The operating stance
Scale performance, not headcount
Definition
Scale performance, not headcount is the operating stance of Atomic Scaling: treat revenue per employee as the number to grow and headcount as a cost of last resort, so that adding capacity does not automatically worsen the economics.
The problem it names: Most companies buy capacity with people, so every increment of growth makes the business less efficient than it was.
Most companies buy capacity with people. It works, briefly, and it is why so many businesses are less efficient at $50M than they were at $5M. There is a different number to run the company on — and an industry that has been proving it for twenty years.
The arithmetic nobody runs
Revenue per employee is total revenue divided by headcount. It is the least fashionable number on the dashboard and the one that decides whether growth is helping you.
Top 25 SaaS companies
Average revenue per employee per year
Supercell
10× the SaaS average — 250M daily active users on 420 people
Valve
170,000 customers served per employee per day; most profitable company per employee in the US in 2012
Figures as published in Atomic Scaling by Ludovic Bodin.
Ten times the revenue per head is not a rounding difference — it is a different kind of company. It is also why Valve can pay an artist several times what a film studio pays: at $4M per employee, paying $1M still nets $3M. High revenue per employee is not austerity. It is what makes generosity affordable.
AI made this the default, not the exception
What gaming proved with product design, AI-native companies are now proving with agents — growing roughly four times faster with seven to eight times fewer employees per dollar of revenue.
Cursor
$500M ARR in 24 months · 60 people
SurgeAI
$1.2B revenue · 130 people · zero outside funding
Midjourney
$500M ARR · 107 people
Lovable
$0 to $100M ARR in 8 months · 45 people
Revenue per employee derived from figures reported in the 3.2.1 newsletter. Company numbers move; the pattern is the point.
Four moves that shift the ratio
Measure the ratio, not the total
Revenue ÷ employees, tracked monthly, on the same slide as revenue. If total revenue is rising and the ratio is flat, you are scaling effort rather than impact — and you will find out expensively, two years from now.
Make every hire pass EPIC
Employees, Profit, Impact, Cost. A role has to improve all four at once. The question is not “can we afford this person?” but “does this person raise users served per employee?” If the honest answer is no, it is headcount, not leverage.
Fix the process before you automate it
QDSAA — Question, Delete, Simplify, Accelerate, Automate, in that order. Automating a bad process entrenches it and makes it harder to remove. Most of the leverage is in Delete, and almost nobody starts there.
Let agents operate, keep humans governing
The AI-native version of this is a Human-Assisted Autonomous Organization: agents as default operators, humans holding judgment, taste and override. That is what makes the ratio move by a multiple rather than a percentage.
It works outside of software
The objection is always that this only applies to games and SaaS. These did not:
- Gigaverse — $10M annualised revenue in six months, with a team of under ten.
- SuySing — $1B in revenue recovered post-COVID by pivoting online. A Philippine retail distributor.
- Mercedes-Benz — 5× online sales while reaching a younger audience. A $160B manufacturer.
Start here
What is your revenue per employee doing?
The Atomic Scaling Score rates your company across the six pillars of the 3P3R Method® — People, Prediction, Playbook, Reach, Retention and Revenue — and names the one keeping your ratio flat. Free, two minutes.