I Ran an A/B Test Instead of an Argument

The CEO had already decided. We were a small team building a mental health app, and the pandemic had turned what we did into something people suddenly, urgently needed. So we were trying to grow, and growth meant choices: which country next, which language, which way to let people pay. He had a view on all of it, formed the way most founder views are, from instinct and a few loud signals and the pressure to move. I had a different view. And I was fairly sure mine was the better one.

That was the real problem. Not that we disagreed. That I was sure.

The Argument I Was Ready to Win

The easy thing, the thing I had done plenty of times before, was to make my case. Line up the arguments, find the supporting evidence, and out-talk the room until my opinion won. Sometimes it worked. But I had started to notice something uncomfortable about the arguments I won that way. Being convincing and being right are not the same skill, and I was better at the first than I had any right to be. A confident engineer can talk a room into almost anything, which is exactly why confidence makes such a poor proxy for competence, and why it is so easy to mistake the talking for the being right.

So this time I said something different. I said: we are both guessing. Let us stop guessing.

We Are Both Guessing

We had the one thing that made guessing unnecessary. We had users. Real people opening the app, often at the worst moments of their week, looking for something that helped. So instead of deciding which payment method to support in the new region, we put more than one in front of them and watched which they actually reached for. Instead of deciding whether to add Spanish on top of the German and English we already had, we tested it, and we tested what they wanted inside it, which turned out to be the audio and the video, the everyday sessions and the small rituals people came back for. It was the same restraint as learning to say not now, turned away from architecture and toward a decision.

The room had opinions. The data had an answer. And the answer was not the CEO's instinct, and I am honest enough now to admit I am not certain it would have been mine either. The test spared both of us from a confident mistake. We took the numbers to marketing and sales, the content team built what the users had asked for without knowing they were asking, and it shipped, to people who could now find help in their own language, in the form they would actually use.

The Ego That Hides Inside Competence

I was the technology person in that room. For years I had assumed that meant I was there to hold the most informed opinion. It took me longer than it should have to understand that more often it meant I was there to take the opinions off the table.

There is a particular kind of ego that hides inside competence, and it does not look like arrogance. It looks like being helpful. You know things, so you offer conclusions. You have seen this before, so you skip ahead to the answer. And in a small company under pressure, where everyone is a little frightened and everything is moving fast, a person who sounds certain is enormously comforting. That is exactly when certainty is most dangerous.

The strongest position in that room was not a better opinion. It was no opinion at all, held open just long enough for the evidence to arrive. It protected the decision from the CEO, and it protected the decision from me, and it handed the choice to the only people with any real right to make it, the ones we were building the thing for.

I still have strong opinions. I just trust them a little less than I used to, and I have made my peace with that. The most useful thing I bring to a hard decision now is not the answer. It is the willingness to be proven wrong by the people it is for, and the discipline to go and find out before I fall in love with being right.