Utilising advances in machine learning to attribute segments to customers for tailored brand communication
Database Attribution - Holiday Traveller survey segments
The challenge
Bonamy Finch created a two dimensional segmentation framework for TUI UK. The first dimension reflected consumers’ enduring relationship with the travel category, and the second dimension captured the needs that change from one holiday to the next. Attribution of these two segmentations to the customer database was essential to help implement a program of tailored communication and improved customer engagement.
The solution
Used the learning data set from the customer survey – where we reviewed 9 million database records, added database proxy questions to the survey, and carefully considered the key dimensions that drove the segmentation.
Multiple machine learning algorithms tested for database attribution accuracy (predicted v actual segments on customer sample). An ensemble of 5 algorithms utilised to provide a very strong accuracy overall.
The result
Successful production of accurate attribution algorithms in R (statistical computing language used by TUI), with supporting documentation and deployment support.
TUI can now easily feed in new customers or new trips into the respective algorithms and generate a segment code, thus helping us to achieve the original project goals. The project has been so successful we are now working with TUI to roll out the approach to all other key markets.
The feedback from very senior stakeholders has been overwhelmingly positive and there is generally a level of excitement around the potential of segmentation in the business.
Kate Sargent
GM Customer Insight & Analytics, Customer Experience, Insight & CRM, TUI UK
