List rental modelling

Improve your third party data acquisition performance

For those with a list rental database, or those brands looking to expand their customer database through direct mail, telemarketing or email, we can support sustainable improved performance. Our modelling approach has been proven to increase performance for the same mailing volume by up to 30%, of sustainable performance.

What does good performance look like

For one eCommerce client their target response rate was 1.15%. Having previously struggled to achieve a cell above a 1% response rate, our modelling approach improved the performance of a like-for-like volume (the top 100k records) by 30%. We also managed to improve the performance of the second cell of 100k, enabling our client to merge to two cells together and doubling their mailing volumes.

By using an external analytics provider you can be sure that we provide accurate estimates of what can be delivered within the time frames.

How do we do it?

Our analysis have a wealth of experience in building best of breed multi-variate models. These models look at all of the information available, pick the most predictive variables and rank the whole “base” universe from top to bottom with a score. The score can be applied to any individual on the list rental database and that provides a probability of that individual completing an action (typically buying).

The output of the model is a gains chart (below) that shows how, the base is evenly split into 20 segments, whereas customers typically have high scores. This is why there is a curved line above the base. A high curved line shows the the model is predictive.

Next we typically build a suite of models, modelling “pre-selected” universes. For example, we may target over 50s only. This will make the model work harder.

We can then compare the strength of each model, and how it works for the target mailing volumes.

By having a standardized process, we can ensure that stability of campaign performance over time. In some cases this stable approach has led to a roll out volume of 33x the initial test volume of 15k.

Overview of our approach

  1. Agree KPI’s; response, initial value, Life Time Value (LTV)
  2. Retrospectively build a model to predict that KPI
  3. Build a test model & create a set of model segments
  4. Assess the performance of each segment in the previous campaign by comparing mailed vs responded
  5. Compare like-for-like test volumes
  6. Re-apply the model to live data for purpose of selections

Making it cost effective

With so many factors to consider, and modelling iterations to review, it was important that we could take the overhead out of the modelling process.

Our TP modelling process, was built to automate the data cleansing and preparation so that we could focus on adding value and interpreting the models – rather than “standard” data prep.

Read our case study to find out more:UK data provider with over 50m consumer records

Why use us?

  • Data providers
    • Improve the performance of your data
    • Increase roll out potential
    • Provide your clients with comfort that we have the best approach in place
  • Brands
    • Have an independent view of what performance you can achieve
    • Consolidate your data suppliers
    • Have access to industry knowledge to get the best deal

Want to hear more?

See our case study for a UK data provider with over 50m consumer records

Get in touch to review our standard process & documentation