Lessons from Brick-and-Mortar Retail

I recently read Emmet Cox's book on Retail Analytics, which focuses almost exclusively on the brick-and-mortar channel. To my way of thinking, that focus is not a drawback: too many times, it is proclaimed that the brick-and-mortar channel is dead (or dying) and the online channel has nothing to learn from it. This is generally the proclamation of people who don't understand the very things they propose should be ignored, and unless we know better, such people can steer us in the wrong direction.

As in many things, there's a great deal to be learned from methods that have been successful in other channels, and a great deal that carries over very well from one channel to another. Whether online or in-store, customers have preferred retailers with whom they shop often. They tend to shop with a certain frequency, buy certain products together, and change their preferences of both items and retailer with surprising predictability (if you know what to monitor).

Online merchants have the ability to perform all of the same kinds of analyses as do traditional ones. Granted, they have a much broader customer base (they do not draw from a limited geographic area), which carries with it certain advantages and disadvantages, but they still collect the same kinds of data (header, detail, and tender) about every transaction.

There is additional data that an online merchant can collect. Unlike a physical store, monitoring and tracking every movement of a customer from the time they enter to the time they leave is neither impractical nor intrusive in the online channel - and unless you turn off a lot of server logging functions, it is in fact impossible not to collect this data.   Such information can grant additional insight - but it is "additional" and not "replacement": the precise number of people who look at an item but decide not to purchase could be helpful (if it drives a decision), but does not supersede more meaningful metrics. 

The problem isn't that retailers in either channel don't have data - the problem is that they don't know what to do with it. There's a great deal of very meaningful information that can be gleaned from the bits of data that are collected that is simply ignored. Or when it is collected, it is viewed with a "Hmmm ... that's interesting" perspective and doesn't guide business decisions.

If a merchant isn't using the simple basket data that is collected as a matter of course with every sales transaction, he's missing a lot. And for online merchants, the additional click-trail data that is gathered as a user browses a site and adds or removes items from their basket is often a distraction from the more meaningful transaction data that is being dismissed as irrelevant simply because the customer is using a different channel.

As such, the analytic techniques use in traditional retail are still valid and well worth considering - and getting them right, even before considering any of the additional data available by virtue of the online channel.  Ideally, the multi-channel merchant aggregates all data, separating it by channel, and channel-specific behaviors become evident. There will be differences, but they will not be altogether incompatible and unrelated.