreein.wts".

7.10.3 Plotting Surfaces

The Plot specification window lists the three inputs, their respective Data Ranges, and the mid-value of of the ranges (“Fixed” column under Plotting Range) along with the eight performancemeasures included in the decision surface model that you just created.

At this point you may choose to make a 2-D plot of any of the listed performance measuresagainst any one of the three inputs or to make a 3-D plot of any measure against any two of theinputs. In this scenario we’ll illustrate 3-D plots.

Let’s start by plotting ‘Service Level’ versus ‘Initial Inventory Percentage’ and ‘Reorder Lead Time’.

First select ‘3-Dimension’ at the bottom of the window. Then select the two inputs and 'ServiceLevel’. Note that the “Fixed” value of “0” remains displayed for the (unselected) input, ‘VolumeError Percent’, indicating that the plot will be for the case in which the volume error is 0.

Click on the ‘Plot’ button, supply a plot title, and click on OK to start the plotting.

7.10.4 Interpreting and Manipulating 3-D Plots

The plot shows that for this particular scenario the service level increases with shorter reorderlead times and/or higher initial inventories. However, note that the length of the leadtime is more important when the initial inventory percentage is small (e.g., 30%) than when it islarge (e.g., 60%). Similarly, you can see that the impact of initial inventory percentage is muchmore significant with longer lead times (e.g., 4 weeks) than with short lead times (e.g., 1 week).

For future reference, save the plot by selecting ‘Save Graph’ on the File drop-down menu. Nameit, say, slvsispandrlt (short for, “service level versus initial stocking percentage and reorder lead-time”).

You can obtain the coordinates of specific points on the surface by clicking on the corners of thegrid squares on the surface itself. For example, by clicking on the grid point at the very back ofthe plot, you find (in the small window which opens) that with an initial inventory percentage of30% and a lead time of 1 week, the service level is around 95.26%.

88

Note: The graphics package used in the Sourcing Simulator may not allow you to obtainthe coordinates of grid points on the front edges of the plot (i.e., when the lead time is 4 orwhen the initial inventory percentage is 60).

By selecting ‘Grid Lines’ on the Format drop-down menu, you can alter the plot to display onlyhorizontal, only vertical, or no grid lines. Select ‘None’ to see what the plot looks like withoutgrid lines.

You can also rotate the plot (either automatically or manually) to view the surface from a differentangle. To illustrate, select ‘Automatic’ on the Rotate drop-down menu. To freeze the plot at aparticular angle, select ‘Stop’ on the Rotate drop-down menu. You might see if you can stop itwith the original view. Now try the ‘Manual’ rotation option.

You can obtain printouts of different views of the surface selecting ‘Print’ on the File drop-downmenu with the display frozen at these views. You can also save these views (by selecting ‘SaveGraph’ on the File drop-down menu) for redisplay at a later time.

7.10.5 Displaying Different Plots from the Same Decision Surface

Many different plots can be obtained from the current decision surface model. To look at others,simply close the Plot window to return to the Plot Specification window and alter your selections.We’ll illustrate several of the possible alternatives.

Changing Value(s) of Unselected Inputs

Let’s try to get a feel for whether the relation between ‘Service Level’, ‘Initial Stocking Percentage,’and ‘Reorder Lead Time’ is influenced by the third input in the decision surface model, ‘VolumeError Percent’,

To do this, replace the “Fixed” value of “0” for ‘Volume Error Percent’ with “-20” and click on thePlot button at the bottom of the window. The new plot displays the relation between ‘ServiceLevel’, ‘Initial Stocking Percentage’, and ‘Reorder Lead Time’ when the buyer has overestimatedthe demand volume by about 20%.

Displaying Side-by-Side Plots

To be able to display a side-by-side comparison with the 0%-error plot, first save the -20%-errorplot and close the Plot window. Then click on Compare menu bar at the top of the PlotSpecification window to open an empty Plot Comparison window. Next click on Graph followed byAdd to display the list of stored plots. Select the -20%-error plot you just saved and click on Opento display it. Click on Graph then Add again and open the plot named slvsispandrlt. The twographs will be displayed side-by-side in the Plot Comparison window. You will probably want towiden and deepen the window to get a better picture.

From the display you can see that the 20% overestimation results in somewhat higher servicelevels, especially when the initial inventory percentage is low and the lead time is high (30% of theoverestimated volume provides a larger initial stock level to cover the 5-week interval untilreorders begin to arrive at the store).

Re-scaling for Easier Comparison

Notice that the scaling on the vertical (Service Level) axis is not quite the same in the two side-by-side plots. To make the comparison sharper you can re-scale the Service Level axis in one orboth to make the two scales identical. However to do this you need to be in the Plot window andthus you will have to re-plot and save one or both graphs.

89

Let's re-scale both so that the Service Level scale runs from 90% to 98%.

Close the comparison window to return to the Plot Specification window.'3-Dimension', inputs'Initial Inventory Percent' and 'Reorder Lead Time', and the 'Service Level' measure should beselected and 'Volume error Percent' should read '-20'. Click on Plot. Enter the label "-20%Volume Error" and click on 'OK.'

In the Plot window which opens, if you wish, select 'Scale Z-Axis' -'Manual' on the Format drop-down menu. In the Manual Axis Setup window set the minimum and maximum values to 90 and98, respectively and click on 'OK' to refresh the plot with the new scaling. Save the plot

Close the Plot window to return to the Plot Specification window. Change the 'Volume PercentError' to '0' and click on Plot. Enter the label "0% Volume Error" and click on 'OK'.'

Re-scale the Service Level axis to run from 90 to 98 and save the plot.

Now open the Compare Plot window and open (i.e., "Add") the two stored plots. Resize thewindow for a clearer picture.

Changing the Response Measure

Let’s now look at a plot of ‘Inventory Turns’ versus the same two inputs, ‘Initial stockingPercentage’ and ‘Reorder Lead Time’, with a ‘Volume Error Percent’ of “0”.

To see this, in the Plot specification window, set the “Fixed” value of ‘Volume Error Percent’ to “0”.Click on ‘Service Level’ to deselect it. Then select ‘Inventory Turns’ and click on the Plot button.

In the plot which appears (You may wish to rotate it.) you can see that the number of turns fallssignificantly (from a high of around 6.4 with a 30% initial stock level and 4-week lead time) as theinitial inventory percentage increases and/or the lead time decreases. Both a higher initialpercentage and a shorter lead time result in more inventory in the store, especially in the earlierweeks of the season.

If you save this plot, you can reopen it along with the (0%-error) service level plot in the PlotComparison window. From the side-by-side display you can see that higher service levelsaccompany lower turns.

Changing the Ranges and Step-sizes of the Selected Inputs

To display the relation between ‘Inventory Turns’, ‘Initial Inventory Percentage’, and ‘Reorder LeadTime’ over smaller ranges of values and with different step-sizes for the latter two, you need tochange the ‘‘Lowest’’ and/or ‘‘Highest’’ values along with the step-size value in the correspondingPlotting Ranges and re-plot.

To illustrate, with the ‘’Fixed’’ value of ‘Volume Error Percent” at “0” in the Plot Specificationwindow, raise the “Lowest’’ value for ‘Initial Inventory Percentage’ to "40" and reset its step-size to"2". Reduce the “Highest” value for ‘Reorder Lead Time’ to "3" and change its step-size to "0.5".Then click on the Plot button, supply a title, and click on OK.

The new plot is similar to the previous except for the reduced ranges of the two inputs (and thecorresponding reduction in the range of plotted values for turns) and the scales on the two inputaxes.

Changing an Input

90

Next, let’s replace ‘Initial Inventory Level’ with ‘Volume Error Percent’ in the plot of ‘InventoryTurns’.

To do so, first deselect ‘Initial Inventory Percentage’ and then select ‘Volume Error Percent’ in thePlot Specification window. Click on the Plot button.

If the plot that is displayed is not easily read, try rotating it manually until the “Rotation angle” isaround 140 degrees. With this clearer angle, you can see the number of turns decreases withshorter lead times (as before) and decreases as the volume error moves from 20% low to 20%high.

Making a 2-Dimensional Plot

Finally, let’s make a 2-dimensional plot, say, ‘Total Revenue’ against ‘Volume Error Percent’.

To do so, first click on “2-Dimension” at the bottom of the Plot Specification window. Then select‘Volume Error Percent’. Note that the other two inputs are set to their midrange values.

Select ‘Total Revenue’ at the performance measure (you may need to deselect another measurefirst) and click on the Plot button.

The plot shows that the buyer’s volume error can have a large impact on revenue, a swing fromaround $100,000 with a 20% overestimate to over $141,000 with a 20% underestimate.

7.11 Scenario 10 Using the Detailed VendorSpecification

In this scenario we will demonstrate the use of the detailed model of the vendor. Open theSourcing Simulator and click on the Sourcing Analysis Data Entry (Tabs) window. Under File,click on Open and then select the file “Scenario10.saf”.

Next, click on the Vendor Specification Tab. Notice the following:

Raw Material Supply

• Under the Material Supply, Raw Material is different for each color (and from the Buyer’s PlanTab, you can see that there are 4 colors).

• The Min Total Order Qty is 1 so there is effectively no minimum on the entire order of rawmaterial.

• However, when ordering raw material, there is a minimum of 250 units of each color. Thismeans that if more than 0 but less than 250 units a given color are ordered, 250 units will beordered. (This is defined under Minimum Order Qty by Color.)

• The Supplier of Raw Material will deliver an order 3 weeks after an order is made.

Cost Data (per Unit)

You can simply look at this group of inputs. Notice, for example, that the Raw Material costs$2.50 per unit.

Inventory Policy

91

• The Ship Backorders input is not checked, therefore, if the vendor cannot meet a retail order,any short items are not backordered. If the retailer still wants the missing items, they willinclude them in the next reorder.

• The Ship Residual FG Units at Season End input is also not checked indicating that theretailer will only accept what the explicitly order and the vendor will liquidate them at residualunits at the FG Liquidation Price defined in the Cost Data section.

• The Collaboration Policy is “Shared Plan and POS Data” meaning that the retailer shares theirPOS data and demand forecasts with the vendor.

• Under the FG Inventory Policy, Make to Stock is specified meaning that the vendor will onlyproduce finished goods that have been ordered and thus will not carry finished goodsinventory.

Manufacturing Process Data

You can simply look at this group of inputs. Notice, for example, that the Processing Time is 2weeks meaning that a shop order released this week will be finished 2 weeks later.

Next, click on Plan Production under Manufacturer Model. The window shown in Figure 7.1 willappear.

Figure 7.1: Production Planning Window

Notice the red values in the Finished Goods Inventory column. Since we have not specified anyShop Orders or Material Order, there is a net need for Finished Goods Inventory to meet theRequired Shipments.

Click on the white box at week –6 in the Shop Order Release column and type in 500 and hitenter. Notice that 488 now appears in the Completed WIP column in week –4. This is because

92

the Process Time is two weeks and the Dropout Rate is 2.5%, i.e., a shop order released in week–6 will be completed two weeks later with the quantity completed being 2.5% less than what wasreleased (500 minus 2.5% is 488 units). Notice also that the values in the Finished GoodsInventory are now increased by 488 units. Also, the Cumulative Material Requirements columnnow has 1000’s and the Material Inventory –1000’s. These reflect the fact that to produce 500units of finished goods requires 1000 units of raw material (2 raw material units per finished goodunit). Since we have not ordered any raw material yet there is a net need of 1000 units.

Click on the white box at week –9 of the Material Order column and type in 5000. Now a 5000should appear in the Material Delivery column three weeks later in week –6. The MaterialInventory in week –6 is now 4000 (5000-1000 of Material Requirements).

In this way you can enter a plan terms of Shop Order Releases and Material Orders such that theFinished Goods Inventory and Material Inventory do not contain red entries (negatives). However,the Order Planner option allows the computer generate the plan. Under the Tools option on themain menubar, click on Order Planner. Under Pre-Season Shop Order Releases, click on FixedNumber of Orders. Enter 4 releases starting in week –6 with 1 week between releases, then clickon ‘OK’. In the preseason, we want to make enough product to meet the Initial Inventory of 2400.Since we have a 2.5% dropout rate and we want 4 releases, each must be of size 615.

The other choices under the Order Planner are described in Section 4.1.6. Select Backward Loadbased upon Capacity in each of the four boxes (Pre-Season Shop Orders, In-Season ShopOrders, etc.) Figure 7.2 shows what you should see.

Figure 7.2: Manufacturing Plan Based on Backward Load

93

Click back on the Sourcing Analysis Data Entry window. In the menubar on the main window clickon Simulate and then Run (or click on the “running man” icon on the Sourcing Analysis Data Entrywindow. Next click on Uniform Average and use the default values for everything. Click on ‘OK’.

Next change the FG Inventory Policy to Make to Stock with the FG Supply set to 500 ‘Units’ andthen run the program again as above. Compare the results for the Make to Order vs. Make toStock Policies.

Sourcing Simulator Users Guide

Documents