The feature nobody asked for, and how long I kept it alive

The most expensive thing my first company ever built was not a mistake we made quickly. It was a mistake we made slowly, lovingly, over most of a year, and we were proud of it the whole way down.

The demo that landed

It started with a demo that got the room. We had built something clever, a feature that did a genuinely difficult thing in one smooth motion, and when we showed it, people leaned forward. You learn to live for that lean. A founder in an early meeting is starved for any sign that the thing in their head is real, and a room leaning forward is the strongest drug there is. We decided, on the strength of that lean, that this was the heart of the product.

What we did not notice was that the people leaning forward were almost never the people who would actually use it. They were other founders, investors, friends who built things for a living. They admired it the way you admire a clever chess move. Admiration is not the same as need, and it took me an embarrassingly long time to tell the two apart.

The sunk cost nobody warns you about

So we built it out. Every month it wanted more: more edge cases, more polish, more support when it broke, which it did, because clever things break in clever ways. And every month a quiet number sat in our dashboard telling us the truth, which was that hardly any customer ever touched it. We had an explanation for that, of course. We always did. They had not discovered it yet. We had not explained it well enough. The right customer was coming.

The trap is not that you build the wrong thing. Everybody builds some wrong things. The trap is that the more you have poured into a thing, the harder it becomes to hear what the numbers are plainly saying, because killing it now means admitting the last nine months were a detour. So you keep it alive. You feed it. You defend it in meetings. It becomes, without anyone deciding this, the thing you are protecting the company from rather than the thing the company is for.

What I ask now before I build

We did eventually turn it off. The world did not end. A handful of people wrote to ask where it had gone, and we helped them, and that was that. The relief in the team was almost physical, the way a house feels after you finally clear the room everyone had been squeezing past for a year.

Now, before we build anything that will be hard to remove, I make us answer three plain questions. Who exactly asked for this, by name, not in the abstract. What will they stop doing once they have it. And how will we know, in numbers, within a month, whether we were right. If we cannot answer the first one with a real person, we do not start. The lean of a room is a wonderful feeling. It is just not a customer, and it will never once pay an invoice.

Success brings its own expensive temptations, and The Second Shop is about the biggest one: the question that sounds like a compliment and lands like a threat.