Why last year's shares are the wrong starting point for headcount planning
The default in annual planning is to take last year's headcount per product line and add a percentage to each. It feels fair and it funds the past. A mature line keeps its share whether or not it needs it, and a growing line stays starved because it started small.
Allocating by revenue share fails differently. Revenue tells you where the business is today, not where the next return will come from. The line that pays for everything may need fewer engineers than it has, if much of its work is running systems that could cost less to run.
So start with measurement. For each line, find what share of last quarter's engineering time went to keeping the lights on: on-call, incidents, upgrades, security patching and operational requests. The capacity split page covers how a team measures this without timesheets. At executive level, the job is to aggregate that measurement across lines and treat it as the first call on the budget.
Investment postures
Each product line gets one posture, agreed with its business owner, and the posture drives the allocation.
A sustain line protects revenue and margin. It gets its run cost plus targeted work to reduce that run cost, and little new feature capacity. A grow line has evidence of demand and gets most of the new investable capacity. A bet line is unproven. It is funded in tranches, with leading indicators and a dated gate that decides whether the next tranche is released. A platform line is funded for the run cost it removes from other teams, which is why its case is made in other teams' numbers.
The worked allocation
The figures are illustrative and come from the assumptions stated. They are not benchmarks for any real organisation.
Assume 200 engineers today, funding for 30 new roles, and four lines. Run shares are measured, as above.
| Product line | Engineers now | Measured run share | Run FTE | Investable FTE | Posture | New roles | Year-end | Effective FTE-years added in year one |
|---|
| Core product | 100 | 45% | 45 | 55 | Sustain, with debt work to cut run share | 0 | 100 | 0 |
| Growth product | 60 | 25% | 15 | 45 | Grow | +16 | 76 | 8.7 |
| New market bet | 10 | 10% | 1 | 9 | Bet, two tranches | +4, then +4 if the gate is met | 18 | 3.3 |
| Platform | 30 | 50% | 15 | 15 | Platform | +6 | 36 | 3.3 |
| Total | 200 | 38% | 76 | 124 | | +30 | 230 | 15.3 |
The effective-capacity column uses two assumptions. A committed role takes four months to fill, so the hire starts in month five. They then spend three months at half productivity before working at full pace. That gives (3 × 0.5 + 5) / 12, about 0.54 FTE-years in year one. A conditional role released at a gate at the end of the first quarter, with the same four-month fill, starts in month eight, giving (3 × 0.5 + 2) / 12, about 0.29.
The core product gets no new roles, and its leader will push back. The answer is in the run-share column. If the debt work cuts core's run share from 45% to 35%, it frees about 10 engineers' worth of capacity within the line, with no new hiring. That is a forecast, not a gift. Count it only once the measured share has fallen.
Headcount is not capacity
This is the point that separates strong headcount planning from adequate. Thirty approved roles are not thirty engineers' worth of output this year.
Add attrition. Assume 10% annual attrition on 200 engineers, so 20 leavers. Each backfill loses four months of vacancy plus three months at half productivity, about 0.46 FTE-years per leaver, or about 9.2 FTE-years in total. Set against the 15.3 FTE-years the new roles add, the net gain in year one is about 6.1 FTE-years.
That figure still flatters the plan. It leaves out the time existing engineers spend interviewing and onboarding, which falls hardest on the teams that are growing. The practical consequences are threefold. Commit roadmap outcomes against effective capacity, not approved headcount. Hire early in the year for work that must land this year. Treat retention as a capacity lever, because in this model each departure avoided is worth almost half an engineer-year.
Showing the trade-offs to finance
Present the plan with its lower-budget version already prepared, so a cut becomes a choice rather than a scramble. Here, if finance funds 20 roles instead of 30, you might first drop the bet's conditional tranche of four, which was gated anyway. Next, take six roles from growth and name the outcome that slips, such as a second-market launch moving by a quarter. Say what you will not cut. Platform roles that reduce run cost elsewhere and the core debt work protect next year's capacity, and cutting them makes the following plan worse.
Express the whole plan in categories finance recognises: run, grow and bet, as both headcount and cost. If finance frames engineering as a share of revenue, bring your own company's history rather than an industry figure, because external ratios rarely compare like with like.
Revisiting the split during the year
Headcount planning done once a year and never revisited becomes last year's shares next year. Hold a quarterly reallocation review against measured run share and each line's outcomes. Release or withhold bet tranches at their dated gates. Move capacity between lines deliberately, through team changes rather than by moving individuals one at a time. The headcount case page covers the manager-level request that feeds this.
Company shape changes the emphasis. A startup plans against runway and may revisit monthly. A listed company plans against guidance and values predictability. A private-equity-owned company may plan against a margin target, which favours sustain postures. A captive centre's headcount is often set by its parent, so its leader's lever is the mix of work, not the total.
Fund measured run cost first, give every line a posture, and plan in effective capacity. In this model, thirty new roles add about six engineer-years of net capacity in year one once hiring lag and attrition are counted.
© 2026 Preptima. Originally published at preptima.com (opens in a new tab).