AI Is Breaking Revenue Per Employee as a Standalone Metric
Revenue per employee flatters lean, AI-heavy agencies. Count freelancers and owner hours, then check contribution margin before you trust it.
By Craig S. Cody, CPA, Certified Tax Coach.
Revenue per employee is still a useful number for an agency. It just can't stand on its own anymore. Once AI and freelancers do a big share of the work, the employee count stops counting the humans who actually deliver it, and the number flatters you.
When I look at an agency's numbers and see a high revenue per FTE, I ask who's actually doing the work. When the answer is contractors, the number isn't useful unless you count contractors and employees together.
So keep the metric, but count the whole delivery model before you trust it. Here's why AI breaks the number, what an honest headcount looks like, and the one figure that tells you whether a lean model is really working.
On October 13, Digiday profiled BaD Mktg, a Toronto agency run by two principals, Ian Buck and Yotam Dor. They work with freelancers, use several AI models to run most of their back office, and built their own tool, Goodie, to handle priorities, payments, vendors and admin.
Buck says the setup lets them function like a shop with a dozen staffers. In three years the agency has brought in more than $2 million in revenue, and it's cut concepting from about six weeks to six days or less.
That's a smart operating model, and I'm not here to knock it. Three details from the interview are worth noticing, though, because they're exactly why the usual efficiency metric breaks:
Put those together and you have an agency where the biggest costs and the hardest working hours are the ones a revenue-per-employee figure can't see.
Revenue per employee is a fraction. Revenue on top, employees on the bottom. AI can genuinely grow the top. But it also changes what belongs on the bottom, and most agencies never update it.
When work moves off payroll, it doesn't disappear. It moves to three places:
Every one of those makes the denominator smaller without making the business any cheaper to run. The agency looks more productive per person because fewer people are being counted, not because each dollar of work costs less.
One more distinction, because the brief I started from uses revenue and AGI interchangeably. For an agency that buys media or printing for clients, revenue per employee is the weaker of the two. Pass-through spend inflates it before AI does anything at all. I covered that in Your Agency's Margins Look Fine. The Number Underneath Them Is Wrong. Use agency gross income (AGI) whenever you have it.
Here's an illustration. These are two made-up agencies, not BaD Mktg or any client, and to keep the math clean neither one has pass-through costs, so revenue equals AGI.
Both bring in $1.2 million a year.
Agency A has eight employees. The owner mostly sells and manages. Delivery payroll, fully loaded, is $600,000. It spends $40,000 on freelancers and $20,000 on software.
Agency B has three people on payroll: two owners and one employee. The owners pay themselves $60,000 each in salary and take the rest as distributions, which is common. The employee costs $90,000 loaded. B spends $300,000 on freelancers, $36,000 on AI and software, and $14,000 a year keeping its custom tool running.
On the usual metric, it isn't close:
| Agency A | Agency B | |
|---|---|---|
| Employees | 8 | 3 |
| Revenue per employee | $150,000 | $400,000 |
B looks almost three times as productive. Now count the work, not the payroll. B's freelancers bill about 3,000 hours a year at around $100 an hour. That's roughly 1.7 full-time people at 1,800 hours each. A's freelancers add about 0.2.
| Agency A | Agency B | |
|---|---|---|
| Working headcount (employees plus freelancer FTEs) | 8.2 | 4.7 |
| Revenue per working FTE | about $146,000 | about $257,000 |
B still wins, and that part is real. AI and a sharp team let fewer humans produce more. Now look at what the work leaves behind after the delivery costs, which is the number that matters most.
| Agency A | Agency B, as booked | Agency B, owners priced honestly | |
|---|---|---|---|
| AGI | $1,200,000 | $1,200,000 | $1,200,000 |
| Delivery payroll or owner time | $600,000 | $210,000 | $370,000 |
| Freelancers | $40,000 | $300,000 | $300,000 |
| AI, software and tool upkeep | $20,000 | $50,000 | $50,000 |
| Contribution margin | $540,000 (45.0%) | $640,000 (53.3%) | $480,000 (40.0%) |
The middle column is what B's P&L shows. It counts the owners at their $120,000 of salary. But each owner spends about 1,400 hours a year delivering client work. If you had to hire a senior person to cover those 2,800 hours at a loaded $100 an hour, it would cost $280,000. Swap that in for the $120,000 of owner salary and B's margin drops from 53% to 40%.
Nobody in that example did anything wrong. B might be a great business. It might pay its owners very well through distributions. But it's leaning on owner hours that the P&L shows at a discount, and that matters the day an owner wants to step back, hire a replacement, or sell.
Change the inputs and B can come out ahead. Some lean shops will. The point is that revenue per employee can't tell you which kind of agency you're looking at. You only find out by counting.
Don't throw the metric out. Put four checks next to it, all aimed at the same question: did the work actually get cheaper, or did it just move?
Then one number decides it.
Contribution margin after variable delivery costs. Take AGI and subtract everything it cost to deliver the work: delivery payroll, freelancers, AI and software used on client work, and owner delivery hours priced at what it would cost to replace them. What's left, divided by AGI, is the truthful answer.
A smaller team isn't automatically a better business. A faster workflow isn't automatically a more profitable service. And fewer employees don't mean more operating leverage if the work simply moved to contractors, founders and software vendors.
If that margin holds up after the full count, your lean model is real and you've earned the headline number. If it drops, you've learned where your efficiency was coming from before a buyer or a burned-out partner learns it for you.
It isn't about how to price AI work or whether AI pays off at all. Those are separate questions, and I've laid them out in order in AI and Agency Profitability: Ten Questions, in the Order That Matters.
It also isn't an argument against freelancers. A flexible specialist bench is a perfectly good way to run an agency. Just make sure the people you pay as contractors really are contractors. Your file decides that, not the title.
It's for marketing and PR agency owners who've gone lean on purpose, who lean on AI and freelancers, and who quote revenue per employee to their partners, their banker or a buyer.
It's not for a traditional agency with every delivery person on payroll and almost no freelance spend. Your revenue per employee is already close to honest. Just make sure you're dividing AGI, not gross revenue.
It's annual revenue divided by the number of employees. Agencies use it as a quick productivity check. For agencies that buy media or other pass-through costs, AGI per employee is more accurate, because pass-through spend inflates revenue.
Yes, but not by itself. AI and freelancers shift work away from payroll, so a smaller employee count can make the agency look more productive than it is. Pair it with a working headcount and contribution margin.
Convert their hours into full-time equivalents. Divide total freelancer hours for the year by the hours a full-time person works, around 1,800, and add the result to your employee count.
It's AGI minus the variable costs of delivering the work: delivery payroll, freelancers, AI and software used on client work, and owner delivery time priced at replacement cost. Divide it by AGI to get a percentage you can compare over time.
Track a rough estimate of the hours you spend delivering client work, then price those hours at what it would cost to hire someone to do them. If that's far above your salary, your margin is being propped up by your own time.
AI can grow the top of the fraction. It can also quietly shrink the bottom by moving work to people and vendors you don't count. Measure the whole delivery model before you celebrate the efficiency.
I also wrote a book for agency owners. You can request a free copy at the link below.
Nobody's looking out for your money but you. Let's talk.
Craig S. Cody is a CPA, Certified Tax Coach, and retired NYPD Lieutenant. His firm works with more than 70 marketing and advertising agency owners every month, helping them keep more of what they make through proactive tax planning.
This article is general education, not advice for your specific situation. Confirm your own facts with your advisor before acting.
Revenue per employee flatters lean, AI-heavy agencies. Count freelancers and owner hours, then check contribution margin before you trust it.
Agencies that narrow their services grow faster, but only when they cut the right line. A CPA's four-question test before you shut one down.
Agencies that track profit project by project are the most profitable. Here's the one-page project P&L your delivery lead should own.