Hiring's response to growth, before and after
One fitted slope per company per era: how much hiring a percent of revenue growth induced in 2013–2019, and how much it induces in 2021 onward. Meta fell from 0.89 to zero and Amazon to slightly negative. Hollow markers carry the deeper finding — in the later era revenue no longer explains headcount at all for most of these companies, rather than merely explaining less of it.
What does it show?
Fitted before and after Covid, the elasticity of hiring to revenue collapsed at every software company: Meta from 0.89 to 0.00, Amazon from 1.48 to −0.02 — and for most of them revenue no longer explains headcount at all.
Methodology
One OLS fit of log fiscal year-end headcount on log annual revenue per company per era: 2013–2019 and 2021–latest, with 2020 excluded as pandemic transition. Hollow markers flag fits with R² below 0.5 — in the later era that is most of the software companies, which is itself the finding: revenue stopped explaining headcount rather than merely buying less of it. Post-era fits rest on five or six annual observations, so the direction is robust and the second decimal is not. Sensitivity, actually run: starting the pre-era at 2012, and starting the post-era at 2022, leave every software company's decline in place. Data sources and M&A caveats as in the companion path chart. Descriptive, not causal.