Our client, one of the biggest airlines in Europe, wanted to quantify its customer churn rate, understand how it differs between different customer segments and proactively manage it, minimising lost customers and sales.
Customer registration, loyalty and sales data were stored in multiple different data sources, often under different customer IDs and/or duplicates. The first step was to clean, merge and deduplicate the records, creating a single source of truth customer dataset. Once that was in place, describing churn through a series of BI dashboards, allowing business users to drill-down to different time periods and customer segments so they can see how churn evolves over time was quite straightforward.
The next step was to apply ML and build a churn prediction model. The model was built over a number of iterations, applying different approaches, eventually managing to reach an > 85% accuracy, in terms of predicting a customer’s probability to churn. The model would then cluster customers into different groups of churn risk, enabling our customer to have a proactive tool in terms of managing churn and safeguarding tens of millions of lost to the competition revenue.
Since then, we have applied churn models to multiple customers across various industries such as Retail, FMCG, e-commerce and Financial Services with success and a substantial analytics ROI, especially if you couple churn prediction with A/B campaign management. A very practical and extremely effective means of boosting retention and customer lifetime value.
- Protect millions in terms of lost revenue
- Prevent churn before it happens
- Very high Analytics ROI