Oct 17, 2015
5/27/2018 111236462 Equations of Retail
1/12
< b> a n(b
0< (a)< ln(b
x(x-1)=x2-x
x=1
xn*yn=(xy)n
FV=CV(1+g)T
4+6=54
x4=16
2
6 =6
Q=
na2/1+a2/2+...+a2n
(xy)z=xyz
x4 = 16
Q= a2/1+a2/2+...+a2n
x4 = 16
The newequationsof retail
5/27/2018 111236462 Equations of Retail
2/12
2
The new equations of retail
A common complaint of the retail CEO confronted with thetsunami of online data is now what?
The new equations of retailMichael Ross Chief Scientist and Co-founder, eCommera
Retail is about thousands of small decisions
so the question should be: what new or differentdecisions should we make given this data?Three things make this difficult:
New activities Google, site sort orders,CRM, delivery to customers are all generallynew disciplines for retailers, and require newskills
New data each of these activities generates
its own torrent of data which needs to becollected and understood
New costs the cost per click of Google,the variable costs of picking/packing/shipping,cost of communicating with customers.New costs change the economics of retail.
New activities mean new decisions. And newcosts mean different decisions. These decisions
require rules. And the new data allows rules tobe codified what we (somewhat) grandly callthe new equations of retail.
This article gives five examples (out of the
hundreds of possible examples) of the sorts ofscenarios where new and different decisions needto be made and where its easy to get it wrong.
How much is it worth investing in the serviceexperience?
What profit per order maximises your overallprofit?
How much should you spend on a keywordon Google that hasnt generated any sales?
What combination of markdown and productviews maximises profit?
When should you target a customer who hasstopped purchasing?
This article sheds some light on these questions.
5/27/2018 111236462 Equations of Retail
3/12
3www.ecommera.com
In physical retail, operations is back of house,focused on getting product to store. In the onlineworld, operations is the last touch point with thecustomer, and is fundamental to success, but it isoften handled by exactly the same people, and inexactly the same way.
For most retailers, this is not the exciting part ofeCommerce. Theres no whizzy new technologyto play with and no social opportunity yetoperational excellence is critical for success.Imagine a typical situation: you get a 600 orderat 2.30pm from a first time customer who hasselected a different billing and shipping address,with next day delivery.
Do you:(i)Do everything possible to ship it same day(ii)Do a standard fraud check but insist that the
first order is delivered to the billing address(iii)Do a rigorous fraud check and probably miss
the delivery promise.
The likes of Amazon and Net-a-Porter can simplydecide to make service a focus and have amessianic belief that its the right thing to do. Forthe majority of retailers, their CEOs and boardsrequire convincing that this is the right place toinvest.
A key operational question therefore is howcan you make the trade-off of between cost vs.service?
The Maths
Intuitively it is obvious that better service drivescustomer retention, but how do you quantifythe value of service? Whats the equation that
tells you when to stop investing? The key is tounderstand the relationship between service andcustomer retention, and then to quantify the valueof retaining customers. A typical analysis is below:
OPERATIONS:Whats the value of service?
Figure 1: Delivery on promise
-3%
TODAY
TODAY
MorePro
fit
LessProfit
Better serviceWorse service
-2% -1% +1% +2% +3% +4%
Incrementalcustomerlifetime value
Incrementalprofit
INPUTS
Impact of delivery on promise on
customer satisfaction
Impact of customer satisfaction on repeat purchase rate
Impact of repeat purchase rate on customer
lifetime value (LTV)
Cost of improvement to delivery of promise
Optimisationopportunity
5/27/2018 111236462 Equations of Retail
4/12
4
The new equations of retail
Once the relationship between cost and servicehas been established new processes need tobe put in place. Of particular importanceare automated fraud processing, orderprioritisation/routing, order cut-off times,carrier pick up time, shift patterns, weekend
working (to avoid a Monday backlog), etc.
For example, Figleaves had a focus that all in-stock orders placed by 4pm should be shippedsame day and the warehouse managers dailyreport showed orders placed by 4pm notshipped.
It is vital to develop processes that make failingon promise unacceptable. In the early days of
Net-a-Porter, orders from VIP customers wereprinted on pink paper so they stood out in thewarehouse, with a mandate that all VIP ordershad to ship same day. Operational magic!
5/27/2018 111236462 Equations of Retail
5/12
5www.ecommera.com
In physical retail, marketing is either non-existent(high street rent obviates the need for marketing)or is a fixed annual spend. As one retailer told merecently, our marketing strategy is deciding howmuch we will spend on TV at the beginning of theyear. In addition, the variable costs of sale in store
are negligible typically a carrier bag and creditcard charge. In physical retail, all transactions(outside of sale and clearance) make money.
The profit model online is different. Its very easyto lose money on an online order. Firstly, mostmarketing is a variable cost whether per order, perclick, or per impression. Add to this variabilitya delivery charge (which could be greater, equalor less than the directly associated costs) and
promotion costs.
Tough-minded retailers often announce thatthey want to make money on every order theyship. This statement sounds sensible but is
(i) not what it seems and (ii) is definitely not theway to optimise profits.
Firstly, aspiring to make money on every orderis primarily a statement about the granularityof marketing allocation. Why? Look at the
marketing cost of an example order, the visitorclicked on Google and that visit cost 27 pence, sothe individual order looks very profitable.
However, you then note that this profitable orderis accompanied by 300 visits that didnt convert.So the order is profitable but the marketingactivity directly associated with the order isnt.
Secondly, ensuring that all orders are profitable is
clearly a different statement to ensuring that theoverall profit is maximised. Imagine a retailer witha 10 delivery charge this may guarantee thatevery order shipped makes money; unfortunately,a high delivery charge has a negative impact on
MARKETING:Whats the right profit per order?
Figure 2 Understanding profit per order
Gross tradingprofit
Gross profitper order
Marketing costper visit
Conversionrate
Promotioncost per order
Marketing costper order
Delivery revenue/cost per order
Average order
value
Gross margin
x
Number oforders
Trading profitper order
x
x
Online visitors
5/27/2018 111236462 Equations of Retail
6/12
6
The new equations of retail
conversion rate, and will depress the volume oforders. So, a core marketing question to ask iswhat profit per order maximises overall profit?
The Maths
Maximising overall profit requires the rightbalance of marketing spend, delivery charge andpromotion cost. To give a simple example, aretailer makes more money from 5,000 orders at5 profit per order, than 2,000 orders at 10 profitper order.
The critical equation to understand is whatdistribution of order profitability maximises totalprofit (and the optimal profit will typically come
with some loss making orders). For example,ASOS offer free delivery and free returns in theUK. It is unlikely they made any money on the 3
Figure 3:Profit per order
AVERAGE18.20
CURRENT DISTRIBUTIONOF ORDER PROFITABILITY
OVERALL PROFIT
40
Profit per order
Number of orders
TODAYMore
promotionalLesspromotional
5 10 15 20 25
Optimisationopportunity
Averageprofit
per order
ILLUSTRATIVE
TODAY
Moreprofit
Lessprofit
order I placed for an earring multipack, but theoverall business is extremely profitable.
Online retailers need to recognise that monitoringand managing the distribution of orderprofitability is a new discipline. They should
monitor competitors delivery charges, freedelivery thresholds and the availability of widelyavailable discount codes. Offering free/reduceddelivery by stealth (i.e., without marketing) isa great way to understand the latent elasticitywithout creating a call to action.
There are creative ways to manage orderprofitability: for example, Zappos treatspromotions, marketing and delivery as a single
pot so that free delivery becomes a promotionalactivity funded by marketing.
5/27/2018 111236462 Equations of Retail
7/12
www.ecommera.com
MERCHANDISING Lower price or more traffic?
In physical retail products either sell or dont sell.For those that dont sell, the action is blunt reduce their price. Whether buyer or merchandiser,the ownership of the performance is clear. Thiscombination of the physical cost of markingdown and the well understood price elasticity
means that most retailers typically have a maturediscount schedule. For example, most fashionretailers have a 30-50-70 percentage pointdiscount approach.
Online the product economics are different.Product views have a cost and you can forexample drive traffic from Google to specificproducts. For merchandising therefore, a keyquestion is Do you get a better return from 100
more visits or a 10 percent discount?
The Maths
Online retailers need to understand the optimalprice/views mix that maximises product
profitability (do you give money to Google orcustomers?). The maths requires understandingthe relative elasticity of price versus traffic. Atypical analysis is below.
Once this relationship is understood the next
challenge is one of cultural change. Merchandisingteams must learn to make discounting decisionsusing analytics data. Buyers incentivised purelyon achieved product margin have little interest inworrying about end-to-end product economics.Similarly, marketing teams incentivised on a costper order dont worry about gross margin.
I asked the CEO of a US department store how hegot his buyers and merchandisers to engage in this
sort of analytically driven markdown approach.He replied: If buyers/merchandisers come tome to approve a markdown and dont have anyanalytics data, its a short meeting.
Figure 4: Price/views to optimise product profit ILLUSTRATIVE
PRODUCT VIEWS
Totalproductcontribution,
000
(Margindirectmarketing)
Optimal strategy to reduceprice and views
20% off Full retail price
200 300 400 600 700 800100
10
20
30
40
50
60
Optimisationopportunity
TODAY
7
5/27/2018 111236462 Equations of Retail
8/12
8
The new equations of retail
Most retailers are spending between 10-30 percent of their paid search budget on keywords thathave never generated a sale. Often retailers simplylook at their blended overall Google performanceand accept (either implicitly or explicitly) thatthe efficient spend can subsidise the inefficient
spend. Its a dangerous game equivalent toaccepting that profitable stores can subsidiseunprofitable stores and is simply a waste ofmoney. Optimising Google requires optimising ona keyword by keyword basis.
For example, imagine you start bidding on anew keyword Canon Ixus 230hs whichhasnt generated any sales (on any attribution
window) after 5 visits, 10 visits, 100 visits,1000 visits when do you stop? Stop too soonand you will miss sales, stop too late and onlyGoogle wins.
A critical marketing question then is How do
you determine the optimal stopping point for anew keyword?
The Maths
The maths of paid search is horribly complex.Google has created a deceptively simple model you pay per click. How hard can it be? In realityit is an enormous mathematical puzzle.
MARKETING:How much to spend on
a keyword with no sales?
10 15 20 25 30 35 405
0
10
20
30
40
50
Figure 5:Optimal stopping point for example keyword
Expectedprofit
Optimal stopping pointLifetime value
Optimal stopping point1st order profit
INPUTS:
Example for CanonIxus 230hs
Cost per click W pence
Product grossmargin X
Expected customerlifetime value Y
Target customeracquistion cost for thiskeyword Z
Informed priordistributionassumption
-20
-10 Number of visitswith no sale
5/27/2018 111236462 Equations of Retail
9/12
9www.ecommera.com
The maths requires understanding after howmany clicks to stop spending. The analysisabove shows the relationship between theexpected profit versus the number of visitswithout a sale.
There are a wide range of new activitiesand processes that are required to managethousands of long tail keywords. If a keywordisnt working, its easy to simply switch it off,however that may not be the best action to take.The new processes include:
Checking price/availability (dont blameGoogle if youre out of stock or overpriced)
Checking match type its easy for broad
match keywords to pollute performance Checking landing pages/copy (looking at
bounce, engagement, quality score its easyfor landing pages to become out of date)
More targeted bidding by geography or timeof day for example, we found at Figleavesthat bidding on the keyword lingerie wasless economic late at night.
Google are the virtual landlords of the high
street. And as on the high street, no one likes toover pay on rent!
5/27/2018 111236462 Equations of Retail
10/12
10
The new equations of retail
In physical retail customers are typicallyanonymous.
By contrast the online world affords untoldinformation about your customers. It is clearly aretailers prerogative to decide (i) whether to look
at this data, and (ii) what action to take. Doingnothing is a strategy and there are many verysuccessful retailers today who succeed withouttapping into any deep customer insight.For those that do look, the question is what toactually do. Take a customer (John) who hasspent 1200 over the last 3 years, making 12purchases. Johns last purchase was 6 months
ago. Do you: (i) do nothing, (ii) send an emailpromotion, (iii) retarget John with a banneradvert, or (iv) call John?
Common wisdom is that a customer who hasntpurchased for 6 months is lapsing and after
12 months is lapsed. These are clearly broadgeneralisations. Every category has its ownpurchasing rhythm.
So a key question on customer next best action toask yourself is: When is the optimal time to targeta customer who hasnt purchased?
CUSTOMER NEXT BEST ACTION:When is a customer lapsed?
Figure 6: Optimal intervention point ILLUSTRATIVE
Incrementa
llifetimevalue
Weekssince purchase
10% promotion20% promotion
Optimal intervention point
1 2 3 4 5 6 8 9 10
Optimisationopportunity
TODAY
5/27/2018 111236462 Equations of Retail
11/12
11www.ecommera.com
The Maths
It is critical to analyse and understand the impacton customer lifetime value of different promotionsat different times post purchase.
Offering a discount on the next order immediatelymay be too soon, after 12 months may be toolate. The optimum time will be different for eachbusiness.
For example the Economist designed a renewalat birth campaign after recognising that the besttime to renew a subscription was just after it hadbeen taken out.
Moving from broadcast to targetedcommunications requires building a customer-centric marketing capability with a culture ofdisciplined testing. Many organisations claimto be customer-centric with a test and learnculture. Few act like this in practice truecustomer centricity is difficult and most retailerslike the concept of test and learn, but dont like tomake mistakes.
***
As Richard Feynmann who won the NobelPrize for Physics often remarked, If anyonetells me they understand quantum mechanics,they dont. I feel the same is true of most
aspects of eCommerce.
eCommerce is still in its infancy and many ofthe new equations are simply unsolved. Thebest ecommerce practitioners know what theydont know.
5/27/2018 111236462 Equations of Retail
12/12
eCommera Limited1st floor,
Wells Point, 79 Wells Street,
London, W1T 3QN
+44 (0)20 7291 5800www.ecommera.com
Take control of your eCommerce business with our flexible,multi-channel platform:
Delivered on-demand for speed to market and in-life flexibility Extensible with pre-integrated best-in-class partner products Continuously innovated and upgraded.
Built, managed and evolved rapidly and cost effectively by oureCommerce technology experts.
eCommera TradingEdge solutions optimise multi-channel growth and profit potential for our clients.
Transform your wealth of eCommerce data into actionable insightand optimised business performance with our data navigator:
Operational dashboard makes sense of vast amounts of data Rapid means of diagnosing the drivers of retail performance Recommended trading actions, with a profitability-impact score.
Backed by strategic advice, planning and operational consultingfrom our retail experts.
Transact your brand online and internationally through our fullymanaged eCommerce service:
Strategic framework for planning and deployment Integrated end-to-end eCommerce operational infrastructure Dedicated team of retail practitioners to manage and optimise all
aspects of your online operation, from marketing to fulfilment.
Our multi-skilled team use best practice methodologies and leadingedge technology to deliver optimised customer experience andonline growth.