A Prospect Pool was needed that would contain a large number of records along with key selectors and discriminators.
Working with a media buyer, 30 million records from three of the UK's main data providers were sourced.
The data suppliers were not chosen based on coverage but different types of data.
One provided comprehensive lifestyle and demographic information, the second contained mail order purchasing information and the third, financial and wealth data.
All data was consolidated into one database, the Prospect Pool.
By matching files of the Agency’s clients against the Prospect Pool all current donor’s could be excluded but more importantly the data was Profiled, key selectors identified and models created for scoring purposes.
The Scored data was then output and used for prospect mailings.
In some cases ‘control cells’ were output that contained records with low scores for testing purposes to ensure the accuracy of the models being applied.
Details of previous outputs (Order History) were factored into the models and ensured that multiple outputs of the same record were minimised.
Results of each mailing were analysed and used to further enhance the models and scoring. |