The Unhired Index
The Unhired Index counts the missing workers of big software: the sum of reported employees at nine leading software companies, against what each company's own pre-2020 revenue-headcount relationship predicts for the revenue it actually earned. The two lines track each other for a decade, cross in 2023, and then split: by 2025 the companies employ roughly 185,000 fewer people than their own histories imply. The companies are chosen by fixed criteria listed in the method — not picked for the story — and the index updates as each new fiscal year is filed.
What does it show?
Had these 9 companies kept hiring the way their own pre-2020 filings predict, they would employ about 185,000 more people than they did in 2025 — a gap that opened in 2023 and has widened every year since.
Methodology
For each included company, OLS of log fiscal year-end headcount on log annual revenue over its pre-2020 filings, held fixed and applied to later observed revenue; the chart sums reported and implied headcount across companies. A company is included when four things hold: it is a software business; it filed at least 7 pre-2020 fiscal years of revenue and headcount; its own pre-2020 history actually predicts its hiring (baseline R² of at least 0.8); and its headcount is readable from filings through 2025. The same criteria were applied to all fourteen companies attempted, and the payload carries the full verdict table. Each exclusion has a stated reason: Amazon (fulfilment labor; including it would roughly double the gap), Intuit and Autodesk (their own pre-2020 history does not predict them, R² 0.11 and 0.27), ServiceNow and Workday (headcount not reliably extractable from recent filings). Robustness, actually run: raising the R² bar to 0.90 leaves 6 companies and a gap of 136,000; requiring 9 pre-2020 years leaves 8 and 181,000; the original five-company set gives 153,000. The pre-2020 baselines track reported headcount within a few percent for a decade, and the small 2020-2021 over-hire visible on the chart was repaid by 2023 — the gap that follows is not its unwinding. Fiscal years as filed; M&A moves employees without matching revenue timing (Synopsys's 2025 includes the Ansys acquisition, which shrinks the measured gap). Descriptive, not causal.