FMCG giant Unilever is constantly looking for new ways to engage consumers and provide new tools and services that enhance the retail experience. As part of its drive to encourage healthier eating it approached Tessella to develop a sophisticated engine that would recommend better options to users based on their existing shopping history. The result is SmartSwaps – a deceptively simple application that belies the complex algorithms within.
Tessella is perfectly placed to provide support on this type of project because it can bridge the gap between hard core mathematical modelling and industry knowledge.
Louise du Plessis
Nutrition Scientist at Unilever
Background and Challenge
Unilever is a world-leading supplier of home care, personal care and food products. 150 million times a day, someone somewhere chooses a Unilever product. In order to maintain its market position, the company continually invests in research to optimise its products. It is also committed to helping consumers make better choices when buying products.
“Initially, we were looking for ways to encourage people to choose our healthier spreads such as Flora instead of butter,” explains Louise Du Plessis, Nutrition Scientist at Unilever. “This led to our team doing a lot of research on behaviour change principles around online shopping habits. We wanted to find a way to connect directly with consumers and influence these shopping habits.”
Tessella has long been a partner with Unilever, providing technical skills and software development expertise to address a range of issues across the company. This made it a natural fit for the SmartSwaps project. Together, Tessella and Unilever set about designing an algorithm that could take large amounts of messy data and turn that into personalised recommendations for consumers. Unilever worked with Tesco in order to access its vast database of customer shopping history. This would provide the fundamental information that would underpin the application.
“Tessella is perfectly placed to provide support on this type of project because it can bridge the gap between hard core mathematical modelling and industry knowledge,” adds Louise du Plessis. “From the outset the team from Tessella worked extremely hard to create a tool that could extract sensible and usable information from a large database that contained inconsistencies and missing information, and convert it into simple consumer recommendations based on levels of calories, fat, sugar and salt.”
Solution and Benefits
The resulting SmartSwaps tool is incredibly easy and intuitive to use. Consumers simply enter their Tesco Club Card details on the website and the algorithm generates alternative shopping items in which ice cream, for example, could be substituted with frozen yogurt or a lighter version ice cream. On their next online shopping trip, users are able to view better alternatives to the ones they have chosen in the past. One upside for Unilever is that because it has a wide range of healthier options, its products will be recommended more frequently.
“The algorithm looks at the customers’ previous purchases and what other people buy in those categories. It takes into account levels of calories, fat, salt and sugar and also specific brands. By crunching all these variables it can recommend relevant better alternatives,” continues Du Plessis. “It gives us a number of benefits – firstly, SmartSwaps allows us to connect directly with the consumer and begins a dialogue whereas our usual relationship is mediated by the big retail groups. Secondly, it helps consumers eat more healthily without compromising on taste. And finally, it helps showcase our own range of healthy brands and drive sales.”
SmartSwaps has yet to go fully live but it has undergone a usability pilot with 200 users which is already giving valuable insight into how it influences purchasing decisions. Of these users, an enormous 87 per cent changed a habitual product for a better alternative. The average number of swaps made per person was five demonstrating the extent of the application’s influence. What’s more the recommended products worked out slightly less expensive than usual purchases.
“The impressive statistics show the relevance of the recommended products and how well Tessella’s algorithm works,” says Du Plessis. “If we were highlighting products that didn’t closely match their needs, it simply wouldn’t have been so successful.”
With the concept now proven to be viable and relevant to the modern consumer, Unilever’s hope is to roll it out to a wider audience. It is hoping that SmartSwaps will play an increasingly important role in helping British shoppers become more aware of better options available in the market place.
“The SmartSwaps tool is versatile and could be used with any retailer. We can plug it into their customer databases and let it work its magic,” concludes Du Plessis. “It’s thanks to the work of the Tessella team that we have such a robust, intelligent and flexible application that really understands the needs of each consumer.”