Churn and sentiment scoring: reading the risk flags

2 min read · Updated Jul 17, 2026

Current watches for two different flavors of trouble: a slow sentiment drift picked up from support and activity patterns, and a sharp signal from customer surveys. Both are meant to catch a problem before it shows up as a lost renewal.

The signals that feed it

The sentiment engine behind each company's AI read (see "The AI company read: what it analyzes and when it refreshes") compares the last 30 days against the 30 days before it, and the last 14 days of email activity the same way. It's looking for:

＋A rise in ticket volume or reopened tickets versus the prior period.
＋Cooling email or activity cadence — an account that used to hear from you regularly going quiet.
＋Survey scores trending negative, when your CrewHu satisfaction surveys are connected.
＋Seat or headcount shrinkage on the account's contract.
＋Escalation language in ticket and email text — repeated frustration, urgency, "still not fixed" phrasing.

Separately, if you have CrewHu connected, a negative or neutral customer survey response can trigger its own alert the moment it comes in, independent of the nightly company read — that one's about a specific bad experience, not a slow drift.

Where it surfaces

＋The company record's AI read shows an overall sentiment (healthy, watch, or at-risk) with a score, plus the specific highlight that drove it.
＋The Relationship Health tile on your Sales home and the team cockpit ranks every customer by an overall churn score — a negative or neutral survey is one of several signals feeding that score, not its own separate panel. See "The sales dashboard: pipeline, pace, and the leaderboard" for the full ranking.
＋Radar briefings and morning briefs call out accounts whose sentiment dipped or that crossed a churn-risk threshold, alongside renewals and pipeline movement — see "Radar briefings and morning briefs."
＋A direct notification (bell and, if enabled, email) can reach the account manager and sales leadership when a survey alert or churn-risk threshold fires.

What to do with a flag

A flag is a prompt to look, not a verdict. Open the company record, read the specific highlight that triggered it — which ticket cluster, which survey, which seat change — and judge it against what you already know about the relationship. If it's real, a quick check-in call or a note to the account team usually resolves more than any dashboard can. If a survey flagged, that's often the fastest one to close: reply to the customer directly, then create a follow-up from the Relationship Health tile so the account steps aside once you've acted.

Note

Nothing here takes action for you

Current never emails a customer, changes a deal stage, or escalates a ticket on its own because of a churn flag. A human always decides what happens next.

Tip

It's a worklist, not a permanent scoreboard

Creating a follow-up from the Relationship Health tile holds that account off the list for 21 days so the next-riskiest one can surface, whether or not the follow-up is finished yet. With nothing flagged in your book, the tile says so plainly instead of just disappearing.

Was this helpful?

← Previous

The AI company read: what it analyzes and when it refreshes