I Sold the Product Before I Built It

Concept-only customer interviews, no demo, no prototype — and why a commitment at the end of that conversation is the only demand signal I trust

The hardest thing I did this year wasn’t writing the architecture. It was sitting across from someone who wanted to know what the product looked like, and refusing to show them anything — because there was nothing to show, and because showing something would have ruined the only measurement I cared about.

I had decided, before I wrote a line of the reasoning layer, to run the interviews on the concept alone. No demo. No prototype. No mockup I could pull up on a laptop to fill the silence. Just the framing of the problem and the shape of the answer, described out loud, and then a question at the end: would you adopt this? If the answer was yes — a real yes, a put-it-in-the-roadmap yes — then the pain was real. If it was a polite yes, I’d know, because polite yeses don’t survive a follow-up about budget and timeline.

I want to describe why I did it that way, because the discipline of it is the whole point, and it is easy to talk yourself out of.

The instinct, when you have a strong idea, is to build something first. A demo feels like progress. It feels like the responsible thing — you wouldn’t ask someone to commit to a thing they can’t see. So you build a prototype, you show it, and people nod, and they say it looks great, and you walk away convinced. The problem is that you have no idea what you just measured. A demo sells. A good demo sells well. The nod you got was a reaction to the artifact — the animation, the clean screen, the feeling of a thing that works — and almost none of it was a reaction to the problem. You learned that you can build something that demos well. You did not learn that anyone needed it.

The concept interview inverts the thing being tested. With no artifact in the room, there is nothing to react to except the pain. When I described the situation — the questions a team wonders about and never asks because finding out costs a day of someone’s time, the decisions that quietly get made on gut feel because the real answer was too expensive to fetch — the person across from me either lit up or they didn’t. There was no screen to be polite about. If they leaned in, it was because I had named something they lived. And when, at the end, I asked whether they’d adopt a thing that made that pain go away, the yes I got was a yes to the problem being solved, not to a piece of software I’d shown them. That is the only yes worth building on.

The pain, doing it this way, is entirely mine. The conversation is harder without a prop. You have to carry the whole thing in language — the problem, the mechanism, the world after — and you have to do it well enough that someone can commit to something they have only heard described. There were moments I wanted, badly, to pull up anything, just to break the tension of describing a product that didn’t exist yet. The discipline is in not doing that. Because the second you show an artifact, you stop measuring demand for the solution and start measuring reaction to the artifact, and you can’t run the measurement twice. The first showing spends it.

Here is the part that took me a while to trust: a commitment at the end of a concept-only interview is a stronger signal than a commitment after a demo, not a weaker one. It feels weaker — you got the yes with less to show, so surely it’s worth less. The opposite is true. The demo-driven yes is contaminated by the demo. The concept-driven yes had nothing to lean on but the problem. When someone says they’ll adopt a solution they’ve only heard described, they’re telling you the pain is large enough that they’re willing to bet on the description of a fix. That is a much louder signal than someone admiring a screen.

And it changes what building means. Most products get built on assumed demand — the founder believes the pain is real, builds the thing, and finds out at launch whether the belief was right. The concept interview moves that discovery all the way to the front. By the time I started building, the demand wasn’t a hypothesis I was about to test. It was a thing several conversations had already established. The product, when it showed up, wasn’t the experiment. It was the confirmation of an experiment that had already run — in words, across a table, before any code existed. The thing I built was built to honor commitments people had already made to a description.

There’s a version of this that sounds like a trick — talk people into a yes, then build whatever. It isn’t that, and the difference matters. A concept interview is not a pitch. The goal is not to get to yes; the goal is to find out whether yes is there. The structure has to leave room for no, and most of the value is in the nos — the conversations where someone heard the whole framing, understood it perfectly, and said that’s interesting, but I wouldn’t change anything to get it. That’s not a failure of the interview. That’s the interview working. A no on the concept is a no you got for free, before you spent a year building the thing it would have rejected.

What I’d tell anyone earlier in this than me: the order is the lesson. Discovery first, building second, and the wall between them held honestly. The temptation to soften the wall — to bring a little proof into the discovery, to start building before the conversations are done because you’re impatient — is the temptation to stop learning anything trustworthy. Keep them separate. Make the conversation carry the weight. If people commit to a thing they cannot see, you have found the right pain, and everything after that is execution against a demand you already know is real.

I’m still doing these. Even now, with a product that exists, the most useful conversations are the ones where I describe a capability that isn’t built yet and watch whether the person reaches for it — because that reach, before there’s anything to reach for, is still the cleanest signal I have ever found.

Key Takeaways

• A demo sells; that’s exactly why it’s a bad instrument for discovery — you can’t tell a reaction to the problem from a reaction to the artifact.
• The concept interview removes the prop so there’s nothing left to react to but the pain.
• A commitment with nothing to show is a stronger signal than one after a demo — the yes had nothing to lean on but the problem.
• Concept interviews move discovery to the front: by the time you build, demand is established, not assumed.
• The nos are the point. A no on the concept is a year of building you didn’t waste.