Case Study: Colruyt achieves rapid churn modeling at scale with TIMi

Churn Model On Structured Dataset

Colruyt, the third largest retailer in Belgium, wanted a churn model and also a way to predict “partial churn,” where customers stop buying certain product categories while continuing to shop overall. They worked with TIMi and its Anatella platform to build the models from a very large structured dataset of purchase history.

TIMi delivered a churn model in two days and then produced 36 additional category-specific partial churn models in two more days. Using Anatella’s automated data-transformation and model-factory workflow, the team processed 1.4 billion rows of raw data into a 1,012-column output table and updated all 36 models in about one hour; the final left-join took just 9 seconds on an ordinary laptop.

Colruyt