Case Studies

Who says rental list modelling can’t be more versatile

When it comes to selling and buying data, most providers recognise they need to deliver the commercial insights and benefits their customers seek. This is often easier said that done.  Whilst data can be selected to meet a particular criteria, it isn’t always easy to deliver all the additional insights and selections customers want.

A classic example of this was when a multi-channel data agency came to us.  They specialised in optimising data for better customer acquisition and presided over a data set which held approximately 200 variables on 50million UK consumers.

When customers came to purchase their data for a specific selection of that set (for example females aged between 25-35), the organisation struggled to then rank the selection by additional measures, such as likelihood of becoming a customer.

Their data systems weren’t able to provide the required scenarios or performance insights of the original data set, in order to distinguish further within standard selections.

 

Creating a deeper and wider modelling approach

There was therefore an appetite for a more versatile and commercial approach in how they organised, packaged and sold their modelled data.

Turning to The Fusion Analytics’ team, we created an automated process which took a set of their customer data and applied a greater number of searchable variables – 700 to be precise.

Armed with this, our analysts quickly built a multi-variate regression model that ranked the 50m people according to certain desired outcomes – for example, the likelihood of becoming a customer.  In doing so, we applied rigorous quality checks including testing and validation samples, profiling, and forensic analytics.

 

Impressive deliverables

The time-saving and easy to use outcome has meant our client can now cost-effectively:

  1. Provide a list of viable Urns to their customers (Urn models are simple ways to represent real life probabilities. In statistics, an urn model is an idealised way of modelling real-life scenarios)
  2. Create a set of models and volume offer for their customers
  3. Have a model report ready within 3-5 days
  4. Have a selection of urns provided back with a testing plan within 5-7 days

With a known time-line, outputs and price our client’s sales team have become more confident in selling the modelled data.  By no longer needing analytics support to define and package specific orders, they have been able to bring a faster production and delivery time to their customers, and a more profitable approach to their business.

 

A long-term outlook

We also produced a range of models and source coded selection files, so our client’s organisation had an ongoing test and learn strategy to apply to their data as new customer needs arose.

 

Initial results from our new partnership

We are pleased to say that the initial results have demonstrated that we can increase performance for the top cell (10% of selection) by 72%.

Overall, we can provide a like for like volume and out perform the existing selection process by 36%. Enough said!

 

Can we help?

If you find your data sets lack the versatility you need to gain competitive advantage, or you want an objective view of your data modelling approach, do get in touch for an initial chat.

Leave a Reply

Your email address will not be published. Required fields are marked *