FY2027 headcount and opex plan

Holding back a third of the R&D hires costs little ARR this year because only the roadmap module depends on them in the model, and it ships one month later. The real cost is the later roadmap items, which the model does not count; that is why the memo holds a third and not all of them. Hiring the account executives a quarter earlier adds ARR that, at a 78% gross margin, earns back its extra cost in about 11 months.

Delaying every hire by a quarter would lift FY2027 operating income by $3.52m, but the hires still arrive, so the saving happens once, while the $1.72m of December ARR it gives up is lost every year after. With the two timing changes the plan's operating margin is −4.8%, already at the benchmark median of −5.0%, without giving up ARR.

Profit is most sensitive to what each hire costs, not to how many there are: each 1% on pay is about $0.84m of operating income. If first-quarter offers run more than about 0.22% above the plan's rates, the recommended plan's margin falls below the benchmark median. Higher attrition costs profit as well as ARR: every leaver is replaced, so each one brings recruiting and equipment cost, the January–March refills of last year's leavers start on full pay, and new account executives take six months to ramp.

Level 1 is pegged to the occupation's San Francisco median, level 2 to its 75th percentile, level 3 to the median of the matching manager occupation. Account executives (level 2 in Sales) use the 75th percentile of technical sales representatives; commission comes on top. Pay is raised from May 2025 to January 2027 at the Employment Cost Index for Information-industry wages, and again by the same rate in April.

The rule: sales and marketing and R&D both rise with growth, so their targets are the medians of the benchmark's fastest-growing third of companies, the third whose median growth is nearest the plan's 30% ARR growth. G&A, gross margin and operating margin use the benchmark's own medians.

Benchmark (projects/opex-benchmark/out/targets.json): SEC XBRL data (data.sec.gov companyfacts), latest 10-K fiscal year ending Jan 2025 to Jan 2026, 124 US-listed SIC 7372 software companies with revenue of $100M or more; GAAP, so every line includes stock-based compensation. S&M 37.3%: the median of the benchmark's growth third 3 (median revenue growth 25.8%, n=38), the third nearest the plan's 30% ARR growth; the all-growth median is 33.2%; R&D 26.7%: the median of growth third 3 (n=41), by the same rule; the all-growth median is 21.4%; G&A 22.4% ($100M-$300M, n=16); gross margin 72.4%, operating margin -5.0% and stock comp 11.8% are the under-$1B medians (n=70). Ranges are the 25th to 75th percentile of all growth rates. Revenue per head stays on this project's nine-company peer median.

R&D is the one line over its target, mostly because the plan pays every engineer at San Francisco rates. G&A sits below the benchmark's 25th percentile; the benchmark notes that smaller companies run higher G&A, so that line is more likely under-budgeted than over.

Revenue per head is not in the benchmark, so opening ARR is sized on nine public software companies' revenue and the employee counts their 10-K filings state: the plan's $327k per average head sits near their $336k median.

Tillbrook is made up, so nothing here describes a real company's plan. The pay anchor is the all-industry wage for each occupation in the San Francisco area, which a software company may pay above; that is why pay level leads the tornado. Opening account executives are treated as fully ramped, commissions are expensed when booked, and the roadmap module is the only revenue that R&D hires drive inside the year.

Next: add a location mix for R&D, and run the month-end variance against this plan.

Tillbrook Software is a made-up company. BLS and SEC data are public; nothing here is investment advice.