The Trap: Confusing Cult for Fit
You have users who open your app three times a day. They file detailed feature requests. They evangelize in WhatsApp groups. By every engagement metric, you're winning.
Then someone checks cohort retention. Month 1 retention is 8%. Month 2 is 3%.
You rationalize: power users are sticky. Early users are always evangelists. Retention will improve with feature work.
It won't. You're watching survival bias masquerade as product-market fit.
Why Power Users Lie to Founders
Power users are not a representative sample. They are the people for whom your product solved a specific, intense, narrow problem. They are not the people your product was supposed to serve.
The Core Trap
Power users are survival bias: the 1% who stayed, not the 99% who left.
Consider Figma in 2014. Dyson Tan's design team (the power users) used it obsessively. But 95% of designers still used Sketch. Figma's power users were not proof of fit. They were proof of partial fit for a specific subtype of designer (collaborative, web-native, team-heavy).
Figma did not declare victory. They studied the 95% who didn't convert, discovered what they actually needed (offline mode, plugin ecosystem, performance on large files), and shipped. That took three more years.
Most founders ship for power users instead. They add dark mode, custom workflows, integrations. The power users love it. Everyone else still leaves.
The Math That Hides the Problem
Consider two products:
Product A: 1,000 signups. 100 DAU. 10 power users. 87% of DAU comes from those 10.
Product B: 10,000 signups. 800 DAU. 50 power users. 42% of DAU comes from top 10%.
Your VC tells you Product A has "better engagement." Your data tells you Product B has better distribution and retention across cohorts. Product A is a hobby. Product B is a business.
Most founders raise on Product A metrics. They show power-user NPS (usually 8–9). They show power-user case studies. They do not show 30-day churn.
Engagement concentration in product A (power users) vs. product B (distributed DAU). A looks better. B scales.
How to Diagnose the Problem
Three signals that power users are hiding churn:
1. Power user intensity is rising while cohort retention is flat.
If your top 10% increase from 20% to 40% of DAU while overall retention stays 8%, you have user concentration, not growth. The product is becoming more niche, not more essential.
Test: take your top power users out of the data. Calculate DAU. If it drops below 5% of signups, you have a cult, not a platform.
2. Feature requests from power users don't move retention for new cohorts.
You ship the collaboration feature power users asked for. DAU goes up 15%. But new-cohort retention stays 8%. The feature served the existing power users, not the market. You optimized for retention of people who were already retained.
Test: run a separate cohort analysis. Track retention on the feature separately for power users (existing) vs. new users. If the gap is >3x, you're designing for the wrong segment.
3. NPS is high, but NPS detractors churn faster than promoters grow.
Power users give you 9s. The 800 people who left gave you 4s. A 40-point NPS gap sounds validating. But if 80% of signups are leaving before they can become promoters, NPS is a lagging indicator, not a leading one.
Test: calculate "net new power users per month" (users who reached power-user threshold, minus those who left). If it's negative or flat, your power users are not reproducing. You're harvesting a fixed pool.
The Messy Middle Trap
Scott Belsky calls the period between launch and PMF "the Messy Middle." Most founders spend 18–36 months there. Most fail.
Power users let you feel like you're winning in the Messy Middle. You have proof points. You have a narrative. You can raise a Series A on vibes and engagement.
But feeling like you're winning extends the Messy Middle. You optimize for power users. You avoid the hard work of understanding why the median user leaves. You build on top of a shaky foundation.
YC's advice: find 100 people who truly love your product, then find 1,000 more like them. Not 100 power users and 990 ghost cohorts.
Power User Masking: Two Cohorts, One Narrative
[A]The power users sustain the narrative. The median user leaves before day 7. Most founders read only the top line.
The test: can you describe your power users in a single sentence? "Design teams with 5+ members using real-time collaboration." If yes, you have a segment. Can you find 900 more of them? If you can't, you don't have PMF. You have a feature.
What to Do Monday
1. Segment your DAU by cohort age. Don't look at overall engagement. Look at Month 1, Month 2, Month 3 retention separately. Where does it flatten? That's your real problem.
2. Interview the bottom 50%. Not your power users. The people who signed up and left. Ask them: what did you come for? When did you stop coming? Don't ask them to explain churn. Ask them what you missed.
3. Calculate power-user lifespan vs. median lifespan. If power users stay 8 months and median users stay 2 weeks, your product is not scaling. It's consolidating around a micromarket.
4. Set a retention threshold. Define PMF as: 30-day retention above 25% AND new-cohort retention stable (not declining) month-over-month. If you don't hit both, you don't have fit. You have power users.
Power users are not a bad problem to have. But they are a problem if you mistake them for proof. They are a ceiling, not a foundation.