Using predictive models to reduce customer churn by more than 20% with 10x ROI

The customer retention strategy was based on randomly contacting as many of the customers whose contract would expire in the next couple of months as possible, and giving hefty discounts to the ones that have already canceled their contract. The client had a huge number of product bundles (aka. offers) that were created “just in case”. The number was so high that it was literally impossible for salesforce to learn how to sell the services, and they ended up learning a few selected offers that were the easiest to sell.

The staff was lacking information about which customers were most likely to churn and what the factors impacting such risk were. The feedback loop on how employees’ actions and campaigns affected sales metrics (especially churn) was extremely long; as a result, decisions on who to contact and what should be offered were based on gut feeling of front-line sales force and their managers instead of hard data.

Many of the employee’s actions were based on guessing and randomly choosing customers to contact or offers to make, so it was difficult to reduce churn, and using the old strategies and techniques made it virtually impossible. Instead of making use of data, they were trying to make use of their intuition. As our client’s employees put it: “We were blind on one-and-a-half eye”.

The pilot was initially scoped for 10 months, including:

• 360 view of each customer,
• predictive models enhancing 360-view with churn prediction
• predictive models enhancing 360-view with product recommendations,
• tools delivering suggestions on how to take care of churn-prone
• customers to front-line employees,
• integrations with client’s systems,
• feedback loop to keep predictive models up to date. After that time, we expected to have initial results on how our solution helped with churn reduction in initially selected customer segments.

We did much better than what had been planned. In just 2 months we released the initial version of 360 view and churn models that we tested on a small sample of customers. That allowed us to optimize those models within 4 months since project inception – initial recommendations for customer retention campaigns based on our solution were available almost half a year earlier than planned.

During the churn model pilot, we trained product recommendation models – that allowed us to increase campaign conversion rates almost twice, not only for churning customers but also for the whole segment.

After additional 2 months of testing a combination of churn prediction and model recommendation, and excellent results (see below), our client decided to roll out our solution for full customer base and introduce machine learning into other departments.