I work with enterprises and I work with genuinely small companies, and the small ones get worse advice.
They get told to track adoption, hours saved, tickets deflected. Numbers that go up without anything happening.
There is one number that actually compounds for a small business, and it is not on most dashboards.
Start with what hurts, not what's exciting
Before the metric, the sequence, because getting this backwards is the common failure.
Solve pain points first. Not the exciting project. The thing that hurts.
Every small business has both, and the exciting project always wins the meeting. It's the one that sounds like the future. The pain point is boring, somebody has been complaining about it for two years, and fixing it is the only thing that will get you a second AI budget.
So do the boring one. Then you have earned the right to do the interesting one, and you will do it with a team that believes you.
Then the number
Once the pain is handled, move an economic metric. For small companies, the one I care about is revenue per employee.
Going from $100,000 per employee to $200,000 or $300,000 is not an improvement. It is a change in what the business is.
The plain version, which is how I actually ask it:
Can I make more money with the same money?
That question is doing more work than it looks like. It rules out the two easy answers immediately.
Growing revenue by hiring proportionally does not count. Neither does cutting headcount to flatter the ratio, because that shrinks capacity and the number goes up while the business gets smaller.
What counts is the same people producing more, which is the only version that compounds.
Why this number and not the others
Hours saved is the metric everyone reaches for and it is almost meaningless on its own. Saved hours go somewhere. If they go into more of the same work, nothing happened. If they go into work that was never getting done, something did. The metric cannot tell you which.
Revenue per employee can. It only moves when the saved capacity turned into output somebody paid for.
It also survives contact with a bad quarter. Adoption dashboards look great during a revenue drop, which is exactly when you need a number that tells the truth.
And it is legible to the person signing the cheque. A small business owner does not need a translation layer for revenue divided by headcount.
What it looks like when it works
I had an idea on a Thursday. The challenge I set myself was whether it could be a company inside a week.
That is the shape of this. Not a large team executing faster. One person with enough leverage that the gap between an idea and an operating business is measured in days.
Every part of that is boring infrastructure doing its job. The tools have the practices already encoded, so nobody has to learn the platform before using it. The work that used to require a specialist now requires a decision.
Scale that thinking across a ten-person company and the ratio moves. Not because anybody worked harder.
Then talk about where you're going
The third step, and skipping it is why AI projects stall after the first win.
No small business wants to stay a small business. They want mid-market. They want bigger.
So ask where they're actually trying to get to, and then ask how AI widens the tent for what they can offer. Not how it makes today cheaper. How it makes a service possible that they could not previously staff.
That reframe changes what gets built. Cost reduction has a floor, and you hit it. Widening what you can sell does not.
What this costs you
Revenue per employee is the right north star for a small company and it has failure modes worth knowing before you put it on a wall.
It is trivially gamed by firing people. The ratio improves and capacity shrinks. If you adopt this metric, pair it with absolute revenue or you have built an incentive to become smaller.
It lags. Pain-point work shows up in a week. Revenue per employee moves over quarters. Somebody has to hold their nerve in the gap, and that is a leadership problem, not a measurement one.
It punishes hiring ahead of growth. Bring on two people in January for capacity you need in June and the metric says you got worse. That was the correct decision and the number disagrees, right? Know that going in.
Pain-point-first is politically expensive. You are telling a founder their exciting idea waits. Some of them will not, and you either build the exciting thing badly or lose the engagement.
It says nothing about whether the work is good. Revenue per employee rises just fine while quality drops, right up until churn catches it. It is an economics metric, not a quality one, and it needs a companion.
When it is the wrong metric entirely. Pre-revenue, or a business whose constraint is distribution instead of delivery capacity. If you cannot get in front of enough buyers, making your team more productive solves a problem you do not have.
The order matters more than the number
Pain points, then revenue per employee, then aspiration.
Teams that start with the metric build dashboards. Teams that start with aspiration build demos. Teams that start with pain earn the right to do the other two.
The short version. The One AI Metric Small Businesses Should Track.
Everything above stands on its own.
Chris