How store-based retailers are moving from general propositions to applying rules to individual customers
Business Analytics is a science so hidden that it is rarely that one finds it splashed in a full-featured article in a widely-published newspapers. In this New York Times article, staff writer, Charles Duhigg, details how Tesco uses high-end data analytics to successfully, in its words, learn their shoppers’ secrets and hence successfully predict their shopping behavior. Let us take the opportunity of using this story to highlight some of the several issues impacting the retail industry, retail analytics and the implications of increasing sophistication and precision of data analytics.
In this article, we will look at why, in today’s world, the traditional store-based retailers (SBRs) need to employ high-end analytics to understand their target customers.
Let us begin with why do retailers need to define a target customer. Defining a target customer is intrinsic to a retailer’s strategy. A strategy is a plan of action into getting where one wants to. Hence, two things define a strategy. First, a destination, because how will you make a plan to get somewhere if you don’t know where that somewhere is. Second, a series of choices one makes. If you decide to go east, you do not go west. You make that choice. You can’t go both east and west, which, unfortunately, many businesses do try to do.
For a retailer, the destination is the customer – or rather specific shopping trips of that customer. The retailer builds the appropriate format for that shopping trip. For example, the same retailer might have a big box to satisfy the big beginning-of-the-month shopping trip, and also a neighbourhood convenience store to satisfy the need to urgently buy a few essentials the customer has run out of, or grab a bite. Here, the target customer may be the same; the difference is only in the shopping need of the same customer.
Hence, it is essential that the retailer clearly know who the target customer is and what shopping trips does he make. But supposing Walmart, which has 100 million footfalls every week in its US stores, asks this question – who are my target customers? Is it possible to interview and profile each of its customers? Clearly not.
That is why, traditionally, SBRs have profiled customers by the aggregate. That is they group the population into buckets and profile them on the basis of which bucket they fall into. So if someone asks them who this customer is, they answer – he’s the customer who falls into this bucket.
The attributes that make these buckets are demographic attributes (age, education level, income, etc.). By building customer profiles at the cross-sections of these various attributes – like white males between 30-35 years, living in so-and-so geography, employed in so-and-so profession and earning so-and-so much – the retailer have aimed to guesstimate what a customer answering to this description might behave like. These attitudes and preferences have been understood by detailed on-the-ground market research and academic research, as the research highlighted in the article. Big retailers have indeed spent millions and billions of dollars in understanding the typical profiles of their best customers. This broad understanding has helped them to tweak their proposition at the general level.
However, unlike web-based retailers (WBRs), SBRs have been traditionally unable to link transactions to specific customers. That is, having no individual history of the customers, but only the history of all the anonymous customers, they have been unable to classify a specific customer into a specific bucket. This has led to an increasing gap in the ability of the SBRs to generalize their broad propositions to specific rules.
For example, let us say that both the SBR and WBR find out that customers who buy product A are likely to buy product B on the next visit 85% of the time. Simple enough. The SBR can generate a coupon for product B at the cash-till once it finds product A in the shopping-basket. The WBR might do the same, and moreover keep sending reminders for product B to the customer.
But, suppose the study further reveals that customer who buy product A on a visit, and product B and C on the second, buy product D, E and F on the third visit 67% of the time. Here, the split between them is revealed. The SBR has no memory of the customer beyond the present visit. If he buys B and C, the SBR has no way of knowing that the same customer bought A in the last visit, and apply this rule. However, since most a WBR traces the individual transaction history of most of its customers, it can apply this rule – and score over the SBR.
As web-based channel in retail steals more and more market share from the store-channel, it is imperative that SBRs think of disruptive ways to link profiles to individual customers. In the story in the article, Target has been able to surmount this problem by its radical Guest ID number, that link a particular customer to all his shopping transactions over time. “If you use a credit card or a coupon, or fill out a survey, or mail in a refund, or call the customer help line, or open an e-mail we’ve sent you or visit our Web site, we’ll record it and link it to your Guest ID.”
Hence, not only are they crunching shoppers’ data into predictions on customer life stage and behavior, they are able to classify individual shoppers into these buckets. The logic is the same as that of fraud analytics where you identify suspicious patterns and then classify whether each transaction, in real-time, falls into that pattern or not.
Hence, while the article focuses on the power of analytics that can build such precise rules on probable human behavior, this capability has existed for some years with retailers, like in other businesses like finance where analytics is applied extensively. The innovation, for me, is in how Target is able to link customers to transactions over time. The opportunities this presents to Target is indeed immense (and part of the reason why they are so cagey to share more on this is to guard these treasure-chests of secrets from their competitors) because big-data analytics is way, way more powerful than non-insiders can imagine. All retailers need to unleash its juggernaut is one more attribute in the POS column, the elusive Customer ID!
Interested in a career in Retail Analytics? Check out Jigsaw Academy’s Retail Analytics Course.