Revenue per employee: useful signal, wrong yardstick for lean AI firms.
Revenue per employee shows how much output a company gets from its people, much like GDP per capita for a country. That's why AI-native firms with small teams stand out on it. But for a very lean firm, the number becomes a statistical outlier rather than a useful signal, because most cost goes to compute and verification, not payroll. Our AMEF paper proposes a complementary measure: the Protocol Efficiency Ratio.
Why it's a useful starting point
Total revenue says little about efficiency. Divide it by the people producing it and you see how well output scales. A small number of AI-native companies have drawn attention because their publicly reported revenue per employee runs far ahead of similarly sized traditional businesses. (These are public market examples, not Secular Labs clients.)
Where it breaks down
| Metric | Assumes | Problem for a lean, verification-gated firm |
|---|---|---|
| Revenue per employee | Payroll is the main cost | A team of four makes it an outlier, not a signal |
| Burn multiple | Most cash goes to payroll | Most cost is compute and verification infrastructure |
| Protocol Efficiency Ratio (AMEF) | Cost includes compute, verification and the core team | Measures new recurring revenue against what it actually costs to run the business |
AMEF proposes this ratio alongside, not instead of, conventional measures. The paper's worked examples are stylized illustrations, not claims about any specific company.
What we measure in practice
In an engagement we track revenue per core employee together with throughput, cycle time, automation coverage, exception rate, verification coverage and cost per transaction. That way one number never tells the whole story.
Frequently asked
What's a good revenue per employee?
It depends heavily on industry and business model, which is why we compare a business against its own baseline, not a headline benchmark.
Is the 3–100x range on your site a revenue-per-employee figure?
It's a modeled range of revenue per core employee under stated assumptions, not a measured client result.
Who should use the Protocol Efficiency Ratio?
Investors and boards assessing lean AI-native firms. See AI systems due diligence.
See what this looks like in one of your workflows.
Related: AI-native operating model · AI systems due diligence · AMEF
More explainers: The coordination tax · The verification premium · Agent orchestration vs RPA · Verifying AI agent decisions