Having revolutionised the world of retail data once through dunnhumby and its work with Tesco, Edwina Dunn, now CEO of Starcount, may do so again with a new approach to segmentation combining postcode data with social likes.

In an article for The WARC Guide to making segmentation work, Dunn outlines her golden rules of data science and how to find the best data to fill the gaps that can be seen in every customer database.

The WARC Guide to making segmentation work looks at how segmentation is evolving, as brands utilise new data sources, and how marketers in markets across the globe are deploying machine learning to identify meaningful audiences. Subscribers can read the full report here.

When working with major grocers like Tesco and Kroger during her time at dunnhumby, she was able to derive “fantastic insights” from the repetitive nature of products and visit frequency, she reports, “but many of the applications were unique to that industry”.

Things have moved on since the 1990s, however, with a number of major new data sources. She identifies six that comply with her golden rules: Google and Facebook – “pretty well closed to independent agencies” – government, banks, telcos and the open APIs of social media.

The use of bank and telco data remains limited thanks in part to GDPR constraints, but government and, especially, social media data offer real potential.

“We have found that people constantly signal their beliefs and motivations by the communities they choose to be part of on social platforms in a way that can’t be seen or measured with pure transaction data,” Dunn observes.

She describes “a breakthrough in terms of new understanding and consumer classification” that uses big data to explore what people love based on the time they spend following “stars” on social media. The result: 350 different and distinct tribes.

Layering these on top of postcodes and social profiles of an anonymous panel of 3.6 million people enables an understanding of how these tribes are spread regionally and opens up new possibilities when it comes to joining the off-line world of geography and retail with the on-line world of e-commerce through people’s passions.

In this way, says Dunn, you can reduce “a complex omni channel world with multiple data sources and conflicting information to a simple code – unlocking the science of knowing customers, knowing catchment areas, or even finding lookalikes”.

For more on the creation of a single classification of consumer behaviour and intention that can be applied to all customers, all prospects and which spans the on-line and off-line world, read Edwina Dunn’s article in full: Mapping passions with postcodes to identify lookalike audiences.

The WARC Guide is a compilation of fresh new research and expert guidance with WARC’s editorial teams in New York, London, Singapore and Shanghai pulling in the best new thinking globally. It also showcases the best on WARC – case studies, best practice and data sourced from across the platform.

Sourced from WARC