The launch everyone told me to stop.

The number on the screen was bigger than any I’d been responsible for, and I was fairly sure I was about to lose it all.

We were rebuilding the entire way the marketplace sold. The old way worked, in the sense that it produced revenue and nobody was actively on fire. But it produced revenue the way a person carrying water in their hands produces a full glass. Most of it ran out between the well and the table, and the only fix anyone knew was to walk faster. I wanted to stop carrying water in our hands. That meant tearing down a selling process that a lot of people had built their weeks, their comfort, and their sense of competence around.

The pushback came fast, and it came from people I respected. Good operators. People who had shipped more than I had. The message, in a dozen versions, was the same. The current thing works. You are introducing risk into something that is not broken. You are going to slow us down right when we cannot afford to be slow. One conversation I still remember ended with a version of “you are going to break the one thing that is paying for all of this.”

And here is the part I don’t enjoy admitting. For a while, I thought they were right.

Because when you tear down a working process to build a better one, there is a stretch in the middle where you have neither. The old process is already half-dismantled and the new one is not yet carrying weight. Every number that matters dips. Every person who warned you gets a small, daily piece of evidence that they were correct. And you are standing in the gap holding a plan and a conviction and nothing you can point to yet. I have never felt less certain of a decision I was more sure of.

I did not back down. Not out of bravery. I’d done the quieter work first, before any of the noise started. I had written down what we were optimizing for, and it was not the number that dips in the middle. It was the number that keeps growing after everyone stops watching. The near-term dip was not a sign the decision was wrong. It was the price of the decision, known in advance, already on the invoice.

So when the fast feedback rolled in, loud and reasonable and from people I trusted, I had something to hold it against. I could ask a single question of every objection. Is this about what I am optimizing for, or about what is temporarily uncomfortable on the way there? Almost all of it was the second thing. Real discomfort, honestly felt, aimed at a metric I had already decided to spend.

The slow number arrived the way slow numbers always do. Late, quietly, and long after everyone had stopped watching for it. The marketplace grew past a billion euros in goods moving through it, and the process we rebuilt let it scale without me standing in the middle of every deal. The people who told me to stop were reading the fast signal, because on the day they spoke it was the only signal there was.

Before this turns into a hero story, let me take the other side of it. Holding the line only works if you did the decision work first. If you tear down a working process on instinct, with nothing written down about what you are trading for what, then the people telling you to stop are the only feedback you have. Listen to them. Conviction without a written-down outcome underneath it is just stubbornness wearing a nicer coat. It was never about ignoring criticism. It was about knowing, before the criticism starts, exactly which number you had agreed to let dip.

I work with founders now who are standing in that same gap. Every advisor and every board member reading the fast signal and offering the fast verdict, and the founder in the middle, sure and unsure in the same breath, wondering if the confident people are right.

Most of the time, the confident people are just early.

So before you tear anything down, do the boring thing first. Write down the one number you are actually building toward, and the one you have agreed to let dip on the way there. Then, when the fast feedback comes, and it will come loud and it will come from people you respect, you will have the only thing that makes it survivable.

You will know which number you were building toward, and the fast feedback never gets to decide it for you.