I need smarter analysis and reporting... and to profile, segment and really understand our customers...


1) Situation...

Month upon month we continue to collect trading information from our customers, but it is only accessible from our accounts system and mainframe.  Even then, what we can see is limited, not in a marketing format and doesn't link to the customer information we hold within our sales lead system - mainly because each system uses different reference numbers. 

In order to understand purchasing patterns and help us decide who to contact, when and why, we would like to be able to extract information from both systems and load it into a suitable analysis programme on a PC.

From here we wish to link the files and create a series of summary fields held against each customer. These need to include:

  • first data/sales source (from which campaign first purchased)

  • last data/sales source (most recent campaign purchased)

  • lifetime number of purchases

  • lifetime amount spent

  • lifetime campaigns in which purchases made

  • annual spend by product category and percentage against total customer spend

  • recency, frequency and value score bands

  • behaviour change flags

We would like to segment the database into groups of customers sharing similar characteristics and describe each group with a profile. Then use predictive modelling techniques to improve our campaign response and feed propensity scores, behaviour change flags and summary profile information back into an easy-to-use marketing database for improved targeting. And of course we would like to see how our database has changed over the years!


2) Solution approach...

3) Real benefits...

4) Overview of Data Mining and Statistical Analysis in Marketing