The project won the CRM category at the recent I-COM Data Creativity Awards 2018, where David Lloyd, head of data and insight at Shell’s agency, Wunderman, explained that levels of loyalty across the fuel market had been dropping 6% year on year.
Consumers tend to see little to differentiate any of the main fuel providers, he said, and they might take a membership card or download a loyalty app, but only use it once or twice before forgetting about it.
But when people did engage with the offers they got through the Shell loyalty programme, they were more likely to remain active users, he reported, and were likely to spend significantly more money with Shell – four times as much as those who didn’t.
“When we went and spoke to them, we realised that relevance and personalisation was a key issue,” he said. (For more details, read WARC’s report: How Shell personalises its loyalty offer.)
Only one in ten were engaging with offers while more than half complained the offers were on products they would never buy.
“Our vision was to humanise the relationship and make every interaction with Shell a personal one,” Lloyd stated.
That involved drawing on hundreds of millions of data points that Shell already had: analysing behavioural data to determine the nature of the driver (eg commuter or weekend driver) and assessing frequency of visits to filling stations and how much fuel was put into vehicles.
Data was then used to determine where in the life cycle of the loyalty scheme each person was, from new joiners to those ready to churn, and determined how generous Shell needed to be – and on which particular products – in order to retain them.
The new system can dynamically adapt to purchase patterns, seeing when people are moving from weekly to fortnightly, for example, “and the whole cadence of frequency of communication is adapted accordingly”.
The project has resulted in an 18% increase in offer engagement, a 24% drop in churn rates, and $38.2 million in incremental revenue, with 549 million litres in incremental fuel sold.
Sourced from WARC