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

Peter Salis, first PM at Snapchat and later product leader at Discord, is between roles: no comms team, no PR filter, permission to let it rip. I wanted the counterintuitive lessons behind his teams, PMs, Snap, AI ads, and taste.

I haven’t been ducking podcasts; mostly I’ve been ducking you. Kidding. I used to assume guests were marketing themselves for the next job. But I show up with little filter, and I’ve got no book, partnership, or anything to pitch. So we can be more honest.

Your ideal organization gets a deliberately wild comparison: a terrorist organization. Strip out the violence, and you mean autonomy deep in the chain plus a very clear shared mission.

Collaboration has a coordination tax, and nobody in leadership says collaborate less. Fast-growing companies need PMs making calls founders would have made months ago. I want people to know the ideology—why we exist—and who can make which decisions without a committee. At Snap and Discord, I turned founder vision into small, repeatable phrases: keyboard shortcuts for strategy. A cult is close, except I care about mission alignment, not the person at the center. Also, yes, my reading list got alarming.

That still has to leave room for high-variance people. You managed Nikita Bier after Discord acquired his company. What does that take?

“Manage” is generous. You don’t manage Nikita; you aim him. He’s Liam Neeson in Taken: for problems requiring his particular skills, almost nobody is better. Ask for the ordinary grocery-store part of the job and, metaphorically, everybody gets taken out. A product org that can’t absorb spiky people is failing. Phil Jackson had it right: the job is combining talent that may not naturally coexist. Don’t demand a perfect nine-to-five shape from exceptional people.

You argue that the typical PM can be net negative, though great PMs are enormously valuable. Why?

The best PMs often leave the function to become founders or broader executives. During cheap money, product became a lucrative route for nontechnical people, with fuzzy entry criteria and every incentive to stay. The middle got weaker than people expect. Great engineers either get value from a PM or refuse to tolerate a bad one.

Snap reached roughly a billion monthly users, yet people expected the next Facebook. Was its young audience the main advertising constraint?

Partly. Teen advertising is harder to target, measure, and prove, with sharp value changes around adulthood. But Snap opens to the camera, not an ad-friendly feed. It’s creation and messaging, neither naturally producing search- or feed-like inventory and intent. We built ads quickly despite that, but camera monetization was one of my real failures. Still, split Snap apart and it’s absurdly strong: leading messenger, camera, map, even astrology product for young people. Snapchat Plus might have eased ad pressure if it arrived earlier; utility subscriptions compound slowly because people need the habit before payment makes sense.

You’re skeptical of vague systems thinking. What makes it real?

A real systems view is modelable. Put stocks, flows, levers, and distributions in a spreadsheet, even if nobody sees it. Asking why repeatedly can help, but the useful version describes how the machine behaves mathematically and which assumptions change the result.

Ads remain the internet’s core consumer monetization fit. OpenAI seems to be learning the marketplace playbook quickly. The question is whether its auction protects trust: when I ask which mic to buy, how much commercial influence can enter before the answer stops feeling reliable? If search becomes sixteen sponsored results before the useful thing, users notice.

That leads to taste. Evan Spiegel rejected ideas that might have beaten what other consumer companies shipped. What did that teach you?

Taste is restraint. A museum curator earns their keep by keeping most of the collection off the wall. In interviews, I want strong work someone built, held, and chose not to release—not just experiments that lost a test. AI lets more people make things, so the bar rises for work deserving human attention. Maybe, hilariously, the answer is more PMs saying no.

Your next contrarian lesson is investing in the best people until they reach their limits, rather than repairing average performance.

With people already making good calls, I give more scope—more rope, ugly version—until I find the breaking point. I defend my team publicly, so I want strong people to know I trust them with bigger decisions. Don’t confuse my lack of attention with failure; often I know they’ve got it.

At Snap and Discord, growth came from the core more often than shiny adjacency. Daily users are dramatically more valuable than monthly ones. Improving Android speed helped restart Snap after the redesign stalled growth. At Discord, making multiplayer gaming with friends better increased use despite assumptions gaming was saturated. There’s always money in the banana stand: make the core experience ridiculously good, then serve nearby groups or remove friction from unexpected use cases. But refinement isn’t immunity; something ten times better can still wipe you out.

SpaceX is the anti-complacency example: Falcon 9 is extraordinary, yet Starship can make its champion obsolete because the mission demands it. That is changing direction and acceleration, not just wanting a higher number. For ads, we called it long-term greedy: advertiser returns today create durable revenue tomorrow, even if this quarter suffers. Consumer products need both a memorable core-value phrase and measurable retention signals.

The great PM is an oxymoron three times: ego to drive direction and humility to give away credit; process discipline and comfort abandoning it in ambiguity; long-range vision and an irritating bias toward action today. My softer contrarian take: your partner may be your most consequential career decision. I’m also obsessed with time; empty stretches vanish. And yes, I led product for a twenty-four-seven Shiba Inu puppy stream. It soothed people.

I’m grateful to say some of this out loud. If it resonates, if you think I’m wrong, or you’ve got beef, I genuinely want to hear it. Disagreement is usually where the interesting stuff is.