Y Combinator partner Michael Seibel explains how to build an MVP in one month or less. Talk to users, set up as a Delaware C-Corp, hire strong engineers instead of outsourcing, then iterate fast. He draws on his experience with Justin.tv, Socialcam, and Twitch.
Chapters
One of the things that YC says all the time is talk to your users. And I've got to say, every time my company hasn't done well, I stopped talking to my users, and I stopped being a user. And every time my company did do well, I talked to my users, and I was a user. So when we decided to do Socialcam and Justin.tv— I'm sorry, Socialcam and Twitch. The Twitch idea happened because there was a small community of people playing video games on Justin.tv. And Emmett, my co-founder, really liked that content and thought that the gaming community wasn't being served by good video. And kind of thought that YouTube, while it had great video content, didn't have great community back and forth conversations, so on and so forth. So we kind of took that idea and ran with it. On the other side, we had a live streaming We had a live streaming app for Justin.tv. And as we measured how people were using it, we realized that they were only taking like 30-second clips, which it would be impossible for anyone to watch those live. And so we started digging into those stats and realizing, OK, well, if everyone's using this to take video clips, why do we have a live product? I've got like 8 things in my pocket now. And so, yeah, so both of those new ideas came directly from user feedback. Now, in terms of user testing and so on and so forth, I like to split up product into 2 basic ideas. One is MVP, and the second is iteration. I think for MVP, everything is kind of gut. And I think people spend far too long getting their product comfortable to release. And it's weird, because it's a problem that I see at Y Combinator constantly, where basically what'll happen is that we can't do anything. Anyone here who's trying to give you advice can't really help you if you haven't launched. The best advice they can give you is launch, because whatever you launch is going to suck. There's nothing that has ever been launched that didn't suck day one. Now, sometimes it's used, right? I'm not saying that there's nothing that's ever been launched that hasn't been used from day one, but universally, if you were to ask the person who built it, is this the expression of my mind's eye? They would say no, absolutely not. And even if you think of products that people universally acclaim, do you remember the V1 iPhone? Like, it was great in a lot of ways, but no 3G, no App Store, kind of janky-looking cover, kind of heavy. Like, I don't think Steve Jobs would have said, this is the realization of my vision for iPhone. I think he would have said, this is step one. This is what we can get out there, and it's better than everything else, but it's not what I want it to be. So far too often, basically, People spend all this time lead-up to launch, afraid to launch, afraid to launch, afraid to launch, and when they're in programs like this or in Y Combinator, the result is they can't be getting the advice that you need post-launch. How do you iterate your product? How do you listen to your users? How do you talk to your users? Because pre-launch, you don't have any users. So, you know, I'm kind of going through my checklist of what is important for a startup. You know, my first set of items is always legal. Like, if you don't set up your startup legally correctly, You're just asking for fucking pain, like serious pain. So that's like Delaware C-Corp, simple. Is everyone incorporated? Yeah. Yeah, beautiful. And everyone incorporated as a C-Corp? And everyone in Delaware? Okay, you should fix that. Like, it's so much easier to fix right now. Every minute you wait, it becomes harder to fix. I've raised money from probably, let's say, 100 investors. If I wasn't a Delaware C-Corp, none of them would have given me any money. And I'm sure you wanted some of my investors. Maybe not all of them, but a couple. Ashton Kutcher invested. You might want Ashton Kutcher as an investor. If you're not a Delaware C-Corp, he's not going to invest. So just change that. You can use Clerky. Have you guys ever heard of Clerky? Yeah. Clerky's like super easy incorporation, so on and so forth. Okay, so I won't belabor that point more. First step is legal. Second step is team. If you're building your product based on outsourced development, you also will not be able to raise money from good investors, period. So if you are either not an engineer, do the work to become an engineer or can't recruit an engineer, I would say stop walking the path. Because if 99% of startups are gonna fail, you're putting another 99% failure in front by having a bad team. So now for you to succeed, it's 99 times 99. That's a lot harder. And also, I think it's a good test. One of the things my dad always said was, The original founder of Justin.tv, Justin Kan, there were many things he was good at, but the number one thing he was good at was he was a good recruiter. Somehow he convinced 4 people who were pretty smart to quit whatever they were doing and help make a reality TV show, which is a dumb idea. So it's like, if you can't operate on that level, if you can't recruit your friends to do whatever your dumb idea is, Can't be dumber than online reality TV show, then how serious are you about it? How likely is it to be successful? And I think that's something that people don't talk about enough. For me, failure always motivated me. Fear of failure always motivated me. And so I liked to know how hard this was. It also made me feel like if I did it, I really accomplished something. It wasn't just like, oh, you graduated from high school, great, everyone does that. It was a real accomplishment. So that second piece is team. First piece, legal. Second piece, team. Third piece is MVP. And the way I like to think about MVP is that if it takes more than a month to build, if it requires a significant business development deal, if it requires— shit, if it takes more than a month to do, it's probably not MVP. What's the best way to find a good MVP?
