Question & Answer with John Collison, the Co-Founder of Stripe – MicroConf 2017

0:00[Music] [Music]

0:16so real quick before we get started how many people in here use stripe a lot so that's your context for this conversation so it's great to have you here I think we're gonna dive into a couple of different questions that Rob and I have talked about and wanted to ask John from the stage and then after that we're gonna take some listener questions or you know it's questions from the audience I said listener questions because the podcast but we'll take some questions from you guys and we will go from there obviously there may be a few things that he can't answer just because of legal reasons or anything like that

0:49but he'll try his best to answer whatever he can so the first question I have for you is what was the biggest initial challenge you ran into when you were building stripe and developing it hmm well that kind of the the first issue we ran into was that we really had no idea how to build stripe and what I mean by that is I think it was I sometimes talk about stripe as a fairly obvious idea in that any of us who you know I was among them or any of us who had actually had to use one of the existing merchant account providers were kind of tearing your hair out and this

1:22was it was fairly obviously the need for us and so Patrick and I we were actually in in college at the time and we decided to build something better you know we said how hard can it be and actually pretty hard it turns out and but with us a kind of aspect that only I think you can have when you're when you're 19 where nothing seems difficult and we set out to build us so we stubbed out the API and what to do you know it's like okay except a credit card here and then like the money arrives in your bank account and then we're like okay what

1:53comes next and and so we did spent a long time kind of searching and casting about for the right partners in the financial industry and learning how to speak their language that was one of the hardest things trying to gain some measure of credibility and you know one of the ways we got our very first partnerships with Wells Fargo was an investor you know an early investor put us in touch and so we're gonna have to have that credibility in that connection in that regard so that was kind of issue number one was we spent a good two years prelaunch working on us because we were

2:28navigating the financial industry and then the second issue with us the I think ever in this community gets that one of the best things you can do is to shorten feedback loops so that you can iterate on the product quicker and so that's why it's so useful to actually sit down and talk to customers and that's why it's so useful to come to things like this you want a really tight feedback loop on your on your product development process that was really hard by four stripes product by its nature because people you know would only use stripe when they're actually ready to go and charge for what they were building

3:01and even then it would take them a while to integrate and stuff like this so we'd fairly quickly give people access to stripe and you know six months later they might sit up and running so I actually went back in in advance of this I was curious to look at what the the shape of the signup curve looked like and we you know the first commit was October 11th 2009 and it was only two years later that we got a hundred businesses on stripe and so you know it was a really slow ramp initially and that was pretty hard because just from a product management point of view it's

3:27it's hard to know what your customers want when you have no customers so how long did it take you before you got your first customer so you said you started in 2008 timeframe you spend a couple years and then it took you two years to get set that hundreds of customers yep what was the time between the zero customer you know customers in one custom no customers to one customer is actually pretty clear what I think of is pretty quick in that it was three months but we were working on the part time and so by January 2010 we had our very first customer and it was a guy called Ross

3:52Boucher who was working on at the time I'd know if any you guys have used the cappuccino web framework which is implement in cocoa in in JavaScript but they were charging for the development kit and actually now kind of full circle he's working on one case which is part of stripe um and and they worked at stripe for a while so it's very much within the family but so we got that first customers in three months I think it was actually pretty helpful because stripe is not you know a one-shot product you integrate stripe and then your honor's and so kind of three months in when we're still working on the part

4:25time we kind of couldn't stop because you know now we have paying on customers who are not just relying on us in some incidental way for a minor task you know their business was running on stripe and so we kind of had no option from there so you're first the first customer that you had paying you for stripe you eventually went on and acquired them or is that so that's I mean that's a different customer acquisition and I think we're both this is kind of much later in life in that you know else works at stripe for a while and then then ended up leaving the

4:54stars in one kiss and then kind of came back in you know so I guess in the the early days look how much of your time was spent working on the the technical stack versus going out and talking to people and trying to get the feedback and doing actual marketing for it because it seems to me that whenever I hear people talk about stripe it's almost never from an advertisement they never heard of it from you know they never saw an ad for example they always heard about it from somebody else so it seems like word-of-mouth is really that primary primary channel but you know like what things did you do in the early

5:26days to try and help drive interest in those conversations yeah so one I'll say you know and I'm talking about this stuff I'm not sure how generalizable any of stripes experience is because I'm pretty sure we were just like a weird special case and so for example to this day if you go and actually google stripe payments right now you'd expect to find like a branded keyword for a stripe but we still haven't gotten around to doing that and so you'll find an ad for Amazon Payments if you do that and but what we did obviously a lot very active customer development in the in the early days I

5:59think did a few things we did that ended up proving really valuable one is messaging and actually telling the story clearly I think people talk about word of mouth as if it's some fundamental constant of the universe that you know you know news about your product will spread as x-rays and obviously that's not true in that one kind of the better you tell your story to kind of how much you tell your story and how much content marketing you do and things like this and thirds how differentiated your story actually is those are all fairly fundamental inputs into the equation of word of math and honestly kind of - at Ryan's point

6:38of you know this feeling of sass being full or you know feeling like there are too many sass apps and you know there aren't opportunities left I think one of the ciao people sometimes run into is you know we've all used or at least a decent number of us have probably used HR systems and you know we use workday at stripe and you know there's other options and it feels like they're all horrendous to use or at least my experience with with with workday has been that from kind of a UX point of view and the obvious conclusion is like oh we should go so you know start a an

7:10HR app there's kind of other challenges with that marker but I think one of the challenges is just from a from a differentiation point of view standing out from the crowd is really really challenging because it's like a hundred HR apps out there and so having something that will actually spread via word-of-mouth you need to make it kind of tangible and and interesting and so for us the space we were competing in it was basically all it was PayPal who were kind of their own thing and and as far as I can tell have a negative NPS and then you know we went to look these

7:42numbers recently and then you had a bunch of banks and kind of people who are not used to speaking to the internet business community to the developer community everything like this and so we made a huge effort early on to show that we were speaking the right language we understood subscriptions we understood SAS we understood internet business models it was instant sign up the reason we put code on the homepage was not that's not the logical place in the you know product cycle to put it and that you want to have the homepage was a broad overview and then get the API Doc's the reason we put code on the

8:12homepage that actually worked and you know to this day you can copy that a lot of people who could in the homepage but no one actually does search where you can copy and paste like the actual code snippet Ranas in a terminal and it will run a valid response that was our way of showing but like we get us you know the API is important and we're trying to communicate this and to people same with the tag line payments for developers you know I think there are lots of less crisp tag lines you could have used so that's my main thing onward Amathus I think people tend to it's like well why

8:38aren't people talking about my thing whereas the the messaging was probably the the most I can talk about other things but I think that's the most valuable thing we did yeah and I think it's interested in how you specifically do that piece of it in that if you're logged in it automatically injects the API tokens and stuff so that the developer hasn't it doesn't have to do an extra step on top of that was that was that something that came up kind of in retrospect you just did it to see how it would work or was it like an intentional deliberate choice early on like hey if we do this people won't have

9:06to do these other things it'll just work for yeah I think we it became yeah we have since retrofitted a strategy on us and so that way we can sound smart for it's like what you do is but I was getting exactly yeah yeah but it's very accidental at the time the way it happened and then we can learn from it where in the beginning I think you just do things that that feel right and then you notice the things that are working particularly well and you pour gas in them and so in our case what we found was that things like you know the fact

9:38that you all the API documentation was prefilled with your API key or we'd have just kind of some interesting and very detailed error messages and things like that we did them just because it seemed like the helpful useful thing to do at the time what we found was that they actually drove word of mouth in that you can get api meta you can get api error messages that are helpful enough that people will tweet hey look at how helpful this api our messages that's not a thing we expected going in and so now we actually do you think about it in terms of when it comes to a launcher

10:06when it comes to functionality in the product what is the the thing that will be tweets worthy what is the thing that is kind of surprising and good enough and that people will actually go ahead and talk about it yeah i mean part of that kind of leads into the the just falling into things accidentally and likely as i said you look back at it retrospectively and say oh wow this really worked and there's like a base level for some of those things works just they suck so bad that when you provide even a you know bare minimum experience for them isn't that bad that it reflects very very positively on you

10:37now when you were going through these processes like do you have do you have a process in place for putting those things together whether it's the documentation now or i guess how early did you start that to be more intentional about those things versus looking back at them and saying oh this worked that worked because there's it be very easy to go through the history of stripe and say well was this a deliberate choice or was this just kind of accidental and lucky and i think most businesses successful ones come across that way and so and when you say those things what do you mean so i mean with like

11:10putting the api key and and providing fantastic error messages that people will tweet out it seems like the strike gets a lot of little things right and the collection of all of those things makes it like stripe like an exceptionally fantastic story to follow and to look back on I'm curious how early it was that you guys started to recognize that these are the little things that add to the greater story and how can we be deliberate about those moving forward yeah I think it's hard to go back and retroactively lis act and add them in I think it just has to be somehow part of the the product design

11:45process and I'm kind of what you decide is the way that you you design products it actually gets a little different once you're you know as stripe now has order of I'd say 200 engineers and you know most of those people either directly or indirectly end up working on and shaping the product and then there's lots of people who are not in engineering who end up working on the product in various capacities and so it ends up being pretty different once you have you know people whose names you don't even know working on this stuff and you want to inculcate it from a cultural point of

12:15view but in the early stages there's a lot of stuff that is not measurable I mean usually because the the the the data are too small you just have to you know you can't measure you know will do these things in the product and see which leads to the you know the largest number you know social engagement or something because again you just you don't versus a statistically significant data set and so you kind of have to go on instinct for that stuff right so it's more or less just guessing at what you think is the best thing to do and then once you've accumulated enough data it's

12:42really kind of looking back on that and saying oh what pictures can we draw from this yeah you know and we've had a lot of time to go back and kind of see what we think worked well and then you know pour gas on those particular activities and call the call the things that didn't and that I think people tend to perceive stripe is this like overnight success you know Patrick I are coming with this tradition of stripe where we someone imported from Microsoft where people bring in n pounds of chocolate on there and and anniversary and so you know in your fifth anniversary at stripe you'll

13:15bring in five pounds of chocolate and obviously if you do the math this leads to a really bad place because you know you have people getting more ten years and a growing number of employees and so you you wonder does per employee go up over time and it does in kind of scary fashion but anyway Patrick and I are coming up on our eighth anniversary working on stripe soon to where we'll have to I guess bring in 16 pounds of chocolates I would rather a last and but anyway we've been given that we've been working on it for this long just year after year you continue to go through this process

13:42of you know what's working well what isn't what should we kind of continue to invent so if you kind of look back at what the history of stripe has been if you were to look point out maybe two things that you would do differently based on what you know now versus what you did then whether they were mistakes or just you know market position or something along those lines what what two things kind of stick out in your mind is things that you would do just completely differently now we are we're fortunate in the position of our there's probably no big giant kind of you know

14:18thing of cratering impact that happens during stripes history so I would like well we definitely wouldn't you know do that part you know all got arrested and but but but it's a bit of selection bias there right and that I probably wouldn't be here talking to you if if there was when those moments you have to find the company this you know ever for an interesting answer to that question you have to find the companies that had somewhat more ignominiously ends and that said I mean that the minor stuff is there are issues that we spent years beating our head against the wall where I think that the answer was pretty

14:54obvious in retrospect and it's always pretty faint painful when you you know you measure something and you know on a chart you know if down is good like latency or whatever you make some tiny change and like the church art just falls off a cliff that's really painful because like we could have made that change two years earlier and so you know just something else have my mind that I was thinking about recently I think we are not particularly happy yet with the usefulness of one stripe subscriptions logic or I think and we could just do a much better job fish and to just the usefulness of the stripe dashboard and

15:30so those are things that you know if I was to go back and do it again I would probably invest much more four years ago versus kind of now what we're doing I actually I'll use this as a quick plug for I brought Adam who is where's Adam gone that corner oh yeah he's waving at the back there he's a back there wood patio 11 Adam runs the stripe dashboard and so we would be very eager to hear any if all your bug reports to Adam please are and any feedback you have and but that's probably something I would have changed a little earlier okay um I guess on that

16:00note you because you recently overhauled your dashboard how difficult was it given the size of the stripe customer base and how many people are using it how difficult was it to kind of retrofit that dashboard or development in parallel with what you've got now and then kind of toggle things over I mean one of the things that strikes me is that within stripe when you sign up for an account you can go for an extremely long time without ever upgrading to like a new version and I've talked to people and so like that's that's unheard of that's amazing that you're still able to use something from 2011 and I know

16:32people who are using that person oh we know them too it's like that like that CI a deck of cards there's like five of them and you know their names yeah so how is it that you were able to to do that like what sort of challenges were were there associated that because there's impacts on your feature development and new like new feature development and making sure that everything's still going to work when you go over managing customer expectations along with making sure that there's a base level of functionality etcetera yeah that forum that for me is maybe a good example of kind of the product feedback and customer I'm three

17:08point and that's you know one thing is when we talk to people who are using PayPal and using the PayPal API is just you know you have integrated it and then it would randomly break at some point and then you have a fire drill and you go fix it and it seems like you know a property of relying on the API should be this is behaves the same over time like that's a pretty useful fundamental construct and and and that's not what they were getting and so from a technical point of view what we designed is it's actually and of cool is basically when you make an API request

17:37to stripe it says they're using the old API you know we are running the newest API and then there's kind of a series of translation layers like an onion that go and kind of roll forward the old API request to the new API requests and reformative such that everything is is backwards compatible so that's internally we can roll out stuff at a fairly fast rage but people can use old versions of the API actually maybe that's another good example to your question of what I might change in retrospect and we didn't set any kind of a cut-off period and so we do have people from like I don't have anyone

18:09from we probably do the people from 2010 who still you know their API integrations are working and I think probably good at some points to roll people forward to the newest version of the API this is just like it makes changes to scary right or if you want to use one new feature then you have to redo everything um but yeah that for us was a pretty obvious one okay looking back on your history a little bit just as a last question what kind of then we'll take some things from the audience when you look at the things that have happened is there a low point that you can kind of

18:36point to that sticks out in your mind is you know you weren't really sure that you were going to be able to pull things off or even get through the other side because you know whether it was funding or just technical issues or you know scaling the business or making different partnerships anything that jumps out of you is potentially like very very scary at the time yeah this is um this is one actually thing there's actually very useful about you know Patrick and I know a bunch of other people running technology companies or whatever and I think having a pure group of people who are in a similar position is is really

19:10useful because the PR story never tells the whole thing right you want to have that you know everything is perfect and all this has been and we've always been kind of known everything about the business and you know you have this very pristine image and underneath you're kind of furiously trying to keep things together I'm actually I think one thing that's being useful for me and Patrick is the fact that there are two of us and always have been in that you you maybe your mood is that the average of the two moods and so the amplitude of the swings is is lessened by that kind of dampening

19:41effect and that actually once you start having employees I think is a really important role of the founder is again dampening the swings that tend to come with both kind of the highs being too high and the lows being too low for us one you know during the early days we had a handful of pretty bad outages and to begin with it kind of scaring the abscess you know in the very early days we would have an outage and then um I'm known cared no one called and that was bad because like oh we have no customers and and then then as you started picking up you know

20:16I remember at one point when there are a lot of people using stripe we had an hour's outage or best part of an hour during the month of December on a weekday and I really would not recommend that as a life experience also there was one you know during the early stages of a company an employee leaving or kind of multiple employees leaving is that at least for us like are we I mean will stripe ever amount to anything are we going to pull through this you know adds an early engineer's leaving that was was hard we're gonna take a couple of questions from the audience Rob's

20:51gonna go around so we've got about 10 minutes or so left first question I'm what I'm wondering what's your big focus and to focus in terms of acquisition now like are you really focused on acquiring new users and how are you doing that now we wouldn't we know about word of mouth but what's what's kind of the new strategy or is it still just word of mouth word of mouth is definitely a big part of it for us and again I think it's helpful of us you know we're in a well-defined space where people need to actually buy get something in this space you know you can't not accept money and

21:33and it's it's kind of interesting to people but again I do want to stress the word of mouth not being a fundamental concept of the universe and we have always worked on PR fairly actively and that is something that's you know you can actually use to help drive word-of-mouth again it's part of the cycle of her you know people are tweeting an article or stuff like this and so you know this week there were three separate events that there were articles written about there was you know we had the acquisition of indie hackers we had a increment magazine that launched today and then we relaunched Atlas and and added some functionality

22:10earlier this week but in all of those you know part of what we wanted to do was just ensure that the story was told in a fairly high fidelity and actually in some detail and to their stories about all of those so that is a a large one for us and if you can get that right if you have something interesting to actually talk about or if you can make it interesting and that is you know when we launched Atlas this blew us away but when we launched the Atlas it was on the front page of the International New York Times and you know if you'd asked us five years ago I

22:41mean we wouldn't have believed you and but that's the kind of mind sure that I mean really you can't I mean look it's you literally can pay for because you can buy ad space there but it's it's certainly pretty expensive so that's one and two is we are always thinking about the the funnel and and we are actually Adam and I were just chatting on the flight over about some funnel optimization stuff that we want to go work on to improve that and you know thinking about is getting further out the funnel and kind of bring people in we have all the Sun I feel like once you

23:17once you say content marketing and kind of gets kind of it's like you you know if you're scuba diving you shouldn't touch the coral because it dies instantly and I feel like there's a bit of that with the term content marketing first loose it's interesting this abyss or it's like haha some fresh content and grist for the mill and but we have always done content marketing on on a bit of an ad-hoc basis and so one of our earliest efforts was capture the flag I know if any you guys played that contest and where we did three of them but that was massive in terms of just having

23:44people hear about stripe and you know established some technical credibility and so I think we're gonna do more stuff like that it is very hard to at scale do creative and interesting things in that I think there are lots of companies that manage to you know figure out the content organ-grinder where you're just churning out the the ad scale stuff I don't think we have figured I was an at scale way to to produce interesting content stuff and so it's still a little ad hoc right now but hopefully scale that up excellent John right here first thank you to you and your brother for supporting Michael confer so it's great

24:21to see you guys here year after year my question to you would be where do you see the mobile checkout experience going over the next few years yeah this is a super interesting basically we think Apple pay and Android pay are gonna win in that both are very good at being that kind of large company phenomenon of being patient and so even if the first version of the product isn't quite right they're gonna keep on iterating until they get it right and then we just have the data and we look at it and they're they're growing like weeds and the conversion rates are way higher and so I'll be really surprised

24:54if that's not what everyone is using I mean aside there's some kind of exceptions to that where if you are Amazon you now have enough stored customer credentials that you don't need that but and if you look as if we want to look at on the on the consumer side is wish they are conversion optimization machines they are just really really down the details on this stuff and their funnels are kind of scary fast because you guys probably seen their ads on Facebook but it just like Hoover people in from Facebook at scale and it's it's really in their interest to have a highly converting funnel and I think

25:26those are they're gonna be big what I'm excited by is on the mobile side and new products actually being possible because the mobile checkouts will get easier enough and so you know occasionally people will treat a link from The Wall Street Journal or the Financial Times and I'll click into it like hey well never mind and you know just the bounce rate is 100% in my personal experience or as if you could actually quickly buy a day pass or something I think it probably would and so there are similar you know you can't today buy an issue of The Wall Street Journal online just because the amount

26:00of work required is is not worth it for that experience I think that will probably change as the checkout to get better now it's interesting to think are what are what the second-order effects will be higher conversion rates is one but at this probably a bunch of new products that are possible that weren't previously next question you know one more of a year high um so it seems from what you're at the way you talk you and your partner on a very much a unified front I was actually wondering what is the decision-making process between you and Patrick behind the scenes if you don't mind me answer yes and so I I

26:39think for a good decision-making process like the one Patrick has for that to work you need a few things one you need to be in in terms of the long-term vision of where things are going and kind of the values and how you look at the world you actually have to be pretty aligned there and that's surprisingly hard at times and that solves a lot of things so Pat and I are pretty aligned on you know what stripe is broadly what we should be doing over the next you know two or three years over over the long term what we wanted to become and that's pretty useful to

27:14begin with second I think you need to be able to distinguish between the the things that matter and the things that don't basically and that sounds again really obvious and then sort of tautological but it ends up being pretty important so there'll be many things that Patrick and I on a day-to-day basis kind of disagree with the others decision but that's kind of I'm happy to go with it and then Patrick and I probably more than most co-founders have have a fairly overlapping set of areas of interest and so I think the you know one of the traditional things you see a lot at least and you know the Bay Area

27:44where I am is you know if the technical co-founder in the business co-founder and they're the delineation of responsibilities is is very very clear and crisp with Patrick and I it's it's it's more overlapping but you still have kind of this is my area and this is your area and so basically high trust when people are operating within their own areas and kind of overturning when it matters for those who don't who aren't aware of John and Patrick or brothers as well yeah we've been resolving conflict cipher for quite a while I can imagine that's difficult yeah next over here

28:23I thinks in any thoughts or plans for stripe regardless or on crypto currencies or bitcoins or stuff like that and is have any you guys asked Patrick McKenzie about that I was annoyed that but um for me I think so strategically the way we look at stripe is it should be you know the the way a business accepts money from its customers and that's obviously fairly broad and deliberately so and so if you're a SAS business we should have a very good support for all the SAS primitives and the things that a SAS company cares about when it comes to payment methods we're kind of like

29:04whatever we need to support whatever these businesses will will need and so we launched support for you know ACH a while back we're a beta testing a bunch of the local European a payment method said matter to people there and we'll kind of continue adding all this stuff and so with cryptocurrencies from a from a pure pragmatic point of view if if if and when any of the cryptocurrencies gets a significant consumer mind chair and consumer scale then we will support them because the businesses will want them but uh we're kind of not dogmatic about it and that I don't think we want cryptocurrencies to happen for the the

29:40sake of the cryptocurrencies happening and even if we wanted that I don't think would help really advance the cause of cryptocurrencies anymore and so we we added Bitcoin supports because we wanted it to be you know some of our the businesses on stripe cared about it and we want to stripe to be the place where they good regardless of the payment method accept money from their customers and you know obviously Bitcoin is still not in wide usage today or or even you know cuz it's you know fairly small and meesh today that there are a bunch of you know we can spend time talking about the interesting computer science

30:13properties and some of the interesting stuff that's happening with some of the newer cryptocurrencies but I would say it's not you know top of our 2018 strategic planning list if I could take tack onto that a little bit do you think of it in terms of like the innovators dilemma as well because I mean that's kind of an up-and-coming thing but you you know it doesn't have any material impact on your business right now but is it something that you're not us not afraid of but concerned about or at least keeping it on your radar specifically because of that yeah I mean we do worry about kind of innovators

30:46dilemma dilemma phenomena in general where you're always trying to look out for what is the good enough solution that you know you will right off but actually satisfies customers quite a bit I don't think for a business selling online I don't think Bitcoin is good enough yet and so it that's not how we think about it yet but who knows well I think yeah thanks again John thanks

31:17you [Applause]