The headcount plan priced at last year's salaries
The plan called for fifteen hires. The comp math came off last summer's bands. By the time offers went out, the market had moved and nobody had told the board.
A plan I reviewed in November budgeted 15 hires for the next year across engineering, sales, and CS. The comp math came straight off the company's existing bands, benchmarked against data from the prior summer. The board approved it in December. It was already wrong.
By April they'd made four offers. Every candidate countered above band. Two accepted after the founder signed exceptions. Two walked. Recruiting put the market move since the plan at roughly a fifth, and nobody had told the board the remaining hires were now underfunded.
This is one of the most predictable variances in annual planning, and almost nobody builds it in. Comp data moves faster than planning cycles. Benchmark reports refresh every quarter, but the plan swallows the latest one and treats it as fixed for twelve months.
The lag between when the data was gathered and when offers go out can run two or three quarters. In a hot market that's a chasm. In a calm one it still bites.
The asymmetry makes it worse. New hires land at the new rate. The people you hired last year sit below it by the same gap. You either move retention comp up, which costs more than you booked, or you don't, and they find the disparity in a hallway conversation and start refreshing their profiles. Both roads cost more than the plan admitted.
The cost lands in three places. Burn runs over. Hiring stalls, because the roles won't fill at the planned number, so they stay open or the bar drops, which is its own expensive kind of hire. And internal parity cracks, as market-rate arrivals reset everyone's expectations. Ignore that and you get a wave of exits a couple of quarters later.
What works instead: price the plan against rolling data, not an annual snapshot. Benchmark comp quarterly. Add an explicit inflation factor for back-half hires, five to ten points in a calm market and more in a hot one. Re-run the plan against real offer-letter numbers each quarter so the delta reaches the board before the surprise does. For the finance and people leads, that's about half a day a quarter.
The deeper habit is to stop treating the annual plan as a commitment and start treating it as a starting line. It's wrong on approval day, the same way a forecast isn't a promise. The only questions are how wrong, in which direction, and what you do next. Teams that re-plan quarterly stay near the truth. Teams that defend the original number run out of cash defending it.