Effective Retail Analytics

There is data and then there is knowing how to use it. The findings on data usage by retail consultants eCommera three years ago don’t seem too removed from the status quo: then, only 23 per cent of UK retailers felt comfortable making the right business decisions quickly based on the data they collected, and nearly half of those canvassed considered their business intelligence tools to be inadequate.

The challenges posed by analytics

Consider the sheer volumes of information. Millions of consumers interact with the biggest retailers through multiple points of contact: brick and mortar, websites, apps, customer service centres. Add to that all the unstructured data (information not stored in company databases) on mobile devices and the plethora of text, images, and videos perused by customers on social media. Collating all of this to deliver a coherent and seamless customer experience is no mean feat.

One aspect of the challenge is understanding cross-channel behaviour, as retailers tend to use different data systems for online and in-store sales. Stores often try to bring things under one umbrella by incentivising customers to add their email addresses to mailing lists, by offering loyalty card schemes, or by using emailed receipts for in-store purchases.

Even so, general consumer behaviour is broad and hard to categorise. Addressing this unwieldy problem is part of the practice of good customer intelligence, gathering the right information reliably in order to deliver data-driven insights into past and predicted buying behaviour.

Predictive models

But how to contextualise all that information? By segmenting and then targeting specialised groups of customers, predictive models can draw actionable conclusions about behaviour. For example, by identifying shoppers’ average order values and drop-offs in their likely patronage in the short term, retailers get a view on high-value, loyal shoppers who appear to be disengaging from the brand.

Propensity and uplift models

Analyse the clicks: if the user appears to be spending a lot of time reading reviews, examining features and trying different colours in the smartphone section, that is reason to believe he or she is primed for a purchase. By sending this user an exclusive limited-time offer (instead of reaching out to everyone in the database), chances of success are sharply increased, and retailers avoid diluting the rest of their mailing list’s attention with a message that doesn’t resonate.

The specific insights afforded by modern retail analytics allow for differentiation between customers to a much greater degree, and when you know who your customer is, you know how to speak to them. Propensity and uplift models even save millions by forgoing discounts for customers who are likely to buy anyway.