Patrick Collison at Startup School 2012: The Early Story of Stripe (Deleted Video)

Good morning. Hopefully this clicker works. So when I sat down to write this talk, I really racked my brains to try to figure out what the most interesting and relevant and useful content would be for a lot of people who are sort of interested in web startups. And when I thought about it in those terms, the topic of my talk became fairly obvious. But then I thought about it more and the algorithm doesn't even seem to be deterministic. I'm not sure anyone even knows how it works. So I guess I'll actually have to talk about Stripe instead. Two years ago, Brian from Airbnb spoke at Startup School in 2010, and he opened his talk by showing a picture of Startup School in 2008, two years previously. And that picture showed a snapshot of the audience and he picked out one particular head. He was just sitting there, an anonymous head in the audience. And I know this because I was at his talk in 2010. And so two years later, he was on stage at Startup School. And somehow two years later, here I am. And I'm sure there's at least one person here who'll be on stage in two years time. And Stripe definitely isn't as far along as Airbnb, but it really drove home the point to me when I realized this that startups are very unpredictable. You know, startups are really hard to get your head around. I remember somebody telling me not all that long ago that Airbnb was just becoming a huge success. They were making something like $10,000 a week. I remember thinking, man, that's really impressive. And now I have no idea what Airbnb's actual revenue figures are today, but it wouldn't at all shock me if they're making $10,000 an hour. Like, what's it even like to try to build a company where that kind of growth is possible? Startups are strange for a bunch more reasons. You know, they're usually kind of counterintuitive because if they weren't counterintuitive, somebody else would probably have solved whatever the problem is already. And on top of that, startups often don't want to be all that well understood. It can actually be quite helpful to be sort of underestimated and undercomprehended. But I think the biggest reason that startups are hard to understand is that so few people still get to see them up close during those first couple orders of magnitude. And when people do tell the stories or when the stories are told, they often tend to be kind of wrong. They're about sort of rocket ships and frantically holding on and trying to add more server capacity as quickly as you can to handle the next million users. They're not about the late night arguments and wondering if your product could ever possibly work or trying to figure out once you've launched your product why it's not growing. And the thing is, even the really successful startups have this phase. That famous summer when Facebook moved out to Palo Alto in 2004. Most people didn't realize this, but there were other people in the same house working on other startups. Like this is the most successful technology company started in the 21st century and in the same house there were people working on other startup ideas six months after launch. It really takes a while. And so Stripe is obviously very, very different to Facebook and Airbnb and nowhere near as sort of far along as they are. But we've now gone through a small bit of growth. And so I want to try to describe how it actually works. In October of 2009, John and I were walking home from dinner in Potrero up in the city and we sort of been kicking around this idea of starting an online payments company. We were really kind of fascinated by the concept of internet payments and all the other companies in the space seemed like total dinosaurs. And as the web sort of spread around the world and more deeply into our lives through mobile devices, it seemed kind of obvious there should be some kind of universal payments infrastructure for the internet that it should be just really easy to transact online. And John just turned to me. He said, "You know, there wasn't debate about it. You know, why don't we just go build it? It probably won't be all that hard." He actually said that it probably won't be all that hard. And so I said, okay, sure, why not? John is my co-founder at Stripe. He's also my brother. And people often ask me about how this works in practice. And so for the record, starting a company with your brother turns out to be a really good idea. John is not only one of the most brilliant people I know. I think he got the highest ever results in Ireland's university entrance examinations, but he's also someone with whom I've literally decades of experience building things. Here's an earlier venture that we worked on together. And so it's October 2009. We decided to work on this online payments company. We decided to call it /dev/payments. You know, the API should be just as straightforward as any other node in /dev/fs. We're kind of okay at programming and definitely not at naming things. But everyone else seriously had been targeting their product at like they thought it was a finance sector product. They were targeting at CFOs and at business people. And we thought the internet was just moving in a completely different direction. We decided to target makers, the people actually building things. We thought that internet payments was a technology problem and we wanted to build a completely new stack for anybody transacting online. So we worked nights and weekends. We were both in college at the time. I was at MIT and John was up the road at Harvard. And so we'd code together in the evenings between problem sets and writing papers. January rolls around which for both of those schools you sort of have that month off. But of course anybody who knows Boston knows that January in Boston is unbelievably freezing. And so we decided to go somewhere else to work for the month. We'd read a few blogs that claimed that Buenos Aires was surprisingly a really good place to get things done. It's pretty cheap. It's really warm. It's friendly. For some reason, I don't know why, there's Wi-Fi everywhere. Everything happens on a really late schedule. Like the bars are open till 5:00 a.m. Nobody has dinner until midnight and nothing starts before noon. It's basically a city on a programmer schedule. And so we read all this and we're like, great, we'll go to Buenos Aires. And so we did and we just worked nonstop in cafes for 3 weeks. I've still never seen a single one of like the tourist attractions in Buenos Aires or I mean I presume they exist. And then we'd code all day and we'd go to dinner at 11:00 or something and then go to bed. I can't emphasize this enough. If you want to get something done, consider going to Buenos Aires. On January 9th, we got our first production user for /dev/payments. I can barely say it. And this is only a few weeks after we started working on it. But we really wanted to get production users shaping the product as quickly as we could. The user was a friend of ours, a guy called Ross Boucher, who was working at a company called 280 North at the time, which is also actually YC funded. And funny enough, he actually became the seventh person to work at Stripe, but that's a separate story. Here's a screenshot of what /dev/payments looked like at the time. And in this screenshot, you can see that John and I are definitely programmers and not designers, but okay. So, January 9th, we now have one user. There's this great story from the early days of Amazon about how they celebrated when they got their first buyer who wasn't any of their moms. And you know, Ross wasn't exactly our mom, but he was a good friend. And so we definitely weren't out of the woods just yet. So, we went back to school and we continued to work on /dev/payments in our spare time. There was this one cafe that I worked out of so much they took pity on me and I'm still Facebook friends with a bunch of the baristas. And so, yeah, Stripe is a kind of unusual company. We're about technology, but we're also about payments. And the technology side requires good reliability and clean APIs and a really nice product and lots of technology things we at least hoped to know something about. But the payment side requires working with banks and dealing with credit card companies and just generally handling a slew of finance industry issues that we really had no experience with and no idea how to handle. We had a lot of meetings where I sort of sat somebody down and said, "Right, so payments, how do they work?" And you know, programmers often, and this is unfortunate, sort of look down at the folks who are trying to learn to code, the "I want to learn Rails for my web startup" crowd, but I actually have a lot of sympathy for them because we were that, but in finance. Summer came around, so we're now six months in. We moved out to Palo Alto, though we hadn't actually yet decided to take leave from school. We found a tiny bungalow just off University Avenue. The living room and the kitchen became our office. It was pretty hot. We didn't have any air conditioning and so John just slept in the garden most nights. He wouldn't allow me to post a picture of this. And by and large we just kept on writing code because that's mostly what starting a software company looks like in the early days. You're either writing code or you're talking to people who use the code or you're probably wasting time. And here's a chart of our transaction volume over the first six months. And that's not a technical error. If you look really closely, you can see a tiny little wiggly line at the bottom. It's admittedly not wiggling all that much. Around this time though, we did have our first person join. We're based in Silicon Valley, home to the best international talent from Berlin to Beijing to Bangalore. And so, of course, we took full advantage of it. We hired a guy called Dara Butley, who was one of my smartest friends from college. And he grew up in a small town called Limerick in Ireland about two miles from where John and I grew up. We also raised our first investment, or our first real investment I guess. YC had obviously already invested. And this was from Peter Thiel. Up then, we hadn't told very many people about /dev/payments, and those we had told had reacted mostly by telling us we were crazy. But of course, luckily, Peter Thiel is crazy. And so he invested and it's been really helpful to have him on board. Soon after that, moved into our first office, which looked something like this. It's sort of, it's at Ramona University in Palo Alto, just next to Coupa Cafe. It's actually a converted house. And because it was a house, it had this wonderful fireplace they were able to light in the winter. The fireplace looked something like this, which I think may have contravened some fire in the workplace codes. And basically, we're just working non-stop. We actually became four people around this time. The fourth person was a guy named Greg. And I remember when Greg was sort of considering joining and dropping out of school, he asked me sort of about our work schedule and whether or not we worked weekends. And obviously we did work weekends, but he really didn't want to scare him off and make it seem like he'd have to. And so I sort of said, "Well, you know, we usually work some of the weekend. We really care about work life balance." And he just cut me off. And he's like, "Great. I'm on the same page about working all day every day." And so we knew he'd fit in. We decided to change our name. I can't even begin to list the problems that /dev/payments had. Somehow it turned out that not everyone immediately got the /dev/ analogy. Nobody was really able to get their head around the whole slash thing. And we started to get mail with things like this. And then there was the admittedly pretty inconvenient fact that Amazon had launched a product, an online payments product called Amazon Dev Pay. And so it pretty clearly just wasn't going to work out. And so we spent hours brainstorming names and we eventually came up with the name Stripe. And we didn't actually think that Stripe was all that great a name when we first came up with it. But we decided to just put a date in the calendar. And if we didn't come up with a better name by that date, then we'd stick with Stripe. And sure enough, it became Stripe. And actually, a few months later, I learned that this is how Apple ended up with the name Apple. It was the exact same date thing. But anyway, we reached the end of our first year, mostly intact. A bunch of Y Combinator companies started to use Stripe and that was pretty good. We were getting some real feedback. And here's another chart. This is our transaction volume through the first year. And so you can see still kind of a ways to go. We spent January of 2011, our one year anniversary, in Rio de Janeiro because January in South America was now becoming a tradition. As anyone who's been to Rio de Janeiro knows, it's one of the most beautiful places on Earth, and of course we took full advantage of it. At this point, we're still in invite only private beta and so there probably is something obvious we could have done to grow at least a little bit faster, i.e., to launch. A friend refers to invite only private beta as the baby blanket of startups and I think that's about right. But the beta period was actually really helpful to Stripe. And we heard Paul Buchheit saying that you should start out by making 100 people really happy rather than many more people only a little bit happy. And we really took this to heart and took advantage of the small number of users to sort of really focus on them as much as we could and sort of really try to figure out what they wanted. And so, for example, not long after we launched the first version of Stripe, we added PagerDuty. So, we'd all get called when the site was down, you know, pretty standard and pretty reasonable. But then we realized that anytime a user gets any kind of error, like even if the site isn't down, that's actually a really bad experience for them. And we could probably make them much happier if we went and investigated the error and sort of proactively reached out to them and helped them fix it. And so we changed the code a little bit so that anytime anyone hit any error, it would send an email that would go sort of straight to the top of everybody's inbox and it would also phone everybody. And we'd go and fix it no matter what. And I really mean that, like we would get out of bed if necessary. We're basically never without a laptop and the means to tether. Here's a photo of one such incident. This is Greg and I. We'd gone to the cinema in Redwood City. A user encountered an error and so of course out came our laptops and we had to go fix it. I think Greg is here somewhere actually. And I think he still feels a little bit bitter about missing the start of the movie over this. The movie may or may not have been Twilight: Breaking Dawn. Startups are high stress. You need to unwind. We also realized that IMing with an engineer is basically the best support experience possible. Like it's really frustrating to have to go into file a ticket or find the support email address and then wonder does this company actually reply to their support emails or if they do reply how long does it take them or all these issues. Like it's much nicer we can just start sort of directly chatting with somebody. And it's also way more productive for us because we can then go and sort of try to figure out what the underlying issue is rather than having to sort of guess based on the user's initial description. So we just opened up this chat system on our website where anyone can jump in and just start asking questions. And we actually still have this today and that was really good. But then we thought, why stop there? Isn't it a really bad experience when somebody asks a question and there's nobody there to answer them? And so we thought, aha, PagerDuty. And so we hooked it up so that if you asked a question in the Stripe chat room and there was nobody there to answer you, it would go and phone one of us after 30 seconds. And so I don't think many people know this, but for many months at Stripe, if you asked a question in our chat room and there's no one there to answer, someone would be woken up if necessary to help you. We don't do that today. And so though Stripe wasn't yet publicly available to everyone, we really tried to sort of turn up the dial on our users feedback and to force ourselves to be extremely sensitive to what they wanted and to what their experience was like. We had our users talking to us during every waking hour and if anything went wrong for them, they were literally interrupting our sleep. The other thing that kept us in beta was the fact we weren't just building a thin software layer. And we thought that Stripe should encompass everything from the API requests to how the money ended up in your bank account. And we wanted to be able to define the complete experience and we wanted to be able to do it at scale. We were sort of really influenced by things like Amazon Web Services and EC2 which it just it's a really interesting innovation because EC2 is fantastic if you're a small company or a startup or a side project or something like this, but it scales right through to being a Netflix or a Zynga or indeed an Amazon. It will work for a company of any size and we decided that we wanted to do that but for internet payments infrastructure. We wanted to make something that was really easy to start with but something that would also work for the largest companies in the world. And so that meant working with really good banks. But the problem is that banks and startups are basically the business equivalent of oil and water. And figuring out how to combine them is pretty hard. The best bank in the business is probably Wells Fargo, which powers some of the largest payments companies in the world. And honestly, it was pretty tough to get them to talk to us or honestly even to return our emails. They probably thought we were some strange kind of Nigerian scam like "make money online fast all from the comfort of your own home." So we asked a friend, an investor, and actually now a partner at Y Combinator, Geoff Ralston, to help out. Geoff had previously been CEO of a company called Lala, an online music startup. And they had negotiated successfully with the record labels. And we thought that if you could, as a technology startup, negotiate successfully with the record labels, you could basically convince anyone in the world to do anything ever. And here's a picture of Geoff on a conference call with Wells Fargo. And that's Dara there on the other side. And you might wonder why is Geoff on the floor? Well, Geoff is on the floor because our office was also flooded at the time. And we also had a security audit that day. Sometimes startups are just like a reality TV show. It's like negotiate with one of the biggest banks in the world while undergoing a security audit while wading through water and I don't know, the water's full of piranhas or something. But thanks to Geoff and a bunch of others, we eventually convinced Wells Fargo to become one of our backends. This meant moving all of our systems to working on top of their platform. And it was a couple of weeks of honestly really intense work. We had to hit a particular deadline. This is the night of our first successful transaction. This is John after another particularly long all nighter. And in general, and I kind of mean this seriously, this is the unglamorous side of startups that people do not get to see all that much. Like you really want to make something work and lots of other people think that it's a bad idea and it's really hard and everything happens much, much slower than you'd like and you've sort of many late night discussions like this and soul searching debates and wondering, "Is this actually a good idea or maybe it is a good idea but it's just too hard for us to pull off?" And startups are hard. And the thing is, this doesn't actually go away. This is the thing that I didn't realize before doing a startup. I thought that you have all these doubts in the early days and then it would start to take off and things would get easy. But no matter how successful you are, you will have lots of doubts. The first American to win the Tour de France was a guy called Greg LeMond and he has this quote I've always really liked. It doesn't get easier, you just get to go faster. And it's kind of like this with startups. The startup might start going faster but it doesn't really get easier. In our case, we eventually got the pieces in place and we got to the point we're ready to launch. We launched on the 29th of September 2011, so just over a year ago. At that time, Stripe had been in production use for 19 months and we'd been working on it full-time for a year and three months. We were 10 people when we launched. I was able to fit all the names in a tweet. And by the end of the year, here's what our daily transaction volume looked like. You know, kind of promising, but launching is definitely not a panacea. But the signs were positive and we kept going. And over the last few months, like two and a half years in, Stripe has finally started to become an overnight success. Jessica talked about how startups are roller coasters, and there's a lot of downs, but there's also ups. And the next slide is our daily transaction volume through to today, and it looks like this. From being four people not all that long ago, Stripe is now 34 people, which is actually twice as many people as are in this picture because everything is always behind on a startup and we haven't got around to taking our new group shot yet. And many thousands of companies are using Stripe and a lot of new ones go live every day. And I mean there's a bunch of well-known brands like Foursquare and Boxee and Hipmunk and BFF and New York MoMA, but there's a lot of less well-known ones too. And they actually tend to be just as interesting. There's things like Good Eggs, which enables you to buy directly from local farmers, or Samasource, which brings computer-based work to people living in poverty, or even stuff like the Bedford Cheese Shop, which is now just selling its cheese online. I guess I was just hungry when I was making these slides. Cheese seemed like a good idea. Okay. Has anyone here ever looked into the history of container shipping? Sorry, I know this is a bit of a context switch from cheese. Container shipping, like the shipping container, those 40 foot shipping containers. On some level, they're the most mundane thing in the world and we see them around all the time, but they were actually an absolutely enormous innovation. 60 years ago, before shipping containers, transportation was a massive issue with physical goods. Transportation costs often accounted for up to 25% of the final costs of a physical item. And overall, transportation cost and shipping costs were like 10% of the value of all the imports in the US. Think about that for a minute. 10%. That means if your margin's 20%, just by bringing your goods to another market, you cut your profit in half. And so, unsurprisingly, most manufacturing happened sort of very close to where a physical product was consumed. And so the shipping container invented in the mid 1950s essentially eliminated the cost of shipping physical goods. It cut the cost of loading and unloading ships by 95%. And it's now at the point where, and I'm actually quoting this from a US government report, it's better to assume that moving goods is essentially costless. Really, and the history here is fascinating. There's a few kind of good books about it, but the point is this technological breakthrough and this elimination of friction, and in particular this abstraction over geography, played an enormous role in facilitating the rise of Singapore and South Korea and Taiwan and Japan and China as manufacturing hubs. And my point is really that a technology that doesn't just monetize itself but actually enables newcomers can have really enormous impact. The shipping container literally reshaped the world's economy. We sometimes describe what we're doing with Stripe as building economic infrastructure for the internet. It's not very flashy, but for most of human history, we've had to buy from the people beside us. But thanks to the internet, that's no longer true. We have a new way to abstract over place. Anyone can now build a global business. But yet, while the internet has revolutionized how we communicate and how we collaborate and how we share, we've only sort of started to explore how it can change what we create and how we transact and what kind of businesses are possible. And a lot of the other companies whose founders are speaking today are pretty good examples of that. With Stripe, we simply want to turn payments into a ubiquitous utility. We want more commerce on the internet. But it's still really early days and honestly most of the time we're not thinking about these kinds of problems. We're trying to figure out how to decrease the load on DB3 or trying to get some particular design just right and working to the 20th iteration, or wondering if some particular product is a good idea or having some debate, you know, in Gmail that's so long we've overflowed the Gmail thread limit and it's now become some completely new thread. It's all the day-to-day stuff. But you know, it turns out that all the high level stuff, the larger motivations and the bigger ideas, and the day-to-day implementation, the debates and the tweaking and the iterations, both of them are really addictive and that's why we keep doing this. Thank you.