Most of what I've learned in a startup, I learned from not knowing.
Not knowing whether the thing we were building mattered. Not knowing whether the person sitting across from me, who had spent fifteen years in an industry I'd never worked in, was describing a real problem or just a familiar one. Not knowing, on the harder days, whether we were early or simply wrong.
People talk about the unknown as if it were a wall. It isn't. It's more like fog. You can walk in fog. You just can't walk fast, and you can't walk carelessly, and you have to keep checking that you're still going the direction you meant to go.
Ambiguity itself wasn't new to me. I started in development, then systems engineering. I worked on mergers and acquisitions in IT, where you inherit someone else's decisions and have to make them fit. I spent years in solution engineering, which is mostly the art of answering a question nobody has asked correctly yet. I started a non-profit, which is a startup with worse odds and no cushion. Every one of those was a walk in fog.
So the fog didn't surprise me. What surprised me was where it came from.
I had assumed the uncertainty would come from outside: the market, the technology, the customer. Some of it did. But most of the fog I remember was made in-house. A vision nobody could state the same way twice. A niche we hadn't actually decided on. A hierarchy that determined who got heard, and politics on top of all of it. That fog is optional. It's also the worse kind, because the market's fog at least has something on the other side of it.
Direction beats speed
The temptation, when everything is uncertain, is to move. Movement feels like progress. And a team moving quickly in a slightly wrong direction looks, from the inside, exactly like a team that is winning.
Direction is worth more than speed. A plan tells you what to do this quarter; a reason tells you what to do when the plan breaks, and the plan always breaks. So the first real asset is a clear, honestly held reason for existing, specific enough that it rules things out. If it doesn't help you say no, it isn't a reason. It's a slogan.
Vision does something no process can: it lets people decide without asking. When the reason is clear, a designer and an engineer and someone who spent a decade doing the customer's job make three decisions that fit together, without a meeting. That is alignment. Not agreement. Coherence.
Where we came from shaped what we saw
Most of our team came from sales. The rest came from everywhere else. I am not a seller. I came to it from building things, which is probably why the pattern stood out to me so early.
A background isn't just experience. It's a lens. And the sales lens says, in effect, that every problem is a sales problem. Not enough revenue? Talk to more people. Customer unhappy? Get in the room. Growth flat? More activity at the top.
I want to be fair to that view, because it is mostly true. Almost every problem does eventually show up as a sales problem, and talking to customers is the highest-return habit a company can have. Nothing gets learned in a building.
But "it shows up in sales" and "it gets solved in sales" are different claims, and confusing the two is expensive. Some problems are the product telling you something. To hear that, someone has to stop and ask how the thing actually works. Not how we'd like to describe it, not what the customer asked for, but what is really happening underneath. That discipline is slower, less visible, and it does not respond to effort the way a pipeline does.
The vision that worked for us was simpler than either instinct wanted. Solve a real problem. Not an interesting problem, not a fundable one. A real one.
And real problems come with a useful property: it doesn't matter who you tell. You can say it out loud, to anyone, including the people building against you, and there is no way for them to copy it. Hearing a problem described is nothing like being able to solve it. That is the test. If telling someone your idea is dangerous, the idea was the whole thing you had. If you can tell everyone and nothing changes, you're working on something real.
A track record is not a prediction
Here is the assumption I'd most want to take out of people's heads: that because something worked before, it will work again.
Years in a vertical, real wins, a reputation earned. That is a genuine asset, and I don't want to talk anyone out of respecting it. But success in a settled world is evidence about a world that had already been figured out. A new venture is a different game with different rules, and the conditions that made someone right last time are usually the exact conditions we're now supposed to be questioning.
The trap isn't the experience. It's that experience feels like certainty. And certainty in an unfamiliar environment is just guessing with better posture.
The loudest voice should not win
For the first couple of years, that's what I watched happen. Decisions didn't go to the best-reasoned case. They went to whoever pushed hardest, longest, in the most rooms.
The quieter version did more damage. A small remark. A raised eyebrow. "Well, in my experience, that's not how this works." Never hostile enough to name, never quite deniable either, and just heavy enough that someone with a good idea decides not to bother next time. I watched people go quiet who should have been the loudest in the room. You never find out what those ideas were.
Alongside it came everything that follows from it. Politics. Decisions made somewhere out of sight and announced later. Executive-shaped titles inside a company small enough to fit around one table. Hierarchy imported from places where hierarchy made sense.
That isn't a startup. It's a small slow company wearing a startup's clothes. And it is expensive in the only currency that matters at that stage, which is time, because every hour spent managing who gets credit, who was consulted and who outranks whom is an hour not spent finding out whether the product is any good.
Flattening it is not a nice thing to do for morale. It's an operating requirement. Two tests I'd hold anyone to. The best idea has to be able to come from anywhere, including from the newest and quietest person present. And everyone should be able to see how a decision got made, not just what was decided.
Politics is not a stage a company passes through. It's a tax you agree to pay.
Reacting is not direction
There's a failure I now see everywhere, and it's new.
Product direction gets set by reacting to the market. Someone watches what competitors ship, asks a model to research the space, gets back a clean summary of what exists, then builds toward the middle of that summary. It feels rigorous. There are sources. There is a landscape.
But a model trained on what has been written can only tell you what is already there. Direction built from that description will produce a product that already exists. You will have worked very hard to arrive at the average of your competitors.
The alternative isn't ignoring the market. It's refusing to let the market do your thinking. Look at the problem itself, from the beginning, and reason your way down to what is actually true about it. Sit with the one customer whose work you can see. What you find there won't be in anyone's summary, which is the entire point.
What is actually hard to copy
Selling is not the hard part, and I say that with real respect for the people who do it well. The market is flooded with tools for selling; anyone can build another one. The playbooks are public, the tooling is off the shelf, and five people can stand up a credible sales motion in a month.
What is hard is a product deep enough to matter in one industry and shaped well enough to travel to the next.
That pulls in two directions. Go deep in one vertical and you earn something real: the exceptions, the vocabulary, the ugly edge cases nobody documents. That depth is not buyable. You get it by being in the work.
But depth alone leaves you with a very good tool for one small world. The compounding comes from the second move, finding the shape underneath the specifics. The part that is the same in two industries that look nothing alike. That is patient, unglamorous engineering, and almost nobody does it, because it doesn't demo well and it doesn't close a quarter.
It's also why it lasts. A competitor can copy your pitch this week and your feature this month. They can't copy five years of knowing which exceptions matter and having built for them.
What it takes
Looking back, none of it was complicated, which is different from saying it was easy.
A real problem, and a reason for existing clear enough to say no with. A flat room where the best idea wins regardless of who brought it, and where nobody is spending energy on rank. Enough humility about what we'd each done before to treat it as a hypothesis rather than an answer. And the patience to build the thing underneath, the part that takes years and doesn't demo well.
The unknown is where the opportunity is. You get paid for going where the map isn't. The reason for going is what keeps you pointed the same way long enough for any of it to compound.
Clarity about why. Patience about how. Respect to all.