90 minutes of unfiltered product advice from Snap and Discord’s product chief | Peter Sellis

What It Covers

Peter Sellis, first PM at Snapchat (7 years) and former Head of Product at Discord, shares unfiltered lessons on team design, why the median PM is net negative, Snapchat's structural monetization constraints despite one billion monthly active users, and how focusing on core users drove Discord's fastest post-pandemic growth.

Key Questions Answered

• •Team Design — Autonomy + Ideology: Structure product teams around two elements borrowed from decentralized organizations: a near-religious shared strategy that every member internalizes, and an explicit map of who owns which decisions. This eliminates over-collaboration, which forces teams to move at the speed of the slowest node. When strategy functions as ideology, individuals make founder-quality decisions without requiring constant coordination, keeping execution velocity high even as headcount scales.
• •PM Quality Distribution: The median PM is likely net negative due to a mathematical reality: genuinely skilled PMs exit into founder or executive roles, while underperformers have strong financial incentives to stay. During the zero-interest-rate era, PM became the highest-paying non-technical career in tech, attracting volume over quality. The result is a power-law distribution where most practitioners sit far below the mean, making the profession look worse than its top decile actually performs.
• •Snap's Monetization Ceiling — Three Structural Constraints: Snap's ad business faces three compounding disadvantages: its user base skews under-18, where tracking restrictions sharply reduce CPMs; the product opens to a camera rather than a feed, eliminating natural ad placement; and its core function is messaging, historically the weakest ad environment online. Despite these constraints, Snap reached $4 billion in ad revenue faster than Twitter, Reddit, and Pinterest — all of which launched earlier with more ad-friendly formats.
• •Invest in Top Performers, Not Struggling Ones: Counter to the instinct to rescue drowning team members, Sellis deliberately loads high performers with increasing responsibility until they begin making mistakes, then calibrates from there. Struggling performers are largely left to self-correct. The rationale: strong performers in a flow state can absorb more scope than managers typically assign, and the compounding output from one exceptional person outweighs the marginal gains from coaching someone at the bottom of the distribution.
• •Growth Lives in the Core, Not Expansion: Both at Snap and Discord, the highest-growth periods came from improving the experience for existing daily users rather than acquiring new cohorts. At Discord, focusing exclusively on intentional multiplayer gamers — a segment already at perceived 100% penetration — produced the fastest growth since the pandemic by moving users from 10 active days per month to 11–13. Moving DAU/MAU ratio by even a few basis points outperforms new-user acquisition campaigns in compounding impact.
• •Saying No as a Taste Muscle: Sellis observed Evan Spiegel personally reject higher-quality ideas than most consumer startups ever ship. The discipline of restraint — holding a prototype, believing in it, and choosing not to release it — is one of the few exercisable taste muscles in product. A useful interview signal: ask candidates about high-quality ideas they built but never shipped. A curator who hangs the entire collection has made no curatorial decisions; the same applies to product leaders.
• •Core Product Value as Operational Language: Define the product's daily reason-for-use in a single phrase that maps directly to metrics — Snap's was "fastest way to share a moment with your best friends," Discord's was "best way to talk and hang out before, during, and after playing games." Each word in the phrase corresponds to a measurable lever. Sellis delivered this framing to every new hire for seven consecutive years at Snap, embedding it as the shared language data scientists, engineers, and PMs used to evaluate roadmap decisions.