Building an Analytics CoE (Center of Excellence)

Building an Analytics CoE

Rahul Saxena

Workshop at IIM Bangalore

July 20, 2013

This presentation is based on the material available in the book:

Business Analytics: A Practitioner's Guide by Rahul Saxena & Anand Srinivasan

Springer International Series in Operations Research & Management Science

http://www.amazon.com/dp/1461460794

http://images.springer.com/covers/978-1-4614-6079-4.tif



Agenda

Introductions – about the group, about me. What is "analytics"? What is an "Analytics CoE"?

What does the organization want to do with analytics? What is the role of the CoE that they envision?

What is the organizational context? Current providers of analytics? Leadership support?

What will the Analytics CoE need to be like (now and in the future, up to the planning horizon)?

Where do we stand with analytics capabilities now, compared to what we need?

How will we evolve the CoE? Set expectations, drive the evolution, establish the value.

Confidential 2CobotSystems.com

About the group …


STUDENT WORKING

0-5 Years

5-10 Years

10-20 Years

>20 Years

Full-time

student, 0-5

years of

experience

Currently

working, 0-5

years of

experience

Demographics TEACHING

Currently

teaching, 0-5

years of

experience

PROVIDER

ANALYTICS

IT

(BI, DW)

BUSINESS

OTHER

(Academic,

etc.)

Provided analytics

as an analytics

professional

Provided analytics

as part of an IT role

Provided analytics

as part of a

business role

Provided analytics

as part of an

academic or other

role

CONSUMER

Consumed

analytics from

professional

analytics provider

Consumed

analytics from IT

provider

Consumed

analytics from

business provider

Consumed

analytics from

other provider

Perspectives

About me


No

wC

ob

ot Syste

ms

2013C

isco

2012C

isco

2011C

isco

2010C

isco

2009C

isco

2008M

cAfe

e

2007M

cAfe

e

2006Ep

iance

2005Ep

iance

2004IB

M

2003IB

M

2002IB

M

2001IB

M

2000IB

M

1999IB

M

1998IB

M

1997IB

M

1996Laitram

Co

rp

1995Laitram

Co

rp

1994Tu

lane

MB

A

1993Tu

lane

MB

A

1992In

dian

Railw

ays

1991In

dian

Railw

ays

1990In

dian

Railw

ays

1989In

dian

Railw

ays

1988In

dian

Railw

ays

1987SC

RA

Jamalp

ur

1986SC

RA

Jamalp

ur

1985SC

RA

Jamalp

ur

1984SC

RA

Jamalp

ur

1983IIT K

anp

ur

1982

1981

1980

1979

1978

1977

1976

1975

1974

1973

1972

1971

1970

1969

1968

1967

1966

1965

Engineering

Mgmt

ERP, e-Biz,

CRM

MB

A

Analytics

ManagementManagement

Consulting

US based

A timeline diagram

Business

Architecture

Business

Analytics

What is “analytics”?


Management

Information

Systems

Industrial

Engineering

Operations

Research

Finance

Planning &

Analysis

Statistics

Business

Intelligence

Strategy &

PlanningBiz Ops

(SCM, CRM,

Svc, etc.)

ANALYTICS

Computer

Science

… diverse

antecedentsBusiness

Strategy

Psychology

6 sigma,

Quality, Lean

What is “analytics” used for?


… create value by

driving decisions from

idea to execution

• Frame

• Inform

• Forecast

• Deliberate

• Recommend

What methods/techniques does “analytics” use?


ILLUMINATE DIAGNOSE FORECAST OPTIMIZE

• Charts & Tables in Reports & Dashboards

• Alerts, Visualizations, and Simulations

• Analysis, Interpretation & Recommendations

• Multi-Criteria Decision Models (MCDM), etc.

… use any method

that helps to drive

from idea to execution

MCDM, etc.

What data does “analytics” use?


TRANSACTIONS CONVERSATIONS OBSERVATIONS

• Structured

• ERP

• CRM

• HRMS

…

• Unstructured

• Documents

• Websites

• Social Media sites like

Facebook, Twitter,

LinkedIn, YouTube,

Blogs

…

• Sensors & Signals (M2M)

• Scans (Bar-codes, RFID,

proximity, etc.)

• SCADA (Control Systems)

• Temperature, pressure,

illumination, etc.

• Location (GPS, etc.)

• Smart-phone data feeds

• Network monitors

• Video feeds

…

… use any data as

needed, internal &

external

What domains is “analytics” used in?


Decision LayersStrategy, Capacity, Scheduling, Execution

… use wherever there

is a need for making

rational decisions

VerticalsTelecom, High-Tech, BF&S, Oil &

Gas, Healthcare, Government, etc.

FunctionsMarketing, Sales, e-Commerce, Fulfillment,

Supply Chain, Product Lifecycle, Strategy,

Finance, etc.

What systems/tools does “analytics” use?

Confidential CobotSystems.com

… use whatever

tools are needed

BusinessEPM, Collaborative

Decision Making,

Decision-Support,

Alerts & Feedback

ITData Warehouses, ETL (Extract, Transform, & Load)

Business Intelligence (Dashboards, Reports, Self-service)

Big Data

AnalyticsInnovation: from idea to value

Operations: continuous value realization

10

… a working definition


Analytics is used to

create value by driving

decisions from idea to

execution

… use any method

that helps to drive

from idea to execution

… use any data as

needed, internal &

external

… use wherever there

is a need for making

rational decisions

… use whatever

tools are needed

11

What is an "Analytics CoE"?Applies insights to data to produce analytics deliverables and drive business value


Variety of Analytical Insights

Process

System

Skills

Knowledge

Decision

SupportData

Variety of

Deliverablesfor various Business

Decision Contexts &

Processes

Drive

Business

Value

Variety

of Data

Sources

&Errors

New

sources

& errors

New

insights

New

Opportunities

& Issues

Agenda








What do leaders talk about when they talk of analytics?


Supply DemandC

urr

ent

Sta

te

Fu

ture

Sta

te

How should we manage

analytics investments to

gain efficiency?

Why do we need so many

reports? Why does it take

so long or cost so much?

How can we develop

analytics capabilities to

meet demands? My peer

has this, I want one too!

What can analytics do to

improve our business

results? Beat the

competition?

What do the leaders want to do with analytics?

The CEO asked for it The CIO has plans for it XYZ business leader wants it


FOR WHOM?

Finance

Planning &

Analysis

Strategy &

Planning

Biz Ops

(SCM, CRM,

Svc, etc.)

… for whom will you

create value?

… what decisions will

you drive?

… do the current

decision-owners think

you’re helping?

What decisions will you drive with analytics?


Analyze contract renewal

data to find opportunities

Feed the workload (renewal

opportunities) to Sales

Schedule follow-ups by the

Sales Call Center

Get the renewals, improve

renewal rate

Finance

Planning &

Analysis

Sales Ops

Call Center

Ops

Call Center

Ops

Answer the “what”, enlist the “who”


Opportunity Pipeline

Bookings Backlog Delivery

Deal P&LImpact

Analytics

Strategic Account Analytics

Delivery Optimization

Analytics

Revenue

Revenue Gap

Analytics

Bookings Trend

Analytics

Talent & Skills

Analytics

Resource Planning Analytics

Revenue & Cost

Alignment

… get clients by offering business value


Protect margin using deal analysis

Improve quality of business by targeting deals

• Analyze each deal for timing of costs (COGS & PCOGS), revenues (milestones), & risks (e.g., pay per SLA); assess FMV and VSOE impacts; calculate deal impact on the P&L

• Plug-in deals pipeline into the P&L projections, provide updates based on actual deals and run-rate

• Work with Geo leaders to drive decisions to improve the Sales P&L

• Approve or modify deals based on insight into P&L impact

• Deal targeting to actively manage and improve the Geo P&L

P&LDeals

Analysis

Improved P&L

Control

M1 M2 M3 M4 M5 M6

COGS $33 $33 $33 $33 $33 $33

PCOGS $21 $21 $21 $21 $21 $21

COSTS $54 $54 $54 $54 $54 $54

DELIVERY HLD LLD NRFU

REVENUE $50 $270 $310

Q1 Q2 Q3 Q4 Full Year

PROJECT MARGIN ($8) $34 $56 $43 $125

P&L IMPACT -1% 4% 7% 5% 4%

Q1 Q2

Deal P&LImpact

AnalyticsPROJECT

METHOD

DECISION

• Geo P&L Owner

• Geo Vertical LeadsOWNERS • Geo Finance Lead

• Geo Sales Ops Lead

… also track costs & value … keep clients by delivering business value

What is the role of the CoE that they envision?


… provide us the analysis,

we promise to use it to

create business value

… help us throughout the idea

to execution cycle to deliver

business value

… provide us decision advice

based on analytics

Agenda








What is the organizational context?


DRIVE

DECISIONS

Analytics

Providers

• Meetings (with customers, suppliers,

project teams, colleagues, etc.)

• Walking around, observing

• Mass media (Newspapers, TV, etc.)

• Internet, Books, Gut Feel, etc.

Suppliers

Substitutes

Relevant

True (data is correct, models are verified)

Actionable (provides the basis for specific action)

… apply the noise filter

What do analytics deliverables look like?


CommunicationDeliverable

Messages (email, SMS) to

Inform & Recommend

Documents (Reports,

Dashboards) to Inform,

Forecast, & Recommend

Decision Explorers to

Frame & Deliberate

Where will we find Analytics Providers?

Analytics providers are

found in many places, it

helps to cluster them

into these five domains

1. Business: managers

and their assistants

2. Analytics: analysts in

“analytics” teams

3. IT: in the BI, DW, or

Reporting teams

4. External BI Providers

5. External Analytics

Providers

Confidential CobotSystems.com 23

Who are the current providers of analytics?


Leader &

Department

Data Providers Noise Filter Analytics

Providers

Tom, VP, Sales IT, BizA-CoE,

MktDudes.com

Jack, Director Sales

Ops

Jack, Director Sales

Ops, MktDudes.com

Jim, VP, Engineering IT, PLM-Specialists.com PLM-Specialists.com Mary, PLM Manager

Harry, CFO BizA-CoE BizA-CoE BizA-CoE

Not

Used

Data

Provider

Trusted &

Relevant

Decision

Modelers

Decision

Advisors

Inconsistent

Decisions

Leadership support?


SUPPLY

Consistent

& Disciplined

Data

Oriented

Decision

Models

Learn &

IngrainD

EM

AN

D

Harry

0% to

+10%

More

than

10%

Reduce

Jim

• The size of the bubble

denotes $ funding

• The color of the

bubble denotes growth

Tom

Agenda








Analytics Provider Network

The analytics function

cannot be the locus of

all the analytics assets

(and talent) for the

organization – it is,

instead, a capability that

enables the effective

use of analytics

… understand the

network, figure out your

partnering strategy

Confidential CobotSystems.com 27

Enable the Analytics Network for the organization


DRIVE

DECISIONSBICC

(IT)

ERP

(IT)

CRM

(IT)

BizA

CoE

Sales

Ops

PLM

Mgr

SAS

D2

IIM B

PLM.com

Dir

FP&A

Zoom out of the “centralized / distributed / hybrid” debate … it’s a network

Is analytics

everyone’s job?

Isn’t it?

http://www.amstat.org/

http://www.amstat.org/

Enable the Analytics Network for the organization


Shared Data in an Agile Data Platform, Rapidly Extensible

Relevance & Trust built into Deliverables via a Rules-base

Decision Models built and provided where Relevant

Learn & Ingrain, rapidly adapt/evolve all layers

Data, in context

Relevance

Data-driven Decisions

Adapt & evolve

Provide the analytics platform to enable everyone in the network

… make “analytics” work for your organization

Where is the bottleneck? Demand or Supply?


Actual demand = Low

Potential demand = High

Actual supply = High

Potential supply = Low

Demand is a function of leadership

& imagination: stimulate the

imagination, deliver successes,

make analytics accessible with a

clear business value proposition

… transform the organization

culture to use analytics

... align supply to demand

Not

Used

Data

Provider

Trusted &

Relevant

Decision

Modelers

Decision

Advisors

Harry

Align supply to demand, and also stimulate demand


SUPPLY: Processes, Skills & Systems

Jim

Tom

Jim

Tom

Harry

Shared Data, Agile Data Platform

Relevance & Trust

Decision Models

Learn & Ingrain

… provide the

analytics platform

… stimulate demand:

in-house successes,

industry examples,

conferences, peer-to-

peer networks,

decision advisors

Analytics Network dynamics

Nodes Evolve: Mature, Add, Re-organize, Disappear

Deliverables Evolve: Innovation Operation

Data Evolves: Federate the data to keep pace

Platforms Evolve: Avoid obsolescence, embrace agility

Talent Flows: Enable people with career paths

Funds Flow: Manage funding for nodes & the platform




INNOVATE

CULTIVATE

EXECUTE

Work with data providers to improve data

Drive value realization through analytics usage

Ensure ongoing validity of the analytics

Provide analytics deliverables per SLA

Develop new insights & concepts

Drive from Idea to Execution

Assess value & need

Explore and Fail-fast

Continual creation of Analytics Deliverables

Continual assessment of Data & Model Validity

Continual governance of Policies

Continual monitoring of Usage

INNOVATION: TALENT-DRIVEN

OPERATIONS: PROCESS-DRIVEN



Junk 1-time Useful

1 100% 0% 100% 80% 16% 4%

2 100% 4% 96% 77% 15% 4%

3 100% 8% 92% 74% 15% 4%

4 100% 12% 88% 71% 14% 4%

5 100% 15% 85% 68% 14% 3%

6 100% 18% 82% 65% 13% 3%

7 100% 22% 78% 63% 13% 3%

8 100% 25% 75% 60% 12% 3%

9 100% 28% 72% 58% 12% 3%

10 100% 31% 69% 55% 11% 3%

11 100% 34% 66% 53% 11% 3%

12 100% 36% 64% 51% 10% 3%

13 100% 39% 61% 49% 10% 2%

14 100% 41% 59% 47% 9% 2%

15 100% 44% 56% 45% 9% 2%

16 100% 46% 54% 43% 9% 2%

17 100% 48% 52% 42% 8% 2%

18 100% 50% 50% 40% 8% 2%

Repeat

Delivery

Total

Capacity

Capacity

for New

New Delivery

The suppliers’ dilemma: the more useful analytics

you produce, the more you get mired in repeat delivery

… which is valuable repeat business but boring workJunkAlways

Useful

1-time

Useful

Repeat

Operations

Repeat

Operations

over time, the operations workloads pile up

… this is a mark of success: your stuff is useful!

… manage it with automation

TIM

E

100% Innovation

Agenda








Analytics roles and functions


Agile & Production Analytics: Innovation & Operation


Workloads: what do your "Analytics" people do?


Strategic Scorecards, Metrics Portfolios, & Scenarios

Reports & Commentaries; including Forecasts

Consultative Analysis Projects (Stats, Optimization, Simulation, etc.)

Enterprise Modeling, Business Modeling, Industry/Market Analysis

Data Stewardship (Assess, Cleanse, Enrich)

Business Case Modeling, Business Value Analysis, Valuation (Financial)

Dashboards

Governance Meetings & Follow-ups

Target, Actual, Outlook, and Revision

Alerts & Follow-ups

Communication & Driving Adoption of Data-driven Management

o Frame the Decision Need

o Generate Ideas

o Support Decision Making

o Monitor Execution

o Drive Results

Regression & Correlation

Propensity Models (Logistic Regressions, Multinomial Models)

Statistical Hypothesis Testing

Forecasting

Clustering and Segmentation

Optimization (LP, NLP, etc.)

Sampling & Surveys

Benchmarking

Pricing Optimization & Revenue Management

Design of Experiments

Financial & Risk Management

Search & Search Optimization

Queuing, Simulation

Statistical Process Control (R Charts, P Charts, etc.)

Planning, Scheduling, & Dispatch; Campaign Tracking

Decision Analysis, Multi-Criteria Decision Making, Game Theory

Project & Program Management, PERT, CPM

Quality Management (TQM, 6Sigma, ZeroDefect, etc.)

Business Architecture & Systems Modeling

Controls & Compliance (SOX, Basel, ISACA, etc.)

Organizational Behavior, Organization Theory, and Strategic Mgmt

Techniques: what methods & techniques do you use?


o Frame the Decision Need

o Generate Ideas

o Support Decision Making

o Monitor Execution

o Drive Results

Systems: what tools do you use? … the BI view

Data Loading & Processing Batch Stream

Data Sourcing Transactions Conversations Observations

Data Marts & Warehouses Closed Access (only via IT) Open Access

Offline Analytics Tools Reporting Dashboards Self Service, Slice/Dice, Drill-down

Data Quality In-line Analytics Surveying & Feedback


Systems: what tools do you use? … the Analytics gap


Innovation Frame decision needs in an enterprise model Manage idea-to-execution pipeline Visualize, Explore, Cluster, Segment, Forecast,

Optimize, Simulate, Machine Learning, etc. Decision support (AHP, MCDM, Bayes, etc.)

Operations Execution (Trusted, Actionable, Responsive) Learn & Ingrain

Data Provisioning Agility, Reliability & Serviceability Traceability & Fidelity

Data Assessment & Scrubbing Manage Data Problems Manage Data Dictionary

Master Data & Hierarchies Enterprise Ontology Model Shared Ontology & Enrichments Privacy, Security and Availability

41

Capability: does the analytics team leverage the spectrum of analytics techniques

1. Get/provide data, measure and improve data quality2. Observe, understand and explain business issues – create analyses,

generate insights, monitor execution, guide value realization

3. Forecast trends, predict outcomes and behaviors (assess possibilities)

4. Formulate, simulate, and assess scenarios, develop ideas and courses of action (develop ideas)

5. Recommend courses of action (select the best ideas: feasible and optimal or Pareto optimal)


Statistics

OR

Decision Advice

Culture: how readily do business people use the analytics?

1. Unaware of how analytics can be used

2. Aware of analytics but do not use it for various reasons

3. Use analytics for specific decisions and actions

4. Support the creation and usage of analytics, metrics and decision rules in an organizational model, can assess the value of analytics

5. Leader in using analytics, recognize applicability, & demand the best analytics; innovate, learn, & ingrain analytics in business processes


Trusted

Actionable

Responsive

Learn

Ingrain

Technology: do the analytics teams get the systems, tools and infrastructure support they need?

1. Unable to provide analytics systems (data, tools & infrastructure)2. Partially provides the analytics systems and data required and

refreshes the data periodically for off-line analyses

3. Fully provides analytics systems and refreshes the data periodically for off-line analyses

4. Enables analytics to be executed in-line, i.e., used seamlessly within business processes

5. Delivers enterprise-wide in-line analytics, so the organization can use analytics at scale and without delay


Shared data

Trusted, Actionable,

& Responsive

Fast decisions

Closed loop

Agenda








How will we evolve the CoE?




TARGET

STATE

CURRENT

STATE



How will we evolve the CoE? Set expectations, drive the evolution, establish the value

Establish the value

Experimentation (A/B Compare)

The model is used in an “experimental” setting, against the current “gold standard” method

Compare the results of the model to the “gold standard”

Before/After Compare

The model is deployed globally

Compare the results before and after the model is deployed (problem remains that there can be other explanatory factors)


Move fast & patiently

What can be done fast

Bring right expertise (skills) into the analytics network

Set up the analytics platform

What takes time

Change the decision-making culture of the organization


Produce relevant, trusted & actionable analytics

Help people to use the analytics deliverables

Track the value created

What must be done right

Evolve the analytics process capabilities

Failure is part of innovation

… manage it by ensuring it occurs in the context of the analytics platform and processes (back those innovation efforts that have an idea-to-execution perspective, remember and learn)


Concept to Value Realization

Process Value Management

Capability Value Management

Organizational Value Management

INNOVATION: TALENT-DRIVEN

OPERATIONS: PROCESS-DRIVEN

EVOLVE

Role

CoverageObsolescence is part of operations

… manage it by ensuring it occurs in the context of long term relevance (don’t lose memory)

Today 1 year 2 years 3 years 4 years

Inconsistent

Decisions

Change the decision-making culture of the organizationConsider leadership, trajectory, overshoot and undershoot


SUPPLY

Consistent

& Disciplined

Data

Oriented

Decision

Models

Learn &

IngrainD

EM

AN

D

Jim

Harry

Tom

Jim Harry

Tom

This presentation is based on the material

available in the book:

Business Analytics: A Practitioner's Guide

By Rahul Saxena & Anand Srinivasan

Springer International Series in Operations

Research & Management Science





Building an Analytics CoE (Center of Excellence)

Data & Analytics