Headcount was always a proxy

Open any pitch deck and find the slide titled “The Ask.” Somewhere on it, almost always, is a number of people. Twelve engineers. Four in sales. Two ops. We treat that headcount as the plan itself, as if the company we’re describing is those forty-two future hires. But nobody has ever wanted forty-two people. They’re heavy, slow to find, and gone in eighteen months. What the founder on that slide actually wants is the output: the code shipped, the deals closed, the books reconciled. The people were only ever how you bought it.

We’ve spent a century confusing the receipt for the purchase. This post is about what happens when you stop.

The number on the slide was never the goal

Here’s the thing we say out loud at every board meeting and never quite hear: we’re hiring to get work done. The headcount is the means. Yet somewhere along the way the means swallowed the end. We started measuring companies by how many people they employ, sizing rounds by how many seats they’ll fund, and judging our own progress by whether the team got bigger this quarter.

You never wanted 50 people; you wanted the work 50 people do.

That sentence sounds obvious until you notice how much of how we build companies assumes the opposite. We don’t write job specs that describe outcomes. We write them that describe a person, years of experience, a list of tools, a seniority band, and then hope the outcome falls out the other side. The proxy got so familiar we forgot it was a proxy at all.

Headcount is a receipt for work you wanted. We’ve been collecting receipts and calling them the company.

For most of business history, that was a perfectly good substitution. If you wanted ten reports written every week, the only way to get them was to find a person who could write reports and pay them to. The work was inseparable from the worker. So we built every instrument we use to run a company, org charts, hiring plans, comp bands, the whole apparatus, around the one unit that could actually do the work. A human.

The substitution was so reliable for so long that we stopped seeing it as a substitution.

Why hiring-as-the-plan quietly fails

Let me stage the version everyone runs, because the pain in it is real and most of us have just learned to live with it.

You need more output, so you open a role. Best case, it takes two months to fill, sourcing, interviews, the offer, the notice period at their current job. Then a month of onboarding before they’re net-positive instead of net-negative on the team’s time. So a need you felt in March is met, partially, in June. By then the need has changed shape, because needs do.

And the unit you bought is lumpy. You wanted maybe sixty percent of a person’s worth of a specific task, but people don’t come in sixty-percent slices. You hire a whole human, give them the one job you needed plus four you invented to fill the week, and now you’re managing a role that exists partly to justify itself. Multiply that across a growing team and a strange thing happens: a meaningful share of the company’s work becomes the work of coordinating the company’s work. Meetings about meetings. Status on status.

Then the lumpiness cuts the other way. Demand drops, or shifts, and the capacity you so painfully acquired is now the thing you have to painfully shed. Layoffs aren’t just brutal to people; they’re the clearest possible proof that headcount and work were never the same thing. You don’t lay off the work. The work is still there. You lay off the proxy you bought to do it, because the proxy turned out to be a fixed cost wrapped around a variable need.

None of this is a failure of management. It’s a failure of the unit. When the only way to buy work is to buy a whole, permanent, slow-to-acquire human, every one of these pains is baked in from the start. You never wanted 50 people; you wanted the work 50 people do, and the lumpy, lagging, permanent shape of “a person” was the tax you paid to get it.

Decouple the work from the worker

So here’s the correction, and it’s smaller than it sounds. Stop treating “a person” as the atomic unit of work. Make the unit the work itself.

The naive reaction to that is “so, automation, you mean scripts.” And that’s the failure mode worth naming up front, because it’s why this hasn’t happened already. The old automation made one frozen task cheaper. You wrote a script, it did exactly that one thing forever, and the moment the task changed shape someone had to go rewrite the script. It couldn’t judge, couldn’t adapt, couldn’t pick up a tool it hadn’t been wired to in advance. So it could only ever take the most rote slivers, and the human stayed the unit for everything that required a decision.

What changed is that software can now make decisions inside the work, read the messy inbox and tell the three that matter from the forty that don’t, look at a half-filled form and know what’s missing, take a goal stated in a sentence and figure out the steps. The instant that’s true, the work stops needing to be bolted to a worker. It can be summoned when the need appears and released when it’s done.

When work decouples from the worker, capacity stops being something you own and becomes something you call.

Watch what that does to the timeline. The need you felt in March doesn’t wait until June for a hire to ramp. The capacity to meet it exists the moment you name the job, because you’re no longer acquiring a person, you’re directing a system that already knows your context to do the specific work in front of it. And when the need shifts, the capacity shifts with it. Nothing to source, nothing to ramp, nothing to lay off. The work is the unit, so the work is what you scale.

What a company looks like when work is the unit

Hold the old org chart up against this and it starts to look like a map of a constraint that no longer binds.

The org chart was always a compression artifact. You can’t direct a thousand individual tasks, so you bundle them into roles, bundle roles into teams, bundle teams under managers, and accept the overhead of all that bundling as the cost of staying coordinated. Most of a large company’s structure exists to manage the fact that its unit of work is an expensive, opinionated, hard-to-coordinate human. Change the unit and most of that structure is solving a problem you no longer have.

What replaces it isn’t a flatter org chart. It’s a different object entirely: a description of the work the company needs done, and a system that meets that description directly. You state the outcomes, the invoices reconciled by Friday, the inbound leads qualified within the hour, the weekly report on the board’s desk Monday morning, and capacity assembles around each one and dissolves when it’s met. The shape of the company stops being “who we employ” and becomes “what we’ve decided is worth doing.”

This is the part that’s easy to mistake for a cost story, so let me be exact about what it isn’t. It isn’t “the same company, cheaper.” A founder who used to spend the budget on twelve hires doesn’t pocket the difference and ship the same thing. They spend it on attempting twelve things that twelve hires could never have been afforded to attempt. The constraint that headcount imposed wasn’t only money, it was imagination. You scoped the company to what you could staff. Lift that, and the question stops being “what can we afford to hire for” and becomes “what’s actually worth doing.”

That’s a much harder question. It’s also the only one that was ever yours.

The turn

Notice what’s left in your hands when the proxy is gone.

For a long time, a founder’s day has been mostly proxy management. Writing the job spec. Sitting the interviews. Doing the onboarding, the reviews, the re-orgs, the painful conversations when the lumpy unit and the variable need stop lining up. It feels like the job. It has the texture of leadership. But step back and almost none of it is the thing you started the company to do, it’s the overhead of buying work through the only instrument that existed to buy it.

Take that overhead away and you’re left holding the one part no system can hold for you: deciding what the work should be. Which problem is worth a company’s whole attention. What “great” means for the person on the other end of it. Whether this is even the right thing to be building at all. The headcount was always a proxy for the work, but the judgment about which work matters was never a proxy for anything. It was the company, the whole time, wearing forty-two job specs as a disguise.

The people you do bring on change too, by the way. Not gone, chosen. You stop hiring to fill a gap in the machine and start bringing in the handful of humans whose taste and judgment you actually want shaping what the company decides to chase. That’s a colleague, not a unit of capacity. There’s a real difference, and you feel it in the room.

Close

The number on the slide was never the plan. You wrote down forty-two people because forty-two people was the only way you knew to buy the work you actually wanted. You never wanted 50 people; you wanted the work 50 people do, and for the first time, you can want the work directly. When that’s true, “how many people are you” stops being the measure of a company, and “what have you decided is worth doing” takes its place. That second question was always the better one. It was just buried under the first.

---

That’s what we’re building at Apollo Space: a company that’s sized by the work it decides to do, not the seats it has to fund. If you’ve ever written a headcount number onto a slide and felt, somewhere underneath, that it wasn’t really what you meant, you were right. It never was. The work was always the point, and now you get to ask for it by name.