Acquisition was strong. Revenue was flat. Nobody in the room could explain the gap, and every explanation we tried was about effort or market conditions. It was neither. It was subtraction.
Accounts were scored one, two and three. Systems kept activity high on the ones and twos. The threes were still called and emailed, just less often. Nobody decided to stop serving them. They simply came last, every week, until they came not at all.
Meanwhile new-company acquisition was running at a good clip and everybody could see it. It was on every dashboard. It was the number we reported when someone asked how growth was going.
And revenue did not move. For a long time.
“They only do two or three deals a year. If they’re at zero they must not be focused on us.”
“We’re winning plenty of new accounts. The engine’s working.”
“The reps are spending time where the volume is, which is what we pay them to do.”
“That account has been flat for years. It is what it is.”
Every one of those is reasonable. The first one is the expensive one, and it contains the whole error: they did two or three last year. The baseline was evidence, and we read it as a ceiling.
You lose a mature account doing two or three deals a year. You replace it with a new one that does one in its first year, because trust and volume take time to build.
On the acquisition dashboard those two events cancel out perfectly. In revenue they do not cancel at all: you are down the difference, every year, on every swap.
And because no single account was large, nothing ever looked like an event. It was a slow death rather than a visible one, which is exactly why it went unnamed for as long as it did.
I have watched accounts move from a three to a two to a one when they were given the right resources. Momentum arrives, and then they start investing their own resources to keep it going, which is the point at which the thing compounds without you.
Which means the score was never a property of the account.
Treat the score as fixed and you make it fixed.
The flywheel runs in both directions, and the downward one is the expensive half, because it looks like good judgement the whole way down. Withhold resources, volume decays, the score confirms itself, so you withhold more. Every step is defensible. The slope never is.
It also inverts the return on acquisition: a three grown into a one is worth more than a new account that arrives as a three, and it needs no ramp, no trust-building and no onboarding.
This is the uncomfortable part, and it is the one I would want a leader to sit with.
A rep who neglects a low-scoring account to work a high one is doing exactly what you pay them to do, and they are right. There is no coaching conversation that fixes this, because there is nothing wrong with the behaviour. The behaviour is the output of a design.
If the comp plan pays for volume, the bottom of the book dies on schedule.
1 · The comp plan was realigned so working the bottom of the book is paid for rather than penalised.
2 · Marketing automation took the bottom tier: a nurture cadence at a frequency no human book will ever sustain.
3 · Self-service. Quoting, collateral, technical answers and ordering, available without asking anyone. This is the one that mattered, and it is not a cheaper substitute for coverage.
The bottom of the market is served by systems or not at all. Human attention will always route to the top, and it is right to. Asking an account manager to nurture low-frequency accounts is asking them to act against their own return, and they will do the arithmetic, and they will be correct.
Which makes the bottom of the market a marketing problem wearing a sales-coverage costume, and it had been one the whole time.
The test is simple and almost nobody runs it: over the last two years, how many accounts entered your book, how many quietly stopped ordering, and what is the difference worth once you account for ramp? If those first two numbers are close, acquisition is funding churn and calling it growth.
Run the model, free No account. Your numbers stay in the link.