Data Science, Analytics & Solutions
This multi-channel retailer had been working with a targeted customer approach, but this had resulted in over 60 different customer segments that was ineffective and watering down attempts to drive sales growth. They asked Beyond Analysis to develop a new more workable model that would support marketing and operations equally well.
What we did
We reviewed and audited the existing customer segments based on purchasing behaviour and product preferences. One of the first observations was the technical approach to developing the segmentation that we felt did not truly represent each individual customer effectively and had led to an over simplification of how customers were attributed a segment. As such we redesigned the solution using raw data and dramatically increased the number of features used in segmentation. Using machine learning, we ran over 50 different segmentations (varying customer features, number of segments and clustering technique) to produce the optimal segmentation.
Seven new distinct and actionable segments were developed with detailed buying behaviours, product preferences that could be deployed straight into marketing and allocation. Further analysis of these groups identified £29m of opportunity in retention and re-acquisition.