The 10x engineer is the most expensive thing your project can buy. Not in salary, in downstream cost.
The standard agency model still leans on the “hero hire.” The senior person who can ship in an afternoon what a junior would take a week to write. And on paper, the math is obvious: pay 2x, get 10x. The unit economics are unbeatable.
Except they aren't. We've seen the receipts.
The 10x myth, briefly.
The original DeMarco / Lister research from the 80s did find a 10x productivity spread between programmers. It also found that the difference correlated far more with environment than with raw talent: quiet rooms, fewer interruptions, ownership of the work. The myth that got popular kept the multiplier and dropped the environment.
What survives in the discourse is just “some engineers are 10x.” What got lost is “and the only way to get them is to build the studio around them.” The agencies hire 10x people. Then they put them on three projects at once, in open-plan offices, in two-week sprints, and act surprised when the work feels rushed.
The 10x engineer is the most expensive thing your project can buy. Not in salary, in downstream cost.
What the hero-shop model actually produces.
A senior writes a clever solution that fits in their head. Nobody else on the team understands it. The hero moves to the next client. The next person who touches the code, six months later, after the hero has left, spends two days reading before they can change a button.
We see this in every agency-built codebase we inherit. The pattern is identical: clever framework choices, custom abstractions over standard libraries, type gymnastics in service of a 40-line file. It works. The launch happens. Then the silence after the launch is interrupted only by the sound of the next developer struggling.
Loose averages from twelve projects we've adopted in 2024-2025
What a small studio produces instead.
The small studio has a structural advantage that hero shops don't: the people who wrote it are still here. The whole point of a studio is staying. The launch isn't the end of the contract; it's the start of a relationship that lasts years.
Ships fast. Walks away.
Optimised for the launch. The deck looks great. The roadmap doesn't survive contact with month four. Cost compounds over the next two years.
Ships at 1.5x. Stays.
Slower at the launch, not by much. Better at every quarter after. Same people in month one, month twelve, month thirty. The code stays legible because the writer is still reading it.
That structural fact, same people, year after year, makes a thousand small decisions different. Code gets reviewed properly because the reviewer will own the change. Abstractions stay shallow because the writer knows they'll be writing the next feature too. Documentation gets written because the next reader is themselves.
Continuity beats throughput.
The bet of the hero-shop is throughput: more done, faster. The bet of the small studio is continuity: same hands, year after year. Most clients don't need throughput. They need their thing to work, to keep working, and to evolve when they evolve.
The hero-shop wins the pitch. The small studio wins the second contract. And the third. And the seven-year anniversary.
What we hire for.
If we're not hiring 10x engineers, what are we hiring? Plain answer: 1.5x engineers who stay. Specifically, we look for three things, in this order:
1. They can explain their work. If they can write a clear PR description, they can write a clear commit, they can write a clear function. The cause is the same: thinking before typing.
2. They're bored by cleverness. Boring code is the goal. We screen out candidates who light up when asked about exotic patterns. We light up when they say “I'd just use a for-loop.”
3. They care about the second-week feeling. Anyone can write code that works the day they wrote it. We hire for the people who think about how that code feels in week two, when they've forgotten the context.
None of these are 10x traits. They're studio traits. They compound.
A defence of the small room.
There's an old idea, well-articulated by Peopleware and quietly understood by every senior engineer who's ever worked from home: the work happens in quiet rooms with one project on screen. Not open-plan offices. Not three Jira boards. Not Slack DMs during deep work.
The small studio gets to have small rooms by default. Three or four people, one project at a time, in a converted living room above a café. That isn't a charming bio detail. It's the work environment that produces better code.
If you can't afford the small room, because you're an agency with twelve projects and forty engineers and a London office, you can't produce small-studio work. You can produce hero-shop work, very well, with a very polished launch. But the launch is the easy bit.
Most clients don't need throughput. They need their thing to work, to keep working, and to evolve when they evolve.
This isn't an argument that big agencies are bad. They're good at things small studios will never be good at: multi-region rollouts, regulated industries, anything that needs a hundred people. We don't do those.
What we do: small projects that need to last. For those, the math really is different. 1.5x for ten years beats 10x for six months, by an order of magnitude, on every metric that matters.
The 10x engineer is fine. The 10x engineering culture is the problem. And the small studio is the alternative most people don't know they can pick.
— A.B., from the studio, in a quiet room.