I ran persona interviews before building a feature — and three unrelated users independently asked for the same thing I never planned

I'm building a BJJ technique app as a side project. My original plan: searchable technique database, step-by-step breakdowns, YouTube links. Clean, simple, done.

Then I ran structured persona interviews before writing any code for the next feature. Three different user types — a white belt IT engineer, a time-constrained working mom, and a blue belt who coaches beginners. Different experience levels, different goals, different constraints.

All three independently asked for some version of a "technique tree" — a visual map showing how techniques connect. A feature that wasn't in my design doc at all.

But here's what made it interesting: they each described completely different things under that same label.
The engineer wanted an RPG skill tree with unlock progression
The mom wanted a filtered learning path — "show me only what's relevant for my level"
The coach wanted a custom game plan builder to share with beginners

Same surface request. Same underlying need. Three different product requirements.

The rule I now follow: one persona asking for something = interesting. Two = worth investigating. Three from different segments = build it.

I wrote up the full process here, including the Stanford research on AI persona simulation that made the interviews actually produce useful signal (instead of generic "yes that would be helpful" answers): full article on Dev.to

Curious — has user research ever surfaced a feature you never would have thought of yourself?