Unpopular opinion: "Talk to 50 users before you build" is overrated advice

I'll probably get roasted for this, but hear me out.

The standard advice is "talk to 50 potential customers before you write a line of code." I did that for a project once. I talked to 40 people. I took meticulous notes. I built exactly what they said they wanted.

And you know what happened? They didn't buy it. Not a single one. The same people who said "I would definitely use this" looked at the finished product and said "oh, interesting" and went back to their lives.

I've since realized that people are terrible at predicting their own behavior. They'll tell you what sounds reasonable, what makes them feel smart for saying, or what they think you want to hear. But when it comes time to actually change their habits or pay money, a different person shows up.

I'm not saying "don't talk to users." I'm saying the value isn't in the answers they give you. It's in watching what they do when they're not being interviewed.

For those who've had a "talk to users" failure like this—what did you change in your approach that actually helped? Did you switch to "sell before building" or "watch behavior not words" or something else?