The Untold Story of Discord: "Pivots and F*ck Ups..."

Yeah. So, fuckups, huh? Well, no, no, you're doing pivoting. I was doing... Oh, you were doing the fuck up. Exactly. It's... Well, see, the thing about pivots is they feel like fuckups actually. Like people talk about pivoting like, 'Oh, you know, yeah, maybe you'll pivot.' It's not fun at all to go through those kind of experiences because the lived reality of going through a pivot is that you've just spent some amount of time pouring your heart into building some product and it's not working and coming to the realization of that and then thinking about like now what. It actually emotionally was very hard. If you have clear eyes and you see what's actually happening around you though, you can learn a lot from failure and think about how do you know what do you do from there? So while we built this game and launched it in 2014 called Fates Forever, it was kind of like a competitive multiplayer game on iPad, three versus three team based game. For a variety of reasons it didn't work. But the plan for the company from the original sort of founding when I started it was to actually build a chat app for gaming and the thought was that we would bootstrap the social network through a game and not actually go to market directly with the chat app. So, we built the game and we had voice and text chat built into it, team play. We actually built a custom forum that was integrated with the game that had a lot of community activity. And when the game didn't work, what we realized was a couple things. One is that we noticed some things about how people were organizing around our game and what apps they were using and how they were cobbling together mass market communication apps to talk with their friends and maintain their guilds across game sessions. And then that made us notice our own behavior using things like Skype and TeamSpeak at the time and how we played games ourselves. We're thinking about like okay what do we do next? Do we try to make a game again or do we... should we... is there a path to go directly to market with a communications app? And ultimately we saw a path to just launch a chat app directly that we didn't have to bootstrap it with a game and that became Discord. So it was a pivot but it was within the umbrella of our search space.

It was obvious that Fates Forever was not going to become a big enough game to bootstrap the network because we could look at the retention numbers and the growth rates of it. While we had a loyal and really engaged superfan group in there, we had the 10,000 true fans kind of vibe going, but it was pretty clear that it was not a straight shot to get to the next step of the master plan. We were actually working on an iteration of Fates Forever to bring it to smartphones and simplify it a little bit when one of my teammates started to have the thought about just launching a chat app, but we were debating these different paths. I just had this sense the game could possibly work, but we still have to get the game to work, which is hard, and then bootstrap the network. And there was actually a path to go to market with the chat app. And as a user of those products, I was just like, I feel like this is probably gonna work. And I guess it did. Sometimes it's just that easy. I mean, it wasn't a straight line either.

No, it did happen in parallel for a hot minute. We spent a year after I realized that Fates Forever in its form wasn't going to make it until we fully committed to Discord. After we realized the game wasn't going to make it, which happened after trying two or three different things of like, oh, maybe we can make it work. And I just kind of realized like, we've got more ideas, but they don't feel like the kind of thing that's going to dramatically change the fate of the title. Nine months through that process, we spun up a small team that was working on Discord. That was early 2015. What ended up happening for us was as we were working on it, we were an 18 person startup at the time. And so I had an engineer working on Discord and then I joined it. I'm an engineer and then I put the third engineer and then the fourth engineer and then within 60 days we got to a place where all the engineers were building this chat app and all my designers, the artists were making a game in Unreal and I was like this is stupid. We're a Series A startup. We can't be making two crazy bets, right? I was like, I have to pick one or we're just gonna fail. So, I was like, let's pick Discord. But at the time, Discord had 5 DAU, right? It was all us. And my friends were like, 'This is cool, but they didn't use it because it kept crashing and it was breaking.' But I could tell it was something in it. And after we burned the boats and went full Discord, four weeks later, we figured out how to make it work. So the focus of leaning into it was a big deal.

I just tend to default to being honest with people. If you're joining an early stage startup, it's because you're self-selecting into the mess a little bit. And so I always felt that being honest with people and bringing them along the journey and involving them in the decisions and getting their input was the best way to get the best ideas and also to get the most buy-in. Now with that said, I think half of that group of 18 people thought that Discord was a horrible idea. Literally they were like, 'This is the dumbest thing ever. TeamSpeak is amazing. It's never going to work.' And the day that I decided we're going to work on Discord, the next morning, one of the employees quit through a Post-it note. Literally, I came in and at his desk was his MacBook closed and a blue Post-it note. And I'm like, you know, where's Pete? I forget his name. And I go over to his desk. He knows his name, let's be honest. I got the whole story. How dumb does that guy feel? I don't know. I haven't asked him since then. But last time I talked to him was through a Post-it note stuck to his desk which is like 'I quit. I wanted to make video games.' And I was like, huh, all right. That sucks. And I felt offended. I was hurt. I was like, oh man, he doesn't trust me, doesn't believe. It was one of those founder moments where I was like, fuck. But now, I suppose maybe he's like, maybe I should have stayed or maybe he's thrilled. Maybe he worked on some great games and he's super happy. Whatever. So it was challenging. We did get to a point pretty quickly where I realized we had a bunch of artists. They weren't going to help. So, I had to lay off a third of the company, which was really hard. And these were all amazing skilled professionals. So, they were grown-ups about it, but it was emotionally one of the hardest things I've ever done over the whole 13 years of the company. And then we had a smaller team working on it, but even then, still half the group was like, 'This is not going to work.' And there's one guy who's still at the company who I also won't name who I know was looking for a job and told me as much. And then things started to work and he was like, 'Okay, I'll stay.' I don't know what the lesson is from that. Maybe it's hard. People are going to quit. It's going to be difficult. But if you have conviction in yourself and trust, faith in yourself, you push through and ultimately winning gets people to stay and come.

One of the things that we benefited from, which is why I think even though some people were looking for jobs and didn't believe in it, people still stayed, is because we were in this interesting cross-section where we were working in the gaming space, but not making video games. And some of the people who were passionate about games, but maybe not qualified to work on video games or didn't want to take the risk of working on video games, would choose to work at Discord. So we had pretty unique mission differentiation. I actually think to this day that's why someone would select to come work here. Of course the company's doing well, but the people who we love to meet in our interview process are people who select to come work at Discord because they love the mission of the company. Those people tend to stay a really long time because if you want to work in tech and you like games, there's not actually a lot of places to go. But if you just want to go work at the company with the latest up and to the right chart, then you have a million choices and we see higher turnover for those folks. We're big enough now to have values. I wouldn't spend time making a values chart. Their values are just like what the four of you are doing. At some point, you got to write it down. But you sort of need to reverse engineer it at some point. Anyway, our first one, we haven't called it this because it's a little bit spicy, but we talk about it as giving a fuck. It's about our customers and the problems. So if you fall in love with our customers and our mission and why we're doing what we're doing, I find that people show up with more energy. They are willing to work harder and they act like owners. Nothing is someone else's problem. And they tend to stay longer.

There were a couple of them. The products that people had chosen to use like TeamSpeak and there was a whole category of these channel-based communications apps that were made for gaming like Ventrilo and TeamSpeak. They sort of have the Discord Slack channel vibe that comes from that tech tree of products. They were never very mass market but they were adopted pretty heavily by core gamers. And then the other pillar was Skype, the mass market communication stuff. In 2015, some of the mobile stuff was starting to become used but none of the mobile products were really great on PC and a lot of multiplayer game happens on PC. The insight was realizing that both of these products were not serving the gaming market effectively in a variety of specific features that when you add them all up there's a 10x gap in quality. I'll just rattle a couple off. With TeamSpeak, oftentimes you have to pay for it. You go to some cloud provider on some janky website, give them your credit card and pay money to have slots for people to hang out on your server. Then when you want them to join, you have to send them an IP address, sometimes a password. Before they could even use it, they have to go download an application to their computer and they're not compatible with each other. So if you have your group on TeamSpeak and you're short one person but you're doing a raid in WoW and you happen to meet someone in game and they have Ventrilo, they have to go download another desktop app. It's a pain. And most of those apps didn't have mobile apps in 2015. So if you wanted to talk to those people when you were away from the computer, you couldn't really do it. On the Skype side, there were a bunch of issues that weren't compatible with gaming. If you were playing a full screen multiplayer game and you got a Skype call, it would minimize your game and pop up a modal. You're like, 'That's fine. I'll answer later. Now I just died.' If you wanted to have a group call, you couldn't have different channels and rooms. You only had one group DM. It would leak your IP address, which is a weird thing because gamers are a bit of a snarky bunch. We would hear stories especially from Twitch streamers where they would get a Skype notification that would show up on their Twitch stream and their Skype name would show up on the screen. If you just knew someone's Skype name, you could get their IP address through the system because of the way Skype was built. Then people would DDoS their home computer. We heard stories of influencers who had their home internet being DDoSed and they couldn't get Comcast to change their IP address, so they couldn't use their internet at home for a week. So you stack up 20 of these things and you have a situation where there's a consumer base that is super passionate, tech-savvy, early adopters that are being underserved by complacent companies or companies that are just not focused on them. And then we just had an end. It was kind of that plus some technology shifts with WebRTC becoming popular in the browser. The year that we launched Discord was the year WebRTC went big in the browser. That was a big innovation that allowed us to reduce the friction and interoperability.

I think they just responded to traction. We raised money to build the game before that. I think I was able to raise money back then to do that because I had previously had a successful startup. So coming off the tail of that successful company, I was able to raise money off of a vision and a pitch. It was still hard to raise. It was still hard to get enough capital to do that because games at the time were still like, 'Huh, can you really invest in games?' Our Series A was led by the games guy at the time, Mitch Lasky from Benchmark. The thing I pitched him was game chat app distribution platform. That's the path that we've been on. We didn't spend a lot of the money and so we used the leftover capital from that to fund the initial version of Discord and then maybe six months later raised the first Discord round.

That previous startup also began as a game studio and we pivoted to social tools for gamers. It turns out this is a playbook. Or at least it's mine. That first company was really about mobile. We launched a game the day the App Store opened in 2008. We won the first 50 titles in the App Store. It ended up not really working out, but we had technology in the game that we built leaderboards, chat rooms, account login, friends list that I thought would be useful for other developers. At the time, my analogy was Xbox Live for mobile. This was before Game Center. So, we took out the tech and launched it as an open SDK for developers. That did really well. What I learned from that experience was really two things. One relevance to what we're talking about here, which is pivoting. The thing that I really took from that was when we were building the game, we got to a point where we were going to run out of money. I remember calling my dad and being like, 'I think I'm going to have to move back home because this startup thing is not going to work and I got to get a new job.' I was probably 24 at the time. But we were sitting around and we had this idea of taking the technology out of the game and releasing it. But it was going to take like four or five months to build to pull it out. We ended up actually rebuilding it. So what we did is we made a bunch of fake screenshots in Photoshop, put up a web page, announced this thing in closed beta, got TechCrunch to cover it back when TechCrunch was a big deal. We took the mailing list of 200 developers signed up and we went and raised the bridge round on it. No problem. Amazing. And then we built it and shipped it and it worked. The lesson I took from that was a couple things. One is I would not have pivoted away from the game if we didn't run out of money. So when I was doing this company, I was like, 'I'm not going to run out of money this time because I raised a lot. So I've got to pay attention to see if there's another path because the world's not going to force me to pivot this time.' So I was looking for a pivot. Not that I was trying to pivot because it was really hard. We built the game, put a lot of passion into it, and it didn't work out. So we ended up pivoting. But I was like, 'I know that this can happen and I've got to be ready.' That was probably the biggest lesson. The second one was around culture and company culture. I know I made a values joke before, but I do think probably around the time you get to 20 people, if you haven't thought about values, you probably want to sit down for a minute and write it down and figure out if you like what you see. At my last company, we got to like a hundred people before we sold it. I remember around 80 people, one of the early employees was a good friend of mine came to me and he was like, 'Jason, it sucks to work here. I just want you to know that.' He told me with love and we're still good friends and he works at Discord now. But he was like, 'It sucks to work here. Can we fix it?' When you're 80 or 90 people, it's really hard to change the culture of a company because the culture is what happens when the founder is not in the room. It's about behavior and it's really hard to change that when you have that many people because there's so much happening. So this time around, we were like, 'We're going to be intentional about making this an amazing place to work for creative people.' We put it front and center and made many decisions along the way that were about making sure we were creating a good environment for people and not just hammering people in ways that we didn't feel was appropriate. I think that's played out.

We did not take the approach of get a billion people and then figure out how to make money later. The intention for Discord originally, I never thought it would get as big as it is today. When I started it, I was like, 'If we get the whole market, we'll have 100 million monthly actives. That would be complete slam dunk. Game over. We win.' And we're more than double that now. So with that came the thought around, okay, we're not going to have a two billion person consumer product. So how will we make money if we can get to 100 million monthly actives? Because we were very focused on gaming, I was trying to think through what is the right business model given that we're building a communications app for people who play games. So from the beginning, the plan was we can directly sell utility and expression. And then we will monetize games distribution somehow. That was from day zero for us. So I agree with your conclusion to a large extent. I think if you're going to have a product that has a billion users, there's only so many things that everyone's going to use. These billion user products have to have this least common denominator. I don't mean that in a negative way, but it has to be extremely approachable, something that everybody can relate to, and then it's got to be better than the things they're doing now, especially if it's an entertainment product. So we always thought about Discord as a social utility, a tool that was solving a problem for specific people. We built it for them in that way. We knew it was going to have to be free to start because communication products need to be free. If you want to have a group of 10 people, they've got to be able to use the product. But we also knew that in the world of free-to-play games, people are used to and accustomed to some power and some utility being paid, and people also expect to pay for games and stuff around games. So the thought was that we would monetize it in that way. So I would encourage you to think about building a business from the beginning, but your plans will change.

It was very organic for us how this played out. We designed the tool. The archetypical initial customer set was the World of Warcraft guild or the Final Fantasy 14 guild. In fact, when we launched it, we timed it with an expansion to the Final Fantasy MMO because we knew that when new expansions came out, that was a moment when people would look to bring their groups back together and look for new tools. So it was an adoption moment in that space. What we realized pretty quickly was even though we focused on the Final Fantasy Heaven Sword users, within two weeks, people were using it to play everything. Originally I was like, okay, we're going to go game by game and make sure it works for people. It turned out that the tool was homogeneous enough that people adopted it for Final Fantasy and then they would go play Rocket League and then they would bring their Rocket League friends in and then they would go play Smash Brothers. It just spread pretty quickly. So we pretty quickly realized that it was covering all of multiplayer gaming. We also thought that this was a better version of internet relay chat, IRC. As part of our go-to-market strategy, we were reaching out to mods on subreddits. The gaming ones were like the Diablo subreddit and the Final Fantasy subreddit. We were like, 'Hey, you have IRC links on your sidebar. You should replace those with Discord links because we think they'll be better for your users.' So very quickly we started with gaming groups. That was our main focus. And then I thought that the Reddit chat rooms, the IRC replacement would be a growth factor for us. Then we really quickly started seeing Twitch streamers using it. So we saw people pulling it into different aspects of their lives. We leaned into it and made sure the tech worked and built tools for those people. For the first couple years, every six months we'd have a bigger new famous person or entity launch a Discord server and it would hit our infrastructure limits. It's real time chat. It was made for 50 people and then they're like, 'Oh, there's 200 people in here, it's breaking.' Okay. Now there's a thousand people. The latest biggest one is Midjourney has like 20 million people in a chat room, which is chaos, but it works. So the market kind of pulled it out of us.

This has been a learning experience for us. In what I call season one of Discord, before COVID, we were very focused on gaming. Even though people would frequently use Discord for other stuff, we were kind of like, 'Yeah, that's nice. We're happy that's happening. We're going to make sure it's not broken, but we're not going to really focus on it.' Because our intention was get all the gamers and then make money around distributing games. So if a bunch of people are on here to talk about house plants, they're probably not going to buy games. So it sort of doesn't matter. But if it helps the gamers stay, that's fine. When COVID happened, the world got turned upside down and everybody needed a place to hang out online. So it became confusing to think about when we're doing market research, the signals were all messed up. So we made a series of decisions for a couple of years where we went more mainstream and tried to make Discord more useful in general population use cases. As COVID ended and we looked back on what was resonating with people and where there was real usage and engagement, the gaming and gaming adjacent stuff continued really strong. People loved it. The gaming features we shipped got tons of adoption and the stuff that wasn't focused on gaming just had less energy and excitement around it. So early last year, or end of 2023, we decided to refocus back on our core of gaming. That's the push of the company now. It's working well. It speaks to my point earlier about if you're going to build a general purpose app for mass market consumers, it really needs to be differentiated and better in a 10x way. For most people, if you're not into gaming, the things that make Discord unique and amazing for people who love games actually make it hard and difficult. We could remove all that stuff to make it easier, but then it's not different and unique. You win in business by being different and unique and over-serving a customer segment. So we love games. It's in our blood. We're going to lean on that. The product strategy we took in the early days and we take now is one of what I call creative misuse. We build these tools for people who play games. We build communication tools through the lens of how players need to organize and customize our app. We try to do it in ways where you can do other things with it, but we prioritize and make decisions through the lens of being a gaming tool. For example, Midjourney and all the web3 stuff exists on our platform because back when we launched in 2015, we knew that players were going to want to customize and connect Discord to external data sources. We had someone on the team who played this game called Eve Online. It's an MMO space game where you get in these corporations, groups of hundreds of people, and you do space battles. Some people call it spreadsheets in space. So we knew that people were going to want to use Discord to coordinate in Eve and that we were not going to go and build the data connectors to make sure the right people with the right rank get access to the right text channels. So we just built an API. It turns out that same API could connect to the blockchain and could allow you to generate images off of text prompts. So we built Discord with this customization capability in it and it just turned out that people who love the product took it in all these different directions.

It's literally the next evolution of what I just talked about. We have this bots platform on Discord. Close to 50 million people a month are interacting with these bots on Discord. It's used in a variety of different things. The web3 stuff, the AI stuff, but also moderation tools, people build games, all sorts of stuff. If you use Discord, you probably have a bot in your server. What we just thought was it's limited to text and wouldn't it be amazing if we could allow people to embed HTML5 experiences? Then we thought, okay, but we're doing everything through the lens of gaming. So what if we take the view of promoting this in a way that it's embedded games that you can play with your friends that are instantly accessible and come with a social graph, but if you want to do other things with it, that's fine. So we've taken the lens of doing it in that way. If you go to Discord today, you can click the shapes button and you'll see there are lots of games you can play in there, but there's also a whiteboard, a YouTube experience, and other bots are using it. I just saw a Counter-Strike matchmaking summary tool where you interact with it in text chat, hit a button, and an iframe pops up and you can look at a table of your data that's pretty. So it's just the next evolution of our platform and we're really excited to see what people come up with.

The way that I've come to think about AI is it's an enabling technology. Fundamentally, what people do when they use our service is they're talking and spending time with their friends and other people. So I think about how does AI help us deliver that value to people more effectively or unlock new things that they can do together. That's one pillar of it. In a strange way, I think Discord is not actually that affected by AI. A lot of communities are on the platform that are talking about AI, open source projects building AI things, and that's all great, but fundamentally the value we're providing is a really easy way to organize with people around stuff on the internet. When you're playing a game with your friends, you want to talk to your friend. So how does AI help in that situation? We have some prototypes of things that we're thinking about that are sort of like toys to fuck with your friends and have fun. Then we think about other situations like at some point you're going to be able to go into a public community on Discord and no matter what language it's in, it will be in your language. Whatever language you speak, it will be in that person's language. We see a lot of this in Europe where there'll be a server with people who all speak different languages, but everyone defaults to English. Maybe a third of the people don't speak English that well. Wouldn't it be cool if you spoke Spanish and it was all in Spanish? We could do it right now. It's just a cost question. You could extend that to voice chat. I'm talking to someone live, I'm speaking in English, you're hearing it in Chinese. That is going to come to Discord. But I imagine that will come to all communication products at some point. Discord has this interesting challenge of summarization. You go to a community and it's a bazillion chat messages. You're like, 'I'm not reading all this.' So can we use summarization to help get the nuggets of value to you that are interesting? These are some ways that we think about it. Fundamentally, it's what is the value we're trying to create for people and then how does AI enable it? We've debated whether we should build virtual friends and this kind of thing. I don't know. We'll see. On one hand, it's gonna happen. On the other hand, I'm like, am I the old guy now? 'Kids these days with their virtual friends.' I don't know. On the operation side, we are using AI all over the place inside the company. Every customer service ticket that comes in is seen by AI first and we use humans to escalate for the harder things. Surprisingly to me, customer service scores went up because the answers are more often correct and the response time is instant. We just deployed Cursor to all of our engineers two weeks ago. So we're like, if you write code, use Cursor. If you don't, figure it out or leave. I'm so much more productive using AI co-piloting tools. So we're just going through every function at the company and trying to figure out how do we make the current people we have 10, 20, 30, 50, 100% more productive with AI.

Yeah, we've talked about this one. I remember thinking like, oh there's 10 ways we could make money at my last startup. And one of my mentors at the time was like, 'You think there's 10 ways because you actually don't know what you're doing.' Cool. Cool. January. Oh, wow. Huh? Like, do I ask more questions or do we just keep with the question? When Jason's ready to announce his next...