Subscription Churn Prediction: Your SaaS Customers Are About to Cancel—Can Beancount Detect the Warning Signs Early?

General
1

We manage around 40 SaaS customers. One of them dropped from $15K MRR to $12K over two months without warning. We didn’t catch it until they canceled last week.

Got me thinking: are there early signals we’re missing? Like, if a customer’s API usage drops, or their support tickets stop, does that predict churn 30 days out?

For Beancount practitioners with SaaS clients: can you tag revenue by customer and set up alerts for MRR decline? Or is this what separate business intelligence tools are for, and Beancount just shouldn’t try?

Curious if anyone’s built churn-detection queries.

2

Lost a major client (35% of revenue) once. In retrospect, signs were there: delayed payment by 2 weeks, fewer questions/support requests, shorter meetings. But I wasn’t watching for it.

Now I track customer health quarterly. Revenue, engagement, communication frequency. Not automated, just a spreadsheet. Caught two churn-at-risk accounts last quarter. One was savable with a discount.

Early detection isn’t magic, but it matters if you’re paying attention.