Understanding predictions and signals

Threadline's Predictions and Signals features are designed to help you act before problems become crises.

What are signals?

Signals are patterns in customer language that historically precede bigger experience breakdowns. They're not just complaints — they're early indicators. For example, customers starting to say "used to be" or "getting worse" are flagging deterioration before it shows up in NPS or churn rates.

What is churn risk score?

The churn risk score is Threadline's estimate (on a scale of 0–100) of the likelihood that customers in this cohort will disengage or stop using your product. It's based on sentiment trajectory, friction concentration, and language patterns.

A score above 70 suggests significant churn risk that warrants immediate investigation. A score below 30 suggests the cohort is relatively stable.

What is revenue at risk?

Revenue at risk is an estimate of the ARR (or transaction value) exposed to churn based on the churn risk score. It's calculated using aggregate signals, not individual customer data.

This number is an estimate, not a guarantee. Use it as a prompt to investigate, not as a financial forecast.

Foresight quality

Predictions are only available when you run an analysis at Foresight quality level. This uses more credits (2 per analysis) but adds the full predictive layer to your results.

What should I do with predictions?

1. Identify the high-risk journey stages — the Predictions tab will tell you which stages are contributing most to the risk score.
2. Cross-reference with Themes — look at what customers are saying in those stages.
3. Act on the top 1–2 recommendations — Threadline surfaces specific actions you can take to reduce risk. Focus there first.