You Cannot Save a Renewal in the Last Thirty Days
The year I took over a book, churn and downgrades were bigger than everything new business brought in. The fix was seeing the risk ninety days out, when you can still act on it.
Forwarded to you? I am Heath. I build go-to-market systems and put AI to work in sales, the right way, then I write down exactly what I built, what broke, and what it moved. One story per week, receipts only. This one is about the churn you could have seen coming ninety days out.
| 1 | Renewals were reactive. A CSM found out an account was leaving about when the account told them. By then it is not a save, it is an exit interview. |
| 2 | Healthy on paper was the trap. 90% of the renewal book read fine right up until it did not. The warnings were in the data, unread. |
| 3 | The fix was the autopsy, then the score. Learn what actually predicts a loss, score the accounts turning cold, and start the renewal conversation while there is still time to change the outcome. |
| 4 | 66% of renewals got zero activity before. After the build, every renewal gets a logged touch 90 days out, auto-flagged if not. Reactive became a motion. |
The reflex: watch the renewal date and hope the book holds
Every team with a renewal book reaches for the same move. There is a date on every account. So you sort by renewal date, you watch the next thirty days, and you assume the accounts that have not complained are fine. Health is quiet, so quiet must be health.
I did exactly that. The year I took over a book, churn and downgrades were larger than everything the new-business engine brought in. We were filling a bucket with a hole in the bottom and congratulating ourselves on how fast we poured. And the renewals themselves were handled reactively. A CSM found out an account was leaving roughly when the account told them. By then the conversation is not a save. It is an exit interview. You cannot save a renewal in the last thirty days.
The block was never effort. It was timing. There was no early-warning signal, so a cooling account looked identical to a healthy one until the cancellation email arrived. The data that would have told us was sitting in the product and the calls, unread. The book looked 90% healthy right up until it did not.
The CSM feels this as "the account told me they are leaving." The AE feels the same shape as "the deal went dark and then it was lost." The marketer feels it as "engagement in that segment slid for weeks before the unsubscribes." Same reflex every time: a slow signal ignored until it becomes a fast, final one.
The reframe: renewal is a motion, not a date
A renewal date is not a moment to work. It is the deadline on a motion that should have started ninety days earlier. The move is not to watch the date. It is to take the calm, useless "the book looks healthy" and drill it down to the one specific account a CSM can still save.
You cannot save a renewal in the last thirty days. You can save it ninety days out, when the signals first turn.
Same move, other seats. An AE stares at "the deal is still open" and drills it down: which stage has not moved in weeks, who stopped replying, is the champion still there. A marketer stares at "engagement is fine" and drills it down: which segment, which cadence, whose opens have quietly halved. The calm number is never the thing you act on. The cooling account underneath it is.
How the best teams frame it
I am not the first person to argue that retention is a leading-indicator problem. The operators and analysts who have built this motion at scale mostly agree on where the leverage is, and it is not the renewal date. It is the early signal and the play that starts before it.
SOURCE
Planhat, "The Ultimate Guide to Measuring Customer Success: Metrics, Churn, and Retention"
What it argues. Churn and net revenue retention are lagging indicators. They tell you what already happened. The forward-looking work lives in leading indicators like usage trends and health scores, which drop into a risk band early enough for a CSM to intervene before a cancellation request ever arrives.
My take. Agree, and this is the part teams skip. They report NRR to the board every month and call it retention work. NRR is the scoreboard. The leading indicator is the game, and it is the only thing you can still change.
SOURCE
What it argues. Drawing on SaaS Capital data, the piece shows a direct link between improvements in net revenue retention and accelerated growth, a relationship they call parabolic. Gross retention holds around 90% and tends to decline as companies mature, so protecting the existing base becomes harder, not easier, over time.
My take. Extend. If retention gets harder as you scale, then a manual, reactive save motion is a bet you lose slowly. The only way the math holds is to catch risk earlier every year, which means the warning system has to do the watching, not the CSM's memory.
SOURCE
Gainsight, "Customer health scores: metrics, models and tools"
What it argues. A customer health score blends usage, sentiment, support, and outcome data into one predictive read so teams can act on risk before a renewal. The value is not the number itself. It is that the score routes an account into a play while there is still time.
My take. Agree hard, with one correction. You cannot build the score first. The reason my early flags were noise is that I had not run the autopsy yet. Learn what actually predicts a loss, then the health score has something true to measure.
Even a curator has to concede when the field agrees: nobody who has actually protected a book thinks the renewal date is where the work happens. The early signal and the play is the whole motion.
The method: Solve, Stack, Split
SOLVE THE CRUX
What is the real problem, framed as work and not a headcount?
The problem is not "we need a health score." It is "we find out too late." So the first work is the autopsy: run churn, closed-won, and closed-lost analyses and learn what actually predicted a loss. Not opinions, patterns. You cannot build an early-warning system until you know what the early warnings are. For the CSM that is the behavior that precedes a cancellation, for the AE the one that precedes a deal going dark, for the marketer the one that precedes a segment disengaging.
STACK THE CONTEXT
What tech and signals turn a raw pattern into a book a human can act on?
Not a shopping trip. Gong runs the autopsies off the calls, patterns not opinions about what predicted a loss. Amplitude carries the churn-risk signal, flagging accounts turning cold while there is still time. Salesforce holds the book-of-business view: every account, its health, its renewal date, and its risk in one place, not a spreadsheet updated from memory. Deepline runs the renewal-proactivity play so the work starts months ahead, in priority order.
SPLIT · CUT THE DRAG
What low-judgment work goes to the system?
Run the autopsies, read the signal, score the risk, rank the book, flag the accounts with no logged touch. Every account, every week, with no human in the loop until there is a reason to be. This is the part that never scaled when a person did it by hand, and the part AI does perfectly because it is mechanical.
SPLIT · KEEP THE JUDGMENT
What stays human?
The save itself. The system says this account, this risk, this reason, ninety days out. The CSM decides what the risk means for this relationship and owns the conversation that keeps them. One owner on the book, a weekly review of what turned around, and the saves feed back into what the score watches for.
The workflow: the board that runs it
Solve, Stack, Split is the shape. Here is the actual board, lane by lane: what the agents run, what stays human, and the tool at each step. Once it is wired, the whole thing runs every week without anyone kicking it off.
The receipt
At a company I was at, a growth-stage SaaS where the renewal book had grown larger than the new-business number, this is the exact build I ran. Autopsy first, then the book-of-business view, then the score, then the play on top. What the diagnosis found was not a people problem. It was a blindness problem.
The fix was not asking CSMs to try harder. It was a 90-day SLA: every renewal now gets a logged touch ninety days out, and the account auto-flags if it does not. The autopsy told us what to watch, the score told us where to look, and the SLA made sure the motion actually ran on the accounts that were cooling. Reactive became proactive because the system did the watching, not someone's memory.
My receipts here are retention-side. Drop your own workflow in. An AE turns the same build into a pipeline-slippage system: the behavior that predicts a deal going dark, flagged to the rep weeks before it is marked lost. A growth marketer turns it into a lifecycle trigger: the engagement drop that predicts a segment disengaging, routed to a re-activation campaign instead of a save call.
WHAT I LEARNED
1. If your CSMs find out about churn from the customer, you do not have a retention motion. You have a notification system.
2. Healthy on paper is the trap. 90% of the book reading fine is where the churn hides, not proof that it is not there.
3. Build the autopsy before the score. You cannot flag an early warning until you know what the warnings actually are.
4. The save is a 90-day motion, not a 30-day scramble. In the last month, the conversation is already an exit interview.
Run this one this week
Do not build the whole system. Pull your renewals landing in the next ninety days. Sort out the ones with declining usage or no logged touch. Add one line of context to each, which team went quiet, what changed, why now, and put a real task on the CSM to open the conversation. Watch what turns around. That is the motion in miniature, and it tells you whether the full build is worth it before you spend a quarter on it.
Two builds that sit next to this one:
- The Product Channel · "Raw usage without context is a colder cold list. Context is the whole channel." Same product signal, pointed at new business instead of churn.
- Scoring the TAM · "You have to score the whole market on one rubric, so a rep can look at any account and know its tier and its next move." The same scoring discipline, pointed at who to acquire instead of who to keep.
This is one build from the Build Log. Every week I take one sales or revenue problem, run it through the loop, and show the receipts. If someone forwarded this, the subscribe button is right below. Keep building. Heath.
You bought the signal. You never built the motion.
Everyone can capture intent now. The pipeline leaks in the gap between knowing and acting. What I got wrong, and what I am asking Adam Robinson on air.
The Seventh Analyst: The Agent That Reads the Other Six
Anyone can stand up six AI analysts. The one that makes the system compound is the seventh, the meta-analyst that reads the others' audit trails and turns every miss into a rule the next run enforces.
The Governance File: How a Pipeline Stops Repeating Its Worst Week
When you automate, your failures go silent. This build turned every past break into an enforced gate, so a fixed bug stays fixed and nothing broken ships confidently again.