Peter Sellis, former Snap first PM and Discord product head, argues the median product manager is probably bad and the best ones leave for founding roles or executive jobs.
Sellis says Snap is hard to monetize because its very young user base, under-17 ad targeting limits, and camera-first app leave no natural place for ads.
Sellis says OpenAI is speed running the components of an ads marketplace and that trust is the key constraint for ads in chat.
Former Snap and Discord product chief on why collaboration slows companies
Peter Sellis, drawing on Snap and Discord and speaking free of any comms team, argues that product success comes from structure and focus. A clear mission, explicit decision rights and strong people given real authority let a company move faster than collaboration ever will. Snap shows that even a hugely successful product can hit hard limits on monetization, and Discord shows that growth comes from deepening the core.
Takeaways
Great product teams pair shared mission with clear decision rights. Peter likens the best teams to a terrorist organization. A shared ideological mission holds people together, and a clear structure of who can decide what keeps them fast. Founders' visions get compressed into short phrases the whole team repeats.
Collaboration is overrated because coordination is costly. Coordination costs mean a company moves only as fast as its slowest part. So PMs at high growth companies should make decisions founders once would have made, and strong performers should get more responsibility and public defense.
The median product manager is probably bad. Skill follows a power curve. The best PMs leave to found companies or take executive roles, while weaker ones stay put. Peter thinks the profession is improving as interest rates rise, and great engineers rarely complain about PMs because they bend them or avoid them.
Snap's audience and camera-first design resist monetization. Users are very young, ad targeting is limited for under-17s, a camera-first app has no natural ad slot, and messaging apps lack good ad products. Camera monetization was one of Peter's biggest failures. Snap still reached four billion dollars in ads revenue faster than Twitter, Reddit or Pinterest.
Growth comes from the core user, not distant niches. At Discord, the team focused on gaming, the biggest niche, and on the core value of talking with friends around games. Retention work moved users from about 10 days a month to 11 to 13. Bets beyond the core should draw on its strengths.
AI ads need trust, and cutting features gains value. Peter says OpenAI is speed running the parts of an ads marketplace, and a modified VCG auction suits chat, with trust as the key constraint. He was disappointed the top tier went ad-free. As AI lowers the bar for creation, cutting features becomes a scarce skill.
Long-term greed and back-casting beat short-term optimization. In ads, tomorrow's revenue depends on the return advertisers get today, so long-term greed is the right stance. Peter favors back-casting from an ambitious future, as SpaceX does by retiring Falcon 9 for Starship and publishing roadmaps.
Tap a name to go there · tap a line to see why they are linked