Note one
More reps, same revenue
The most common growth decision in software, and the one that most often fails — for a reason the plan could not see.
A board asks why growth has slowed. The answer that comes back is almost always coverage: not enough salespeople in front of not enough accounts. So the company hires. Twenty reps become thirty.
Twelve months later revenue is roughly flat and nobody can say exactly why.
What the plan modelled
Thirty reps, each carrying a quota, each closing at the historical rate. Revenue is reps times quota times attainment. Add reps, add revenue. The arithmetic is impeccable, and the arithmetic is the problem.
What actually happened
Each new rep needed onboarding, and onboarding came from the same sales engineers who support live deals. Support per deal thinned. Deal cycles lengthened. Attainment fell — and not only for the new reps. It fell for the existing ones too, who were now sharing a resource that had been quietly reallocated.
The company added fifty per cent more sellers and lost roughly the same share of productivity per seller. The first-order effect was real and positive. The second-order effect was real and negative. They landed in different quarters, and only one of them was in the model.
The thing the spreadsheet assumes
Revenue per rep is treated as a constant. It is not. It is an output of a system that the hiring decision itself changes. The moment you write it down as a fixed number in a cell, you have assumed away the entire mechanism that decides whether the plan works.
This is what a second-order effect looks like in practice, and it is why these decisions are so hard to learn from. By the time attainment drops, the hiring is a year old, three other things have changed, and the decision will never be blamed.
The better move
Not to stop hiring. To ask what is currently limiting revenue per rep before adding more reps. If the constraint really is coverage, hiring works and works quickly. If the constraint is onboarding capacity, or deal support, or fit in a segment, hiring makes it worse.
A straight-line model cannot tell you which one you have, because it has no mechanism in it to be wrong about. A model with feedback in it can.
One of five short notes on why decisions miss. Back to the others, or tell me where I'm wrong.