layout: page title: “Probability To Convert Prediction” permalink: /conversion_prediction/

Probability To Convert Modeling

Problem

If you don’t understand what makes a consumer more likely to convert, your marketing is likely to be directionless or driven by hunches and general heuristics.

Understanding what behavioural actions make a user more likely to convert has a twofold benefit. Firstly, you can accurately target users that your actions will have the largest impact on. Secondly, you can understand what behavioural actions you should be driving users towards.

Impact

Adtech client - Optimised bidding algorithm, how much to spend against the marginal impact of a new session. B2B client- Better understanding of what website sections & email types made users more likely to convert realigned the content team’s strategy towards webinars.

Method

Flexible, gridsearched, GAM models were able to model the dynamic and semi-linear distributions of user behaviour accurately while also providing interrogatable prediction curves and smooth simulated responses for marginal impact analysis.