Talking about culture fit means you've already stalled
The other day I had dinner with some people from a startup in Shenzhen.
They were about three years into the company’s life, and their headcount had jumped to roughly 150 last year. They were fired up about hitting around 300 next year.
From my own experience running startups and managing small and medium-sized companies, I know how hard it is to grow headcount. If you’re close to management, or if you’ve managed a team of any size, you’ll know exactly what I mean. What am I getting at?
The difficulty of culture fit.
As a team grows, people come in from different jobs and different backgrounds. Clashes of good intentions start happening everywhere — in a single task, a single meeting, a single product spec. Everyone means well, but differences in culture and experience create friction, and that friction breeds conflict. Weighting culture fit is essentially a gatekeeping strategy: you either fill the pipeline with people whose intentions are already aligned, or you bring in people who are still close to a blank slate and dye them in the company’s culture.
Going from 150 to 300 people in a year is an extremely precarious move. So I asked one of the company’s early members, a manager, about it.
“How do you think about culture fit?”
His answer was simple. “Well, everyone’s got different taste in food, right?”
We weren’t on the same page. If I’d said this to someone in management in Japan, it would have landed instantly as a familiar story. But this wasn’t Japan. I figured I needed to explain more carefully.
So after explaining, I asked again.
“As headcount grows, you start seeing differences in culture and differences in development style show up inside the company, right?”
His answer still missed the point. “Well, guys from Sichuan love super spicy food — oh, this restaurant serves Sichuan food, but toned down for people in Shenzhen. Good, right?”
He looked a little annoyed, though. Since we were both speaking English, neither of our native languages, exchanges like this sometimes don’t quite connect. We were both a bit drunk by that point too, and I’d explained at some length, tossing it at him with a “you get it, right?” tone, but in the end it still didn’t land.
He never said anything like:
“What matters is whether people commit properly to the mission.”
What follows is simply my own impression. I hope it rises to the level of analysis.
Many of you already know this, but the first thing you notice interacting with Shenzhen startups is how young the members are. In most of these startups, the average age rarely reaches the late 30s — it’s early 30s, or often just the 20s. They’re not just full of momentum; they’re diligent, brimming with the drive to work 24/7, 365 days a year.
There’s tremendous energy there.
I remember something like this myself. In my twenties, a certain number of people around me worked at an almost deranged level. There was a stretch when leaving the office at 3 a.m. was normal for me. Three in the morning. I’d bike the ten minutes home, sleep, and come back in at 10 a.m. I repeated that over and over. I wasn’t calculating my pay, and I had nothing much to protect in my personal life. If trouble broke out at an overseas factory, I’d fly out the same day.
Why was I doing that? Because it was fun.
I have another experience like this. I once joined a just-launched startup in Silicon Valley as its first employee. Individual judgment calls were vague, nothing was managed, no processes were fixed — it wasn’t so much chaos as a blur over everything. But we pushed forward anyway, brushing all that aside. Just build, just ship, just tinker, just sweat.
The chain of command and a certain kind of loyalty were clear, though. Direction came from the C-suite. The details flowed fluidly on the ground. That was enough.
I also had the experience, starting a startup in Japan, of running into that familiar friction of fragmentation. Looking back, it was without question a bitter experience. At the time, I pushed forward with “direction comes from the C-suite, details flow fluidly on the ground” — except this time I was the one in the C-suite. The launch of the first product went well, but after that, under the pressure of fragmentation, everything started going fatally wrong. That was the first time I truly understood what the phrase “aligning culture” meant.
What was different?
At the time, I had no idea. Looking back calmly, the reason I, as a lowly engineer, had been able to throw myself completely into working toward a goal, day and night, without calculation, purely — it wasn’t because the management was good. My boss at the time was understanding and kind, but I don’t particularly remember him doing anything that struck me. I think I did it on my own will, with my own passion.
Yes, passion.
During the period when I was getting crushed by the pressure of fragmentation, passion was what was missing.
There’s something else I noticed spending time in the office of that Shenzhen startup.
There’s little conversation.
Almost no conversation between employees. Everyone is focused, absorbed in their work, but it’s not that Chinese people are particularly quiet by nature. I think it’s simply that there isn’t much need to consult each other.
China is still a country with abundant labor, and because of that, division of labor is highly advanced there. I felt that even the US has more advanced division of labor than Japan. Japan, I suspect, is relatively speaking a country where division of labor is not very advanced by global standards.
The downside of division of labor is that employees’ understanding of the business as a whole declines, which in turn makes internal department transfers difficult. I don’t know much about China on this point, but in the US, department transfers basically don’t happen in the first place, so there was never anything to worry about. Here too, Japan is a bit of an odd country.
The upside of division of labor is, needless to say, depth. By making each person’s area of responsibility clearer, they can pursue a specific category relentlessly without getting distracted. You get positive effects across the board — efficiency, adoption of new technology, cost reduction, schedule adherence. What I worked on during my 10-to-3 days was, simply, “get this specific product’s spurious radiation to pass.” I remember it well. I had to study, I had ideas but not enough time to test them, but I just kept tackling whatever I could, relentlessly. Sometimes it didn’t work out. But I did it anyway, out of pride and stubbornness. That was inevitably a period when a lot of knowledge and experience piled up.
And as division of labor advances, there’s less need to consult other members. As a result, you can concentrate on the work right in front of you.
Of course, within what a team does, there are naturally layers of division of labor — between teams, and within a team. But the unit of team management stays small: five or six people, up to maybe twenty at most. That’s actually a group size where you don’t need to think about “culture” at all.
It’s probably through this kind of structural accumulation that, even once an organization reaches 200 or 300 people, there’s hardly any need to think about culture.
You could also hypothesize that division of labor and the resulting increase in management overhead roughly cancel out. For instance, something very important when dividing labor is setting detailed responsibilities and goals for each individual. That increases management workload. On the other hand, it becomes much easier for a manager to understand what a given person is doing, so judging whether results are being delivered, whether improvement is needed, becomes very simple. If someone is active across multiple teams, evaluating that person gets hard. In one team they might show almost no results and seem unmotivated, while in another team they belong to, they might be delivering enormous results, full of energy and drive. So how do you evaluate someone like that? Is this scope of responsibility even appropriate going forward? All sorts of worries like that arise. With division of labor, none of these worries exist at all.
Design is hard, but evaluation and correction are easy — that might be the nature of division of labor. Does it cancel out to zero? No — maybe, on net, the total management overhead is actually negative, because you don’t need to manage culture fit.
Here’s my hypothesis: in a company with sufficient passion and sufficiently advanced division of labor, managing culture may become unnecessary.
Anyone who’s spent time in the startup world probably recognizes this a little. A startup on the rise often experiences a situation far removed from division of labor — “there aren’t enough hands, so everyone has to do everything.” But that’s really only true in the earliest phase. Soon enough, a deep valley of division of labor emerges. It’s a situation that’s essentially a cluster of single points of failure. Nobody but so-and-so understands the cloud infrastructure at all. Nobody but so-and-so grasps this PCB’s circuit. Nobody but so-and-so understands what this spreadsheet means. That’s the situation. But all of that comes together as one mass, and the business moves forward — truly, as one mass. At that point, nobody’s mouth-breathing on about something as yawn-inducing as culture fit.
But at some point, people start picking at other departments’ performance, issues that come up get passed back and forth with nobody taking ownership, and people start worrying about insufficient communication. This is precisely the gateway to big-company disease. If people are enjoying the work in front of them, focused and absorbed, pouring their heart into it every day, none of that happens.
At the management layer, the reason sufficient division of labor isn’t achieved is that it’s unclear what to focus on. If it were clear where the lever is that determines whether the business succeeds or fails, pouring resources, focus, and passion into it would be a very straightforward story.
But because that isn’t clear, people end up doing a bit of this and a bit of that.
There’s something I’ve consistently told startups and executives when advising them: “R&D shouldn’t divide time — it should divide people.” This is advice that’s almost never been followed (laughs). People just aren’t that versatile.
Management is rarely something you can get through on a single leg. You need portfolio design for everything. And here too, the best solution it leads to is division of labor.
The link between division of labor and passion isn’t all that clear, but the link between division of labor and focus is comparatively clear. That’s true from my own experience as well.
In that sense, passion might be the minimum spark, and division of labor the kind of enclosure you build around that spark. An enclosure isn’t just about blocking wind from outside. Put an enclosure around a spark, and the air gets funneled straight from bottom to top through a narrow space, creating a chimney effect that boosts the fire’s intensity. A little tidbit from camping experience.
Are they doing what I’ve described here strategically? I don’t think so, really. But it can still be analyzed structurally and learned from. I hope people learn from it, and that a Japanese-style version of it emerges.
Originally published in Japanese at https://clazytech.com/2026/05/1615/. Translated with LLM assistance and reviewed before publication.