Lead scoring propensity model drives 20% increase in conversion for business comparison service

Profile

PE-backed B2B price comparison and switching service across a range of areas including electricity, gas, insurance, and telecoms.

Situation

A PE-backed price comparison service wanted to optimise their outbound sales funnel with the aim of increasing monthly sales by improving conversion rate. They wanted to the ability to rank and prioritise their lead base in order to better allocate their finite outbound sales team resource towards higher-value potential customers.

Action

The engagement initially focussed on generating a deep understanding of the mechanics of the business’s sales funnel, in order to ensure that the subsequent modelling work was optimised for practical impact. This involved the careful construction of a modelling dataset as well as comprehensive analysis to identify key individual drivers of lead value. The modelling stage was an iterative process where various scenarios and configurations were tested in order to converge on the optimal solution that was most effective at predicting valuable leads.

Following the construction of the Machine Learning model, we worked closely with the business to implement the solution within their systems and processes as well as design a robust testing framework to calculate and monitor model performance into the future.

Tools and techniques used in this work

• Advanced SQL

• Machine learning algorithms

Impact

The business ran a month of testing of the model to ensure its effectiveness in practice, before rolling the scoring algorithm out across the entire outbound sales funnel. The model testing and reporting framework allowed uplift to be measured easily, with a 20% increase in conversion rate recorded.

The additional insight generated alongside the core modelling has also been used to advise optimal call strategies for leads with certain profiles, driving incremental improvements.

So what?

QuantSpark’s hybrid strategy and data science consultants were able to effectively design and develop a highly sophisticated Machine Learning model for the client, whilst ensuring that the solution was rooted in commercial understanding and business practicality. This approach ensured the maximum potential was achieved with the model, driving real, measurable business impact.