It did exactly what it was designed to do. That is the part that kept it alive for eighteen months, and it is why I now distrust an experiment that is working.
We replaced static territories with a dynamic book, reallocated on performance. Accounts had windows to be quoted and closed. Miss them and the account went back in the pool. Eligibility for more was gated on attainment, idle accounts were reassigned, and a book could be cut for a quarter below the line.
Net-new acquisition climbed. The pressure worked exactly as designed, reps chased new logos, and the headline metric went up and stayed up.
“New logos are up. The model is working.”
“Give it another quarter to bed in.”
“The reps who are complaining are the ones being asked to work harder.”
“Revenue will follow the accounts.”
Revenue did not follow. It went flat in the partner channel and stayed there, and by the time that showed up the damage was already a year old.
The experiment did not fail. It succeeded, at the tier of the market it was aimed at, and that tier was not where the revenue was.
At the top, the largest accounts could not be opened at all without someone local who could show up in person. Trust at that end is built face to face, and no amount of activity from a remote seller substitutes for it. If trust could not be built, the account could not be opened.
And the reps were not being difficult. They stopped investing in relationships they were about to lose, which is the system working as designed rather than people behaving badly.
They will do the arithmetic, and they will be right. An allocation system that punishes long-horizon investment produces short-horizon behaviour every single time, and blaming the seller for that is the mistake.
The fix for a rational response to bad incentives is never coaching. It is the incentive.
We were winning the middle and losing both ends, for opposite reasons. That is much harder to see than a single failure, because the aggregate looked acceptable the whole time.
| Tier | What it needed | What it got | Result |
|---|---|---|---|
| Top | Depth. In person, continuous, trust built face to face | Remote, high activity, rotating | Could not convert |
| Middle | Neither extreme | A uniform coverage model | Performed well |
| Bottom | Frequency. Constant nurture to convert at all | Whatever attention was left over | Revenue fell away |
The ends are where uniform models fail, and they fail in opposite directions, which is why one fix cannot address both.
The bottom is the half most people miss. Low-frequency accounts need constant nurturing to convert at all, and when they moved to account management they simply stopped selling. Not resistance. Account managers spent their time on the more active accounts, which was the right call for them and fatal for that tier.
Which makes the bottom of the market a marketing problem wearing a sales-coverage costume. Asking a person to nurture it is asking them to act against their own return. Give it to a system instead.
What kept it running was that the headline metric was healthy. New account acquisition was climbing and highly visible. The accounts being won were small and bought once or twice a year, rather than the high-frequency accounts that actually carry a territory.
New logos went up while revenue went flat. The metric that looked best was measuring the wrong thing, and that is a pattern rather than an incident.
Territories came back, with a local person, in-person touchpoints and a materially better product. Both were required. Neither alone opened the top of the market.
But the structural answer is the pairing, and that is the transferable part: you do not have to choose between coverage and continuity. Staff for both, and pay them as one unit.
| Role | Owns | Because |
|---|---|---|
| Inside seller | The tier that can be won without presence | Coverage economics. Reach and volume |
| Outside seller | The largest targets, anchored, never reallocated | Trust economics. Showing up is the product |
Two design decisions make it work and both are easy to get wrong.
The inside role is a closer, not a qualifier. Make it an account executive with its own book. Make it a hand-off role and you have built a funnel rather than a team, and accounts it could have closed get escalated for no reason.
One shared growth number across both. Separate quotas on the same geography turn two colleagues into two claimants arguing over which deals count. A single shared number makes the hand-off free, so the account goes wherever it will actually be won.
Split your accounts into three tiers by what they need rather than by what they spend. If the top needs presence you are not giving it, and the bottom needs a frequency no book will sustain, then a single model is quietly failing at both ends while the average looks fine. The model is free and it starts with what your book is actually made of.
Run the model, free No account. Your numbers stay in the link.