We had uncapped upside and would not move a base by ten thousand dollars. Both of those felt generous and disciplined at the same time, which is how you lose people you meant to keep.
People were hired at the lower end of the band. A year or two in, the good ones would ask for ten to fifteen thousand dollars more on base. We would often not entertain it. In the early years we did not increase base salaries consistently at all.
The upside was there. Commission was uncapped, the ceiling kept rising, and by any reasonable reading the opportunity was real.
They left anyway.
“The upside is uncapped. If they want more, they can go and earn it.”
“If we do it for one we have to do it for everyone.”
“They know we value them. We tell them constantly.”
“It is only ten thousand dollars. That cannot be the real reason.”
The last one was right and I drew the wrong conclusion from it. It was not the real reason. It was the only available evidence about the real reason.
They want to know they are valued, and a conversation about how much you value them is a good start. It does not mean very much unless you back it up with money.
Not because people are mercenary. Because a raise is the only signal in the system that cannot be confused with politeness.
Uncapped commission does not do this job either. It says the company will pay for more output. It says nothing about whether this person is going anywhere, and that is the question they are actually asking.
This is worth sitting with, because it is one mechanism wearing two costumes and most leaders have both problems at once.
| The message | Why it did not land | What fixed it | |
|---|---|---|---|
| Telling someone their performance is a problem | Said often, in one-to-ones | A verbal warning is indistinguishable from ordinary coaching | A published standard and a formal document |
| Telling someone you value them | Said often, sincerely | Verbal appreciation is indistinguishable from politeness | The money |
The channel carries everything else too. Which means "but I told them" is not a defence in either case.
Same model as a delayed exit, pointed the other way. A ramped performer holding 95 percent leaves. The seat sits empty for about two months, then a replacement ramps over six.
| Quota-months lost | |
|---|---|
| Vacancy two months, nobody in the seat | 1.90 |
| Ramp six months at the gap between a new hire and a ramped one | 3.00 |
| One departure | 4.90 |
A model, not measured actuals. At a worked monthly quota of $100,000, roughly $490,000.
Fifteen thousand dollars against roughly four hundred and ninety thousand only has to be about three percent likely to pay for itself. Even treating the raise as permanent and the departure as one-off, three years of it is still better than ten to one.
And notice it is the same calculation as keeping a low performer too long. Both are quota-months lost to a seat not held by a productive person. One is voluntary and one is not, and they are computed identically. That is one instrument serving two modules rather than two instruments.
A standard annual increase of about five thousand dollars for solid performers. The size is not the mechanism. The reliability is. A performing seller needs to know a raise is coming without having to ask for it, because asking is itself the thing that makes them start looking around.
And up to about fifteen thousand for the very best, on request, backed into from what replacing them would cost. Not a negotiation and not a reward. An arithmetic decision made against a number you can write down in advance.
A visible path to new titles and roles. People want to make those changes on their public profile, and that is portable proof that they are going somewhere.
It is free, and it gets ignored routinely because it does not appear on the compensation line. For someone whose real question is about momentum rather than money, it can answer a large part of the question on its own.
If someone good asked for a base increase in the last year and did not get one, they are already running the arithmetic you have not run. The model is free and it will put a number on what that seat is worth, which is the number the decision should be made against rather than the size of the ask.
Run the model, free No account. Your numbers stay in the link.