Guide

Churn prediction for B2B SaaS: what works and how to test it

Churn prediction is the practice of flagging the accounts likely to cancel or shrink at renewal, early enough to act. In B2B SaaS it works differently from consumer churn prediction: fewer customers, bigger accounts, and most of the warning in conversations rather than clicks.

Three approaches

ApproachHow it worksStrong whenWatch out for
Rules and health scoresPoints or bands from usage, tickets, NPS and CRM fieldsYou want a score everyone can understand and auditAverages that let one weak area hide; stale manual fields
Machine learning on outcomesA model learns which patterns came before past churnYou have many renewals and clean historyToo few churn events to learn from; hard to explain a single score
AI that reads conversationsLanguage models read calls, tickets and email for riskMost of the warning is in what customers sayMade-up or unsupported claims unless every risk must cite its source

They aren't exclusive. A practical setup uses AI to find evidence in conversations, and plain rules to turn that evidence into a score people can check.

The data you need

How to test any churn prediction

Run it on last year's renewals as they looked before the renewal date, then check:

  1. Coverage of churned revenue (recall): what share of the ARR that actually churned was flagged 90, 60 and 30 days before the renewal?
  2. Precision: of the accounts flagged Red, how many actually churned or shrank?
  3. Lift: how many times more likely is a Red account to churn than the average account?
  4. Lead time: how early did the first warning appear?

Judge precision against your base churn rate. If 10% of your renewals churn and 30% of the accounts flagged Red churned, the score finds churn 3 times better than picking accounts at random.

Avoid leakage (using information the model wouldn't have had at the time). If the model uses a cancellation notice to "predict" a cancellation, it will look accurate and help no one. Test with what was known before the decision.

Questions to ask a vendor

Horizon combines AI that reads conversations with a rules-based health score, lets admins backtest it against past churn, and measures actions against a holdout group.

Updated November 3, 2026

See which of your renewals are at risk.

A 30-minute call on a demo book, or on your own top 25 accounts.

Book a call