Attrition modelling for payment cards
This client was experiencing significant and growing attrition across its B2B SME base. It had little experience of data analytics but needed to find a solution that would alert the business as soon as it appeared that a customer may be looking to leave.
What we did
Beyond Analysis extracted all of the customer and transaction data and hosted it in a data lake where it was cleansed and made fit for analysis.
Every individual customer was profiled against their behaviours and product usage and machine learning techniques were used to develop an automated attrition prediction model.
To supplement this and facilitate the actions a value based segmentation was developed to enable the business to know 'where to invest' and prioritise its efforts in retention.
An automated data feed is now enabled that continuously runs and optimises the models with data APIs into Salesforce to drive the trigger actions within the CRM retention plan.
Attriting customers can now be identified with up to 25 weeks of runway and attrition has been reduced by 18%.
We chose BA because they have a strength in analytics and they are more dedicated, engaging and also capable of dealing with the results.
Vice President, Business Performance