Development of a scoring engine for optimised campaign management

Starting point

With a multi-million customer base, deciding on what will appeal to different target audiences is no trivial matter for direct marketing teams. The effectiveness of advertising hinges on it being relevant to the recipient. Our challenge was to design a system that predicts which campaign and contact channels will achieve the greatest possible effect on the individual customers.

Solution

In an iterative process, together with GALERIA’s purchasing and marketing departments, we piloted and launched a scoring engine. Using conventional data mining approaches, LAYA Solutions tested a subset on the scoring methods, applied them in campaigns, and incorporated the findings into further optimising the scores. The scoring engine is an application programmed in Python that is continuously developed with the help of neural networks. Its scores are historicised in the data warehouse and made available to marketing automation systems via our API gateway, so that they can be used as selection criteria for operational campaign management.

Result

  • Calculation of customer scores 37 times faster than with manual data mining approaches
  • Calculation of response probability based on 140 attributes from master and customer behaviour data
  • More than 50 different scoring models can be configured for the customer base
  • Identification of the relevant customer group according to the campaign goal
  • Definition of customer groups via “best customer score”, brand or product affinity, content relevance or contact channel affinity

0 x

faster calculation of customer scores than with manual data mining approaches

0

attributes from master and customer behaviour data for calculating the response probability

>0

different scoring models can be configured for the customer base