Churn analysis for Techdata
We used machine learning to improve marketing effectiveness, including when communications to specific customers should be sent to reduce the chance of churn.
Challenge
TechData is a global technology distributor who sells the likes of Apple and Dell products to high-street retailers. TechData chose Satalia to boost the effectiveness of their marketing, using years worth of data they had collected.
Outcomes
84 days
communications to reduce churn were sent 84 days earlier that previous system
Outcomes
84 days
communications to reduce churn were sent 84 days earlier that previous system
Solution
We used machine learning to detect changes in the behaviour of each of their B2B customers, including purchase frequency, regularity and volume. Using this data, we built individual churn models for each customer, predicting exactly when they would stop buying. This proved significantly more accurate than a generic churn model, which failed to account for the diverse behaviours in their customer base.
In addition we developed a ‘change point analysis’ model, which uses variances in product baskets to identify the point of the customer journey at which they decide not to buy.