Guide

AI churn prediction: what it adds and how to trust it

AI churn prediction is the use of a language model to read every call, ticket and email on an account and flag the risks in what customers say, which no CSM has time to do by hand. The risk is that it sounds sure when it shouldn't. The guardrails below keep that in check.

What AI is good at here

For how AI compares with health score rules and machine learning, see churn prediction for B2B SaaS.

Where it goes wrong

Guardrails that make it trustworthy

  1. No quote, no signal. Every risk must cite dated evidence, or it's thrown away.
  2. AI finds evidence; rules do the scoring. Keep the arithmetic out of the model so the score is stable and explainable.
  3. Merge duplicates by evidence. Two signals that cite the same call or ticket are one signal, however they're worded.
  4. People close risks. AI can suggest a risk looks resolved; a person decides.
  5. People approve customer email. Drafts are reviewed and sent by the account owner.
  6. Respect permissions. Each person sees only the accounts and data they can already see.
  7. Measure it. Compare accounts you acted on with a random holdout, so you know whether acting on the signals changed renewals.
  8. Know where the data goes. Check which model provider processes your calls and email, whether it's used for training, and how long it's kept.

A rollout plan

  1. Connect the CRM and one conversation source (calls or tickets).
  2. Review the first week of signals with the team: which were right, which weren't, and why.
  3. Add the other sources, then turn on scoring.
  4. Start one playbook, with a holdout, and review results after one renewal cycle.
  5. Keep tracking the share of signals the team accepts or dismisses. If it falls, find out which kind of signal is wrong.

This is how Horizon is built: see renewal risk signals and playbooks and outreach.

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