How to Run a Churn Save Play on an At-Risk Account

What is a churn save play?

A timeboxed motion with one owner, a diagnosis step, and a written exit criterion. The exit criterion is what separates a play from a check-in. Save attempts often end when the owner feels better about the relationship, which is a feeling rather than an outcome, and the account churns anyway.

A play has four fixed properties. It has a named owner rather than a team. It has a deadline. It has a diagnosis step that happens before any offer is made. And it closes into a defined state that gets recorded: saved, lost, or escalated.

Build the routing before the scoring. A precise risk model with no defined play produces a dashboard nobody opens, while a crude flag with a named owner and a deadline changes outcomes.

When is it too late to start one?

Once a replacement has been selected or the budget has been moved. At that point the decision is made and price is the only tool left, which is why late saves so often become discounts that delay the loss rather than prevent it.

The behavioral signals move much earlier. The earliest indicator on any account is the absence of a signal, meaning no responses, no data changing, and no notes. On the customer side that shows up as dropped business reviews and unanswered scheduling requests, and it appears months before a renewal conversation.

Support volume gives a second read, and it works as a curve rather than a line. ORM data shows accounts filing zero support cases carry elevated churn risk, since nobody is using the product hard enough to hit a question. Accounts filing seven or more cases in a year also carry elevated risk. Accounts filing three to five ordinary tier 2 or tier 3 tickets are the healthiest group. Both tails deserve a play. The middle deserves to be left alone.

What are the steps?

Diagnose, sponsor, commit, prove, close. Run them in order. The most common failure is jumping to an offer before anyone knows what actually broke. The diagnosis step is worth more than the rest combined. Talk to someone who uses the product daily before you talk to the person who signs. The daily user knows whether the workflow works. The signer knows only what they have been told.

Who belongs in the room?

One owner, one matched executive sponsor, and one person who can commit to a fix. Three people, and everyone knows why they are there.

The owner runs the play and holds the deadline. The executive sponsor exists to match seniority on the customer side, since a director-level relationship cannot reopen a decision made by a vice president. The technical or product resource matters because a commitment without a delivery date is a promise, and at-risk accounts have usually heard promises already.

Resist the urge to add attendees. Larger groups slow the sequence, and the timebox is doing real work here. An account gets one focused effort or it gets a long series of well-meaning conversations.

When should you offer a discount?

Only when the objection is genuinely price and the account is otherwise healthy. A price concession on an account that stopped getting value buys a single renewal cycle at a lower number and delivers the same loss later, with less revenue collected in the meantime.

Test the objection before you respond to it. Ask what would need to be true for the current price to be defensible internally. A budget-constrained account with strong utilization gives a concrete answer, often about term or payment timing. An account that has disengaged cannot answer, because price is the socially easy reason rather than the real one.

When you do concede, take something back. Term length, a case study commitment, or a reference agreement all convert a giveaway into a trade and keep the precedent from spreading through the base.

How do you measure whether save plays work?

Save rate against a holdout, not save rate alone. The share of worked accounts that renewed at or above prior ARR looks impressive on its own and proves nothing, because many flagged accounts were never going to leave. Hold back a sample of flagged accounts, work them normally rather than with a play, and compare.

Report two other figures. Coverage is the share of churned ARR that carried a flag more than 90 days before the loss. Timing is the median days between flag and play start. Rising coverage with a flat save rate means the signals work and the plays do not. Flat coverage means the signal set is missing something. Those two failures look the same on a churn chart and need opposite responses.

Require a written outcome note on every closed play. Two sentences on what was found and what changed. A year of those notes becomes the most useful retention dataset in the company, because it records which signals preceded real problems. Judge the whole program against gross retention rather than net revenue retention, since expansion can hold NRR flat while the loss rate accelerates, and apply the same grading discipline described in sales forecasting best practices.