Meal-kit churn
2019
11% better 3-month retention
Finding the week customers decide to leave, and what changes their mind.
The question
Meal-kit churn started with one question from the team, asked in plain words. I wrote it on the first slide and kept it there until the end.
The data
Two years of transactions, an export from the CRM and a spreadsheet someone had kept by hand. Most of the work was making them agree.
What I found
One driver explained most of the problem. The rest was noise that had been on every dashboard for a year.
What changed
11% better 3-month retention. The model now runs every morning, and the team owns it.