Shell, one of the biggest petroleum companies in the world with over £300 billion in revenue in 2017, processes millions of transactions. Three out of four UK residents live within a 15-minute drive of a Shell station and many of them will have made a purchase in the last year. Formerly, these customers were seen as aggregated cohorts of particular demographics in different locations. Last year, though, Wunderman sought to change that.
"We looked at how we can predict the next purchase data at an individual level."
“How we wanted to use data was to shift to more of a customer-centric model and focus on the customers as individuals, rather than looking at them as groups of people with some shared commonalities,” Chris Parker, Wunderman’s analytics director told DataIQ.
Wunderman developed different levels of predictive modelling and gained a better understanding through statistics and distribution. Going from the macro to the micro view meant not just looking at the frequency in purchase patterns, but also whether that frequency was increasing, decreasing or remaining stable. This allowed them to start predicting when the next purchase would be made.
“We’ve tried to look at how can we predict the next purchase date at an individual level and deliver one-to-one offers when a customer is likely to purchase,” said Parker.
Weather data is being used to understand the context of transactions.
Parker and his team also started to use 13 weeks of customer activity as the measure of retention. This allowed them to identify betterchanges in behaviour. Parker also said that as well as looking at share of wallet as a metric, they looked at geo-data to understand the impact of competitor location on share of wallet. In addition, weather data is another type of data that is being used to understand the context of transactions. “We’re hoping to enrich and enhance our models there.”
"We made an analytics-pure department."
When Parker joined Wunderman two years ago, his remit was to manage all the analytics for the agency, but at that time the analytics department worked in silos. “The first change we made was to make an analytics-pure department so that there was a sense of identity within analytics. But the big change for Shell is that we started working as a hub, so thinking beyond analytics and think about data as a whole.”
To do this, he implemented an agile way of working, using sprint processes, and created mini working hubs for the bigger projects. Within those hubs, every person regardless of their skillset or the department would have a clear understanding of the problem and how they could use their skills to help solve it.
“That created a seamless hand-off between teams, but also made sure that when we were speaking to the client, we had a coherent story.” This meant they didn’t have any gaps between being briefed and doing analytics, so they answered the business question in the way that the business expected them to, and the operations team could deliver meet the analytical brief.
Parker emphasised that, fundamentally, Wunderman didn’t restructure the teams that worked with Shell, but changed their ways of working so that existing teams could work better together. This better way of working ultimately fed into improved predictions, better offers and happier customers.