We collaborated with an online marketing platform to increase their A/B testing capabilities, allowing them to fully measure the impact and optimise their marketing to drive repeat customer purchases.

An online marketing platform that helps companies to incentivise their existing customers to make repeat purchases and refer their friends. 


With clients operating in the highly competitive online retail sector, the company needed to find a reliable method for determining which offers encourage customers to make repeat purchases and increase their spend in the future. Various types of offers were being shown to customers after they made purchases online, but the company needed to ensure its performance calculations were robust, scalable, and able to work for a range of different clients. This meant determining which offers encouraged customers to make repeat purchases, thereby maximising revenue for clients. 

The company enlisted Faculty to help them reliably determine which of its offers perform best, measured by their impact on future customer purchasing behaviour. Ultimately, this would inform its clients on which offers to roll out more broadly across their customer base.


We carefully modelled the distributions of the repurchase rate and spend of existing customers and used a Bayesian approach for computing the confidence intervals for these metrics, for each offer within a given A/B test. This allowed us to infer the spread of likely outcomes (repurchase rate and spend) we would see if the offers had been sent to the entire sample, rather than only the sub-sample that were randomly assigned to them. 


The approach we developed allowed for continuous iteration within an A/B test, balancing speed and certainty with respect to tracking the performance of a campaign. The client decided to deploy the work into their own platform so that the approach could be integrated across all campaign types, allowing rapid, data-driven decisions to be made across the business.  

The team now has much greater confidence in their approach to analysing the performance of their A/B tests and a clearer understanding of how customers respond to various offers.