Customer Profiling and Analysis: Part 3 – Proving the Concept
In recent blogs I have been looking at how customer profiling can be used to pinpoint prospects, subsequently improving the accuracy of marketing and increasing ROI. In this third instalment I hope to demonstrate the true value of customer profiling through highlighting a recent ‘proof of concept’ exercise completed for a client.
Proving the concept of customer profiling is a consistently valuable activity that allows both us and the client to identify marketing opportunities. It does this by providing statistical evidence of the worth of a particular market segment, rather than relying upon supposition. It is also the ideal method of evaluating the potential for larger national campaigns, prior to significant investment.
Proof of concept exercise
In our recent work for one of our clients we used this approach to identify data segments within the Consumer Universe which represented the largest potential for direct marketing and ultimately lifetime customer value.
To complete the exercise we used data from a particular outlet within the client’s retail chain to create a customer profile. This profile was based upon a range of demographical information such as home ownership, property type, income and weekly supermarket spend to build a view of the clients’ ideal customer.
Using this understanding it was then possible to analyse the data, identifying the same characteristics within the prospect data and segmenting it using a system of propensity scoring.
Propensity scoring is based upon the propensity of a particular record to convert to a customer; it achieves this by ranking individual variables in terms of respective influence. For example, weekly supermarket shop may have a greater impact upon propensity than property value, which would increase its predictive weight of evidence (PWE) and consequently its affect upon the propensity score for that variable. If you would like to learn more about propensity scoring, please read Customer Profiling & Analysis: Part 2
In this particular instance the data was divided into 10 segments based upon propensity scores of -8.13 (lowest propensity) to 3.12 (highest propensity). These segments were then sent direct mail marketing materials. After 6 weeks, conversions from the mail campaign were measured based upon specific metrics from the client. The results can be seen below:
The results clearly show that those segments with a higher propensity score achieved a higher conversion rate, whilst the trend graph reveals that those segments with a positive propensity score generate the largest conversion rates. As part of the process, another analysis exercise will be completed again after 21 weeks, to prove the concept further.
Going forward this now allows us to understand which customer segments have the greatest potential for conversion. Once applied on a national scale with the wider Consumer Universe it will be possible to only select those records with a positive propensity score. Subsequently, this will improve the targeting of prospects for the client, boosting the response rates of campaigns and improving the ROI of direct marketing.