Run voice-of-the-churned analysis
Most accounts that leave gave warning signs months before the renewal, buried in calls and tickets no one went back to read. NEXT reads the conversations from accounts you already lost, finds the language and concerns that showed up before they left, and watches your current accounts for the same pattern. Customer Success gets an early-warning read: a plain description of what tends to come before churn, plus the live accounts now showing it.
What the early-warning read looks like
Example output based on grouped conversations from past churned accounts, matched against the current book.
Pattern built from
14 accounts that churned over the last four quarters, drawn from their calls, support tickets, and QBR notes.
What they had in common before leaving
The exit surveys mostly said "price" or "reorg." The conversations said something earlier and more specific: reporting that never answered the questions their executives asked, and a champion who left without a handoff.
"We keep exporting everything to a spreadsheet because your reporting still can't answer what my VP asks in the weekly."
"The person who brought you in left in March. Nobody really picked it up, and we never got going again."
Current accounts showing the pattern
9 accounts, mostly mid-market, including three with renewals in the next two quarters.
Commercial exposure
About $1.4M ARR sits in accounts now matching the signature.
Where the signal is strongest
Clear and repeated on the reporting complaint and the unreplaced champion. Mixed on a third precursor — slowing usage in a single team — which showed up in some churned accounts but not enough to weight heavily.
Demand summary
The accounts that left were not loud. They quietly stopped getting value from a core workflow, and the person who used to advocate for them went silent. Both are visible in current accounts today.
The read is ready before the QBR, not reconstructed the week a renewal slips.
How NEXT does this
NEXT reads where churned accounts spoke — calls, support tickets, survey responses, and account notes — and groups the language and concerns that recurred in the months before they left. It builds a description of those precursors and keeps a running record of which current accounts use the same language or hit the same friction. When a live account starts matching, NEXT writes the match into the account record and can notify the CS team where they already work, with the specific quotes and the affected workflow attached. It does not decide the account is at risk or change the renewal forecast. It surfaces the pattern and the proof behind it; the CS lead reads it and decides whether to act.
Why churn risk surfaces late today
The signs are usually there. The problem is that no one has time to reread two years of conversations across a hundred accounts to find them.
Exit surveys arrive too late and too clean — by then the account is gone and the stated reason is a tidy headline, not the real story. Health scores compress everything into a number that tells you something moved without telling you why. And the original detail decays at every step: a frustrated quote in a call becomes a line in a CSM's notes, then a yellow cell in a dashboard, then a half-remembered worry in a renewal review.
The weekly review still depends on someone remembering to open the health dashboard and read between the cells. Ask an AI assistant and you get the loudest recent thread, not the quiet pattern that played out across a quarter. Neither comes looking for you.
NEXT pushes the read to the team instead of waiting to be opened or asked. It watches each account against the pattern drawn from real churn and surfaces the match as an action with the quotes attached, not a chart someone has to go find.
How this compares to the tools you already know
Approach | Where the signal lives | What CS does at decision time |
|---|---|---|
Exit surveys | A form filled out after the account leaves | Reads the stated reason, too late to change it |
Health-score dashboards | A composite number per account | Sees the score dropped; reopens notes to find out why |
AI assistant | Whatever you remember to ask about | Gets the loudest recent thread, not the quarter-long pattern |
NEXT | A living record matched against the churn signature | Reads the named precursor and the quotes, then decides whether to intervene |
What changes for the CS lead
Today you find out an account is at risk when usage craters, the champion stops replying, or the renewal date is six weeks out and the deal has gone quiet. By then your options are narrow.
With the early-warning read, the same risk shows up earlier and with a reason attached. You open your book and three accounts are flagged not because a score dipped, but because they are using the exact language two churned accounts used last year — the same reporting complaint, the same disappearing champion. The account looked healthy on the dashboard; the conversations said otherwise.
One mid-market account looked fine on every number until NEXT matched its tickets against the signature: the admin who ran the integration had left, and the replacement was guessing. That is a save you can make in week one, not a post-mortem you write after the loss.
NEXT already supports product and GTM teams at companies like Deel and Visma in connecting customer evidence from calls, tickets, and reviews to the decisions teams make. Here, the same evidence feeds your renewal calls.
The judgment stays with you. NEXT brings the pattern and the quotes; whether an account is genuinely at risk, and what intervention fits, is your call.
Downstream effects
Earlier saves are cheaper saves. Catching the reporting complaint two quarters before renewal means a fix or a workaround, not a discount fight in the final weeks.
Product hears the real precursors, not the exit-survey headline. "Price" turns into "the reporting can't answer executive questions," which is something the product team can actually act on.
QBR prep gets shorter. The account's risk signals and the supporting quotes are already assembled, so the brief starts from current signal instead of an hour of digging.
Where the human stays in control
You set how close a match has to be before NEXT surfaces it, and you can require a human to review matches before they are written into the account record. Set the bar high and you see only strong, repeated precursors; set it lower and you catch fainter patterns at the cost of more to read. That is configuration you tune over time, not an approval queue you babysit. NEXT never marks an account as churning or moves a renewal forecast on its own — it presents the match and the evidence, and you decide what it means.
What to configure first
The signature is only as good as the churned accounts it learns from. Make sure NEXT can read the full conversation history for accounts that actually left — calls, tickets, surveys, notes — not just the closing exit survey, which is where the real signal is thinnest. Decide which churn counts: a contractual non-renewal and a quiet downgrade may have different precursors, and lumping them together blurs the pattern. Set your match threshold deliberately, and agree as a team what happens when a match lands — who reads it, who reaches out, and on what timeline. Expect the signature to need a refresh each quarter as new churn teaches it new precursors.
Where this breaks down
Thin history on the accounts that left
If your churned accounts have almost no recorded conversations — a few tickets and an exit survey — there is little for NEXT to learn from. The signature will be vague, and vague signatures produce weak, low-confidence matches.
Churn that has no early signal
Some accounts leave for reasons that never showed up in conversation: an acquisition, a budget freeze, a sponsor who simply changed jobs. Those will not have a readable precursor, and the pattern should not be forced to explain them.
Over-broad matching
If the threshold is too loose, half the book matches a precursor as common as "wants better reporting," and the signal drowns. Calibrating the threshold reduces this, but it takes a few cycles to find the level where matches are worth reading.
Treating a match as a verdict
A match means an account resembles ones that churned, not that it will. Read as proof of doom, it creates false alarms and wasted outreach. It is a prompt to look closer, not a forecast.
FAQ
How is this different from a health score?
A health score compresses an account into a number and tells you it moved. It does not tell you why, and it usually reacts to usage that has already dropped. This works the other way: it starts from the language real churned accounts used before they left, then watches for current accounts saying the same thing. You get a named reason and the quotes behind it, not just a color.
Does NEXT decide which accounts are at risk?
No. NEXT surfaces accounts that match the pattern drawn from past churn and attaches the supporting quotes and workflow. Whether an account is genuinely at risk, and what to do about it, stays with Customer Success. The match is a prompt to look closer, not a forecast or a forced action.
What if our churned accounts left for reasons we'll never see coming?
Some will — acquisitions, budget freezes, a sponsor changing jobs. Those leave no readable precursor, and the signature should not pretend to explain them. The value is in the churn that did give warning signs in conversation, which in most books is a meaningful share. NEXT marks where the signal is strong and where it is thin so you know which matches to trust.
How far back does it look?
It reads the full available conversation history for the accounts you point it at — calls, tickets, surveys, and notes — rather than a fixed window. The further back and more complete the record on accounts that left, the sharper the precursors it can find. Sparse history produces a vaguer signature and weaker matches.
How often should we rerun it?
A quarterly cadence fits most teams, since each new set of churned accounts can teach the signature new precursors. Between refreshes, NEXT keeps matching current accounts against the existing signature, so you are not waiting for the next analysis to hear about a new match. Rising churn-risk match counts between refreshes can also signal that the signature has shifted — worth a mid-cycle check when the volume spikes unexpectedly.