Hiscox: 60/70/80

Making data work harder

Objective

To help Hiscox deliver their ambitious targets to grow their market share of mid net worth home insurance customers we needed to increase the number of new customers acquired by direct mail without increasing the budget.

Strategy

Our strategy was to look to increase mailing volumes of qualified prospects by identifying new sources, while simultaneously improving the overall efficiency of this activity. Working with our data planning department and data bureau, we looked to achieve this by maximising the use of Hiscox’s prospect pool.

The prospect pool contained profiled Lifestyle data, both with and without renewal month. We had already proved that we could make profiled Lifestyle data work for Hiscox without the key variable of renewal month. The challenge was to make the most of this prospect data while minimising the risk to our client. Direct Mail results were typically taking 60 to 90 days to mature, but we didn’t want this fact to delay our plans to expand our volumes.    

What we did

We built a bespoke response model using data from previous direct mail campaigns and the variables held in the prospect pool both to improve targeting and to highlight the value of other data picked up in the profiles. The model was designed to enable all records in the prospect pool to be scored, ranked and selected according to their propensity to respond to a direct marketing offer. This would also offer additional selectivity on the prospect pool over and above the presence of renewal month and the Lifestyle profile score.

In addition to using the variables held in the prospect pool, w e audited the range of industry standard geo-demographic tools available to help verify and improve the response model. We produced a gains chart to highlight the best combination of products, and this highlighted CACI’s ACORN and StreetValue as the most indicative geo-demographic tools to overlay the response model.

The records on the data pool were tagged with their response model score and their geo-demographic indicators, and an initial framework was developed. The data was initially split by ACORN categories, with StreetValue scores applied. The response model was then introduced.

We then carried out a retrospective match back using previous campaign data to evaluate the likely effect of using the geo-demographic selections and the response model to make data selections. This analysis showed that on average 60% of the volume mailed had produced 70% of the response and 80% of the sales. This also highlighted the parts of the model that were generating the most responses and sales.

Results

We tested this model with a high degree of confidence that it would work effectively by generating more response and customers, driving down our cost per sale. The subsequent 60 days saw immediate improvement in response and conversion rates, and our average cost per sale dropped by over 35%.

We have continued to develop this model to refine our selections further within the framework of ACORN Groups, Categories and Types, StreetValue Bands and the Grades within our response model.  

Summary                 

Since the introduction of the model, the average CPS on a month by month basis has consistently fallen far below the target average CPS, while mailing volumes have actually increased by 150% year on year.