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

Lenny's Podcast: Product | Career | Growth

Peter Sellis, former product leader at Snapchat and Discord, shares unfiltered, contrarian lessons on product management and team design. He

Key takeaways

• Peter designs teams like 'terrorist organizations,' prioritizing ideological alignment and decentralized decision-making to reduce coordination costs.
• The median PM is often net-negative because high-performing PMs exit the profession to become founders or executives, skewing the talent distribution.

Main topics

• Team Design Philosophies
• Product Management Talent Distribution

Notable quotes

If a museum has all of their collection on the walls, then the curator hasn't done anything.

Conclusion

An insightful deep dive into the structural and philosophical elements

Transcript preview

Speaker 1 (0:00) I asked you what are some counterintuitive lessons that you've learned that go against common startup Speaker 2 (0:05) wisdom. I'm no longer beholden to any corporate comms team right now, so I can let it rip a little bit. Do you like to design teams like a terrorist organization? A good terrorist organization essentially has two things. One is it has this ideological culture that everyone understands why they are doing what they're doing. And then it has a very, very clear organizational structure of who can be trusted with which decision. You Speaker 1 (0:29) like to spend time writing your best people into the ground versus... Improving the average people. If Speaker 2 (0:34) you see somebody drowning, you want to save them. I did the absolute opposite. If somebody is strong and crushing it, I'm just handing them more and more rope to hang themselves with. Speaker 1 (0:43) Something that I know you believe most product managers are bad and are often net negative to the company. Speaker 2 (0:50) The great product managers are exiting the system. They're becoming founders. They're becoming executives. It's kind of set up to be this profession where... The median product manager is actually probably pretty bad. Speaker 1 (1:00) Snap is quite famous for not hiring PMs for a long time. The baseline is we should Speaker 2 (1:04) just be able to operate this product without any product manager. So this person has to provide value over nobody. Speaker 1 (1:10) Feels like an interesting new skill for PMs is pushing back on things AI is adding to your products. I think I saw Evan personally say no to Speaker 2 (1:17) better ideas than I've seen almost any other consumer startup. launch, exercising that kind of muscle of saying no, of restraint, is probably one of the few muscles of taste you can exercise. If a museum has all of their collection on the walls, then the curator hasn't done anything. Speaker 1 (1:33) Today, my guest is Peter Sellis. Peter is a legend in the PM community. He was the first PM at Snapchat, where he spent seven years building one of the most delightful and popular consumer products in history. More recently, he was head of product at Discord, which he helped grow significantly. And now he's between gigs, which means he was finally able to come on the podcast without a comps team, without a PR team, which means that he's able to be especially honest and real about everything that he's seen and learned and has been excited to share. We cover so much ground from how the best teams are organized like terrorist organizations, what it's like to manage Nikita Beer, why most PMs are really bad, why Snapchat has struggled to build a truly massive business in spite of having a billion multi-active users, and it's not what you think, and so much more. This episode is full of so much gold, and that's why it went so long. I couldn't stop digging into stuff that Peter kept bringing up. A huge thank you to Jack Brody and Sri Ram Krishnan for suggesting topics and questions for this conversation. Before we get into it, don't forget to check out Lenny'sProductPass.com for a free year of the hottest and most beautifully crafted AI and non-AI products in the world, available exclusively to Lenny's Newsletter subscribers. With that, I bring you Peter Sellis. Speaker 1 (2:53) Peter, thank you so much for being here. Welcome to the podcast. Speaker 2 (2:57) Thank you, Lenny. It feels like it's been a long time coming that you and I have known each other, but I'm really excited to be here. Speaker 1 (3:04) You've never done a podcast before. You've been in product. You've been out there for so long. I'm very honored you chose to do this. Why have you avoided podcasts? Why have you decided to finally do this podcast? Speaker 2 (3:15) You know, it's not avoiding podcasts. I've really just been avoiding you. No, I'm kidding. It's been, you know, I always, I think there are two things. To be totally frank with you, a lot of times when I saw, especially early on, because you have such wonderful distribution among so many influential people and teams. When I saw somebody going on Lenny's podcast, I was like, oh, that person's probably about to get fired. They're going on and they're pimping themselves out for like their next role, which was like maybe one view of it. I was like, why aren't they just grinding? Now I think when I kind of take a step back. you know, a lot of the people I've worked with, a lot of teams I've had kind of the opportunity to lead. I think one of the things that I generally do is show up pretty authentically every day at work. You can call it no filter, you can call it whatever you want. And a lot of the stuff that I say and do, I think it's probably better suited for behind closed doors. But as you and I have chatted about now, I'm between different roles. So I'm no longer beholden to any corporate comms team right now. Yeah, I guess I can let it rip a Speaker 1 (4:20) little bit. Okay, amazing. These are the best conversations. No PR people involved, no comps people. Speaker 2 (4:25) Yeah, nothing to promote either. I have no book coming, no other podcasts, no collaboration. Yeah. Speaker 1 (4:29) Wow, just pure alpha. Speaker 2 (4:32) Just pure, yes. About to drop some pure alpha, baby. Speaker 1 (4:35) This episode is brought to you by our season's presenting sponsor, WorkOS. What do OpenAI, Anthropic, Cursor, Replit, Sierra, Clay... and hundreds of other winning companies all have in common? They are all powered by WorkOS. If you're building a product for the enterprise, you've felt the pain of integrating single sign-on, skim, RBAC, audit logs, and other features required by large companies. WorkOS turns those deal blockers into drop-in APIs with a modern developer platform built specifically for B2B SaaS. Literally every startup that I'm an investor in that starts to expand upmarket ends up working with WorkOS. And that's because they are the best. Whether you are a seed stage startup trying to land your first enterprise customer or a unicorn expanding globally, WorkOS is the fastest path to becoming enterprise ready and unblocking growth. It's essentially Stripe for enterprise features. Visit WorkOS.com to get started or just hit up their Slack where they have actual engineers waiting to answer your questions. WorkOS allows you to build faster with delightful APIs, comprehensive docs, and a smooth developer experience. Go to WorkOS.com to make your app enterprise ready today. The first thing I want to talk about, something that stood out in our jamming on what to talk about is the way you think about designing teams. The way you described it to me is you like to design teams like a terrorist organization. Speaker 2 (5:58) Yeah, yeah. I should probably workshop this one a little bit. Like this is probably an example of the one that's better for behind closed doors. But yeah, when I actually talked to the chief people officer at Discord about this, and I think she was like simultaneously horrified. and intrigued in a way, which credit to her for not judging me right off the bat. But yeah, I think it comes from essentially like two primary beliefs from me. One is that there's actually like a hidden cost to collaborating that no one really internalizes extremely well, meaning like you'll never hear really any executive at any company be like, stop collaborating. And to me, that's kind of like a warning sign. Collaboration comes with incredible coordination costs and coordination costs essentially assure that you're moving as slow as the slowest node in the system. And so I try and like aggressively internalize the cost of collaboration sometimes. And then I think the second thing separately from this is that At high growth companies, especially in so many of the people that, you know, listen to your podcast, I think aspire to be at these types of companies. Product managers should be in a position where they're making decisions that the founders would have made, like sometimes like six months ago, definitely like a year or two ago, these are decisions that the founders themselves would have made. And so like, set aside violence for a moment, like where I'm not saying that, you know, I designed my teams for violence. But a good like terrorist organization essentially has two things. One is it has this like ideological culture that everyone understands why they are doing what they're doing. And then it has a very, very clear organizational structure of who can be trusted with which decision. And I think like. creating a culture where there's a very clear, like ideological, almost like religious level goal, essentially that the strategy is almost like a religion to the entire team. And an organizational structure where people understand who are the directly responsible individuals who single threaded on something, who's the throat to choke, we use whatever MBA, blah, blah, blah ism you want. I think the combination of those two things can be leveraged in a way such that you don't need to over collaborate. And you can be assured of achieving the results that you wanted as if you had collaborated. And so like, I think the components of what is this kind of like great ideological setup, a lot of things I focus on, for example, because I think oftentimes the head of product at high growth companies where the founders are still in charge, the head of product is really kind of like a vessel for their vision. both with Snap and at Discord, where I had the privilege of working with the founders in both cases, I tried to take and absorb and internalize their vision and then create like small phrases and language that could be used and repeated that were essentially shortcuts or almost like synecdoches of like the whole strategy. And so, yeah, I could go more into that, but that's kind of how I think of it. And obviously giving it like branding that, you know, I'm designing it like a terrorist organization at least gets people's attention. But maybe I need to workshop that a little bit. Speaker 1 (9:30) No, you don't. This is perfect. This is exactly what I was hoping this conversation would be. So just to be very clear about the elements of the terrorist organization that we want to keep. One is this idea of autonomy, essentially kind of trusting people further down in case the leaders get caught or are incommunicative. So basically giving more autonomy to folks further down the chain. And then two is having a very clear mission. ideology, you might say, so that everyone's very aligned and decisions are quick and that kind of thing. Speaker 2 (10:01) Yeah, yeah. And to be clear, I didn't start with, I'm going to build a terrorist organization. I bet you'd build a great one, Speaker 1 (10:10) by the way. Speaker 2 (10:11) That would be a great one. But when I thought about the elements of what I wanted from the organizations I was building, I couldn't help but notice that there were organizations out there operating like this, and they just so happened to be terrorist organizations. So, yeah, that led to a lot of interesting reading lists. I think that the Peter Salas product reading list is like a, it's almost like a banned book list at this point. Speaker 1 (10:33) It's interesting because I've heard many people describe you want to build a cult. And this is not that different because, you know, cults are very effective at getting stuff done. It is not. Speaker 2 (10:43) Yeah, you can't spell culture without cult. And so I think, yeah, it is not very different from that. I think there are elements of a cult. that are more about the person leading the cult, whereas terrorist organizations are more about a higher power or like kind of like religion or ideology. So that's why I tended more towards that. But yeah, I think you could read up on a couple of cults and then just take it all the way to the penultimate moment, like before they do the thing that makes them well known as a cult. And you probably could learn a lot, honestly. There's Speaker 1 (11:18) a Speaker 2 (11:18) lot of alpha in that. Speaker 1 (11:19) Speaking of... a higher power. A fun fact about you is you managed Nikita Beer. Speaker 1 (11:28) who came on the podcast, one of the most popular episodes of the podcast, quite a character, was head of product for X recently for, I think, a year. Very few people have worked with Nikita. I'm curious what it was like to actually manage Nikita, work with Nikita. Speaker 2 (11:41) Yeah, Nikita was on my team at Discord. We acquired his startup, and he was on the team at Discord for over a year with some phenomenal engineers that came along with him. I think managing Nikita, that's a very loose term of use of the word manage. I think much like, yeah, you don't manage Nikita. You only like seek to control him or like direct him. I liken it a lot to like, I mean, Nikita is an awesome personality and just actual operator in the product space, like pretty much one of one. And I have a ton of respect for a lot of the stuff that he does. I think of him a lot like Liam Neeson in like the Taken movie series. Like, yeah, if your daughter gets kidnapped, like he has a very special set of skills and he's going to employ those skills. And those skills are very hard to come by and very useful in certain situations. But then when you like need him to like raise the daughter or like, you know, buy some groceries or stuff like that, all he knows how to do is just like kill everybody in the grocery store or something like that. And so now he's like, yeah, incredible person to work with when you have him on kind of like the areas that he is an expert in and truly like one of one for that from that perspective. How Speaker 1 (12:57) do you feel about having folks like that on your team? I mean, a lot of people talk about having very distinct, you know, I don't know. roles for different different people for different roles and sometimes you need someone like that just drive i think Speaker 2 (13:08) it's like essential to be able to create an organization that can adopt that type of talent i i Speaker 1 (13:12) i Speaker 2 (13:12) wouldn't i view it as like a failure state if you have kind of like product orgs that can't um essentially handle really spiky talent i mean nikita took it to extremes like you remember when he was on intro um you know doing that things at one point i like was trying to get a one-on-one with him who by the way he reported up into me And I couldn't get him. So I was like, dude, if you don't fucking answer my DMs, I'm going to go on intro and book you so I can just have a one-on-one with you right now. Luckily, we didn't get to that. He answered. I think I hit him up on like four different messaging apps before I got a hold of him. But yeah, I think this ability to handle spikiness, to build teams that can essentially adopt the type of talent that isn't necessarily like the perfect nine to five employee from a product perspective. is, you know, I look up a lot from like a management style to Phil Jackson, the coach of the Bulls. And Lakers. The Lakers. Like, I think his ability to mesh talent of very different types that literally does not get along otherwise to create these winning teams, to have done it both as a player and as a coach in two different places. to use philosophies from different, from Eastern philosophies, Western philosophies to combine all this stuff. Like, um, I, I, I think he's like one of the most meaningful leaders, uh, from a leadership styles perspective of the last half century. Um, and I think like the thing about product, like especially when it comes to consumer and anything that involves more creativity and taste is I think of a lot of product leaders, more like bands putting out albums and like, you can like, love a band, but sometimes they have a dud album. And what you do with the band as a fan during that time, like, do you stay a fan? Do you still listen to the album? Like, do you ride it through? Do you do you just listen to the old stuff? Like, that's kind of how I think of a lot of the spikier product managers that I've worked with, which is like, man, sometimes they whiff on an album, but if I still believe in them as an artist, like, I'll stick with them. Speaker 1 (15:17) Something that kind of coming back to Nikita, but going in a different direction. When he came on the podcast, he had this line that he's like, I'm honored to be on a podcast of our product management as someone that doesn't believe product management is real. And it's funny that he went on to be head of product at X. But this connects to something that I know you believe, which is that most product managers are bad. and are often net negative to the company. Talk about that and just like why you think the kind of the median PM is not actually good. Speaker 2 (15:47) I feel like this might have been a contrarian take during like the Zerp era. But now it's like kind of, I think it's kind of bleeding into the mainstream of like holding a higher bar for product management. And it is funny, like you have this like kind of, because you had Tom, Tom, the CPO of whatnot as well. um, say something similar. And I think I agreed with a lot of the stuff he said. And I think, by the way, I think some of his most senior PMs are actually former, uh, snap PMs from, from my team, which is great. Uh, cause it kind of like, I think, uh, verifies a lot of this, but yeah, now we're a bunch of self-hating PMs who've like, uh, made a bunch of money being PMs are now like, Oh, three of them are bad. Um, but no, I think there's, there's actually like kind of a mathematical explanation for, for this. Um, First of all, I think any craft that is, especially when you're at the top end of any kind of skill that is normally distributed, you're at the tail end of the distribution. And so for product management, if product management is normally, the skill of product management is normally distributed across the population, the ones who are actually product managers are all the way on the right-hand side of the distribution. So you lop that off and it becomes a power distribution. And for any power distribution, the median is going to be far below the average. Most people are on the left side of it. And I think this actually gets excessively emphasized within product management for two reasons. One, especially during the Zerp era, it became essentially the most lucrative career in tech for non-technical people. So once it has a very subjective definition, right, which is part of the reason why your podcast has been so valuable to people over the years, because they're like, what do I need to do to be one? And so once you're in it, you're being compensated essentially as if you're on a technical ladder. So you have kind of a very high incentive to remain part of it. If you're very good at it, the skills you develop as a product leader are almost perfect for either other executive roles, just like overall leadership roles, or starting your own company. So you're on the right-hand side of the distribution, like the great product managers are exiting the system. They're becoming founders, they're becoming executives, etc. And on the other side of the distribution, anybody who's in is doing anything possible to stay in. So the distribution is, it just becomes more weighted, weighted, weighted like this. And I think during Zerp, like this, you saw this, and there were memes about it and stuff like that. And maybe there's been some correction. But yeah, I think mathematically, it's kind of like, set up to be this profession where the median product manager is actually probably... pretty bad. Speaker 1 (18:28) Do you feel like it's getting better? And do you think there's something I think about a lot that a lot of people say that like PMs suck, they're useless. And I find that that's usually the case because they've just not worked with a great PM. And a lot of people haven't worked with great PMs. And so it kind of like makes the whole profession feel, look bad and sound bad when a lot of people have, like, as you describe, are working with these kind of mediocre PMs. Thoughts there? Speaker 2 (18:59) Yeah, I think you're exactly right. I've also like kind of noticed this. I feel like so much in the world right now is explained by like some form of like the midwhip meme in some way or another. But yeah, I think you're exactly right. Like that is a... that is a second order effect of this kind of distribution that I think governs the whole profession. I do think it's getting better because I think, again, like when a lot of people reference like the Zerp era, but I truly feel like when interest rates go down, that means opportunity costs are lower. And so when opportunity costs are lower, people are doing things like just hiring that extra person, like doing stuff like that. But these opportunity costs, I think, are higher than are internalized because of some of the stuff we started talking about, that a lot of these costs are not internalized. And so like I've been, and we can talk about this, but like I've been really focused on developing a system for evaluating product managers, not just the person, but the person and the place that they are at that time to make sure. uh that the team is actually just like net positive and in my experience and we can talk about what some of those methods are but like in my experience like the way this kind of ends up is that i really don't hear great engineers complain about product managers that much because great engineers a they know how to bend product managers to their will they get the the use out of them that they want and b they just like don't put up with bad ones so they're they like they don't get assigned or they don't get to work with bad product managers i think ironically in kind of a mid-week wean or the horseshoe theory of things. Really bad engineers also don't complain about product managers because they know they can just like hide behind their decisions and stuff like that. So it's really just the ones in the middle that are complaining. Speaker 1 (20:49) I do want to ask about Snap. I feel like it's a company that doesn't feel like it lived up to the potential of what... I think Evan and everyone imagined it could be. I imagine the main reason from, and I had Evan on the podcast, is just the users that it's going after are just hard to monetize. Is that kind of the core of what has kept Snap from becoming a bigger success or is there more to it? Speaker 2 (21:14) I'll try and be as objective as possible, Lenny, because this is a little bit of my life's work and maybe I do have some level of chip on my shoulder. Yeah, I'll try and be as objective as possible. So I'll say... Speaker 2 (21:28) two things, maybe three things about Snap from a monetization perspective, at least an ads monetization perspective. And then we could talk about like, they've had a lot of success with Snapchat Plus and subscription products in general are doing well. But ads are truly the magical kind of like product market fit of consumer monetization on the internet. And there's a lot of mathematical reasons why that's true. And there are three things that hold back Snap from an ads perspective that I think the company, like we still kind of talk about as we. deserve a little bit more credit maybe than we get, but I understand why we don't. The first is, yes, the user base is extremely young. Like, Snap appealed to Gen Z at an excessive rate early on. Now even, I think, Gen Alpha as they, like, kind of graduate from, like, Roblox and stuff like that. And there is a step change, like, literally a suspicious discontinuity in advertising CPNs between 17 and 18 and between 18 and 21 in the United States and abroad. If you're under 17, the rules around how you can track and measure and target ads are extremely, extremely constraining and justifiably so, right, to protect teenagers. That means that it's essentially much harder to drive ROI from this group of people versus an 18-year-old versus a 21-year-old and 24-year-old. And you can say all you want about like, oh, you know, teenagers influence their parents, blah, blah, blah. The key thing to think about is like those are just harder to measure. And it's harder to build performance products around things that are hard to measure. And so, yeah, the snaps audience is definitively like less valuable from that perspective versus like a Pinterest or an Instagram or a Twitter. The second thing is almost every ads business, every ad product you'll see that isn't search based, every display ad product opens to the screen where you put the ads. There's a lot of studies around attention and just mathematically where you're going to get the impressions where you need to kind of like maximize that for advertising and snap opens the camera. And that makes it awesome as a product, right? Like in terms of its ability to be the fastest way to capture a moment and share it with your friends. But in no part of that flow, the core flow of taking a snap and sending it to your friends, is there like a logical place for an advertisement? It is just not a feed based product. And it is not a consumption-oriented product. It is a creation-oriented product, a visual creation-oriented product. And there don't really exist a lot of