Cornhill new policy acquisition
20:20 Dialogue reduces cost per sale by 40%.
Brief
Cornhill Direct offers life, home and motor policies directly to consumers over the age of 50. Direct mail is the primary channel for their acquisition activity, with sales enquiries handled through the company's call centres.
Cornhill mailed approximately 35,000 motor policy prospects each month, sourced from their database of previous quotes not taken up. They wanted better targeting of this monthly activity, whilst maintaining mailing volumes in order to sustain business growth.
This called for Cornhill to adopt different communication strategies for segments of their prospect database, leading to more cost effective mailings. The initial requirement was to develop their understanding of converters to the monthly requotes activity, and use this to develop a targeting model.
Approach
20:20 Dialogue believed that a model would enable Cornhill to segment their mailing files based on the likelihood of each prospect converting. This would provide an opportunity for differential investment and varied communication strategies, thereby boosting conversion in certain segments whilst minimising the reduction in conversion in others.
20:20 Dialogue extracted a sample of prospects who had been quoted over an eighteen month period. This represented a balance between obtaining the most recent data but also allowed for people who had received more than one quote (which might prove to be related to their likelihood of conversion). All data collected at the point of quotation for both converters and non-converters was available for profiling purposes.
The data that discriminated most effectively between converters and non-converters were used in logistic regression modelling. The resulting model combined the various predictive fields into a single score, representing the likelihood of an individual converting, based on their attributes at the point of application.
The model was used to rank a sample of known converters and non-converters by their predicted likelihood to convert - it showed some impressive results. The model predicted a 25% conversion rate for the top 10% of scores compared with only 2% for the bottom 10% of scores. Furthermore, the top 10% of all prospects were nearly 3 times as likely to convert compared to the file as a whole. The model therefore showed encouraging signs in its ability to discriminate.
20:20 Dialogue ran a campaign planning workshop with Cornhill's direct marketing team. This provided a platform to share the results of the analysis, understand the model and plan a specific test campaign that made best use of the new targeting tool.
Given the need to maintain mailing volumes, Cornhill decided to use the model to develop a three-fold communication strategy to the top, middle and bottom scoring bands of the model. The best scoring bands (top 40%) would receive the regular marketing pack and two-stage mailing approach, the middle group would be incentivised and the bottom group (bottom 20%), with the lowest likelihood of conversion, would receive a low-cost postcard. The initial test was run over a period of three months.
Results
The principal objective of the test was to reduce cost per sale through reduced mailing costs and increased conversion rates. Cornhill's mailing costs for the requote activity were reduced by 10% and reinvested in activity elsewhere.
The reduced pack price for the lower groups helped to make them the second least costly to recruit, and due to increased response and conversion in the middle and upper groups, overall cost per sale reduced by 40%.
Don't just take our word for it
Mike Hayward, Cornhill's Direct Marketing Manager said "We are delighted with the support provided by 20:20 Dialogue which has enabled us to develop and implement a sophisticated targeting approach in a relatively short space of time. It is now a key part of our sales activity and continues to provide significant benefits in terms of reduced acquisition costs."