The master user and the K factor

The investor’s most actionable idea deserves its own record: find the perfect user, someone genuinely desperate for an app like this, willing to test everything and say everything they want, and then implement what that person asks ▸ 12:17, “el usuario maestro” ▸ 12:50. It’s the complement to learning to filter feedback: once you can distill signal, go find the loudest honest source of it. The investor’s tactical suggestion, message the people leaving one-star Duolingo reviews, dies on platform reality, the Play Store doesn’t let you contact reviewers ▸ 13:09, so the executable version becomes Reddit: find users publicly dissatisfied with existing tools, write to them, and iterate with them before doing anything else ▸ 12:57.

cada botón de invitar a un amigo es un instrumento de medición →

The measurement side got a one-pager ▸ 13:48. The KPIs he drafted: spaced-repetition efficiency, does Severo make you memorize words in less time, honest caveat attached that he doesn’t yet know how to measure it ▸ 14:14; streak retention, how few people lose the racha ▸ 14:32; and the K factor, the virality coefficient ▸ 15:09. That last one reframes a button he’d never examined: Duolingo’s invite-a-friend isn’t a growth gimmick, it’s an instrument, because an invitation is the strongest signal a user can send that the app is working on them ▸ 15:35, worth replicating for the measurement as much as the growth.

On advertising he keeps his small heresy alive despite the investor’s no: run a 100,000-peso test and compare it against the paying users it produces; if revenue beats spend, the channel funds itself ▸ 16:05, though he closes with the orthodoxy intact, what matters is improving the product ▸ 16:21. Julia’s design warning rounds out the docket: every game type currently shares one aesthetic, which reads as more-of-the-same, and she predicts the interface still changes a lot ▸ 11:24. He doesn’t disagree; he just wants the master user to be the one who says it…