Know who your customers are - part 4
In follow-ups to recent posts some people raised a point about using basket analysis in “reverse” to segment customers. The legend that young men buy beer and nappies (diapers) has been around a long time in the BI community (perhaps because having young families restricts traditional access to bars.) But can this type of knowledge really be used to segment customers?
Possibly not! In a selling organisation the bottom line objectives are to maximise profit and to increase market share to the detriment of your competitors. Basket analysis can help here by allowing creative sales promotions to be developed and perhaps optimising store layouts to extract the most cash from a customer. It is, however, far more important to know that 8% of customers buy products X and Y at the same shopping trip than to know that they are likely to be male, mid-to-late twenties. The key here is the word “likely”, we are using demographic data collected on a small sample of shopping baskets and extrapolating it back to the whole set of customers – not necessarily a statistically sound approach, nor one worth the cost of collecting the raw data. The other significant failing with this single basket approach to segmenting is that is assumes that people always buy the same things each visit; suppose a 25 year-old male buys beer once every two weeks and nappies on the alternate weeks, could we identify him as “male with a young family”?