How to Scale: Lessons from Stripe CEO | Patrick Collison

so what I wanted a Wan to focus on mostly is point out real quick before he starts I think there's a kind of a cruel irony in the organization of this session and that it is true that one of the people on stage here has 20 years of experience building company cultures but I think the directionality of the interview might be you can turn it around describe new mistakes we'll talk about AEI which you know much more about than I do so what I wanted to focus on because it be most relevant to them both the greatest asset of the audience is both recruiting and a culture and there's obviously a relationship between those two things stripes been amazing at creating and forging a reputation among developers and technical talent as the go-to place in Silicon Valley particularly among startups what I really wanted to ask is how did you start how did you build that at first now that you have a reputation and a brand and a lot of traction it's a lot easier but how did you do it when nobody knew who we were um at the time I guess we sort of hadn't figured out that there was necessarily a particularly good strategy to accomplish the whole thing although I guess in hindsight I think there are a few things for a sort of we lucked out um I think the first one and this is something that I think has maybe become a little bit tear over the last couple of years as we've sort of come to see a bunch of companies be successful at it and that it's in some ways easier to build a really good team if you're solving a hard problem that an easy problem and I think even more broadly there's almost a way in which it's just easier . to solve a hard problem than an easy one and so at the time like when we were starting stripe in early 2010 it was the era of sort of social games and there was kind of a bunch of semi ephemeral or at least ostensibly kind of lightweight business models kind of percolating around the valley right that's we were pitching people on this like infrastructure was gonna be hard there were legal problems they're going to be international expansion problems all these things right and I think first sort of the best people that's actually kind of more attractive the idea of building something really durable than the proverbial next recipe amp right um and so I think that certainly helped out a lot the bigger one I think actually was we we actually should have started stripe gradually we were in school most working on us we weren't sure if we wanted to work on a full-time or not and and we use it even once we did so working on at full time was going to take us quite a while before we could launch it publicly and so that effectively pushed out our time horizon in terms of recruiting and so you know this kind of the great Jeff Bezos quote about how you can just use time horizons as a competitive advantage in the if you were operating on a five-year seven-year time horizon and your competitors are operating quarter to quarter they're sort of a set of strategies that are open to you that aren't open to them I think the variant of that that was kind of again not necessarily consciously at the time but in hindsight kind of opened to us where because we knew there was going to be say almost two years before we launched we were sort of hiring on a two year time horizon and everyone else was hiring for like almost literally next quarter right and so it took us almost two years to recruit our first seven people right and so in terms of like people per month that's sort of an atrocious record actually if you if you you know discount myself and John's actually five people in almost two years and we talked to a ton of people and we sort of they started working with some a little bit and then to figure there wasn't quite the right fit and so forth and such as that I think really made a huge difference and and then I guess maybe thirdly I I think people try this in various ways and so forth but I think that it's kind of critically important or at least extremely valuable if you can thread the needle to just work with people before definitively deciding one way or the other and so there's like all these IP issues and sort of potential pitfalls I think if you if you kind of go about things that way we decided to just kind of run the risk and so there were a bunch of people who we sort of did a trial week with or a trial a couple couple weeks things like this right and I ended up deciding not to work with them and we have to sort of you know again be careful on the paperwork side of things there but it enabled us to not make you know bad hires we might otherwise have um if your that idea is well pause there for a moment yeah so one of the things you generally advise people is to really obsess on your first ten hires because each of them is going to replicate themselves ten times so what was the criteria you were thinking the back of your mind when you met those people and they were looking for one of those top ten at first ten positions right like you know III guess it's kind of all the the standard things obviously you want someone who's extremely smart and pleasant to work with and whatever else I think the things we sort of particularly over indexed on were and you for the first ten people I think you want people who sort of will hit the ground running and already know what they're going to do in that I think it's right to sort of take bets on on sort of inexperienced talent and have them develop with a company and things like that right but I at least more perspective for the first ten people we sort of knew where we wanted to build and we did not have time in the first couple of months to be training them basically and the sort of the well one way in which I think it's kind of helpful to think about startups is you want to be sort of making decisions on a time horizon roughly equal to kind of the longevity of the company right and so kind of after three years you can afford to think on a three year time horizon but after six months you you sort of don't get to look at that much more than kind of six months ahead right and so maybe nine months in this company were thinking how can we maximize the productivity for the next nine months and so they had to be people who were sufficiently experienced that they could be sort of really productive in that period and a somewhat personal one but I think now the were sort of 250 people one that's kind of paid a lot of dividends is you know lots of companies have the proverbial no rule I think obviously that that's a good thing to do but but we tried to hire people who are sort of like particularly nice and just generally happy in that you know if you're to describe the characteristics you want to do your company to have right you one of the adjectives you probably ascribe to the culture well if this personally speaking is happy right but it would seem sort of it seems almost like a tough startup in itself to figure out how to sort of create a culture that like ingests unhappy people and magically makes them happy right and you know if we succeed at that we probably something even more valuable and so instead of sort of trying to figure out how to solve that at the culture level we just try to hire people who are intrinsically happy already and kind of cheat in terms of getting to the answer so we really sought that out and then and then I guess thirdly it wasn't sort of merely enough for people to be good but they had to be sort of known to be good by others and this kind of comes back to the the 10x point in that if you sort of in hiring the first 10 people if you sort of visualize your company at a hundred people and like by and large those residual 90 people are going to do a kind of a disconcerting degree the the subsequent 90 are going to be determined by those say other eight people you hire and and so you don't just want them to be good you want them to be known to be good such that they're going to attract you know a bunch of other super talented individuals alongside them and you know that really played out ok again comes to the time horizon thing there are multiple people at stripe like somewhere between maybe 12 and 24 who took who took us in each case years to recruit at there are multiple at stripe who took us literally now more than four years to recruit and when you're starting out sort of it seems I was almost useless to sort of be investing in things that might or might not pay dividends a sort of on that time horizon but the way we can uh nationalize it was that well we're already sort of we already have sort of taken the batch that we're going to succeed if we don't succeed no nothing matters so we might as well assume that we are in which case recruiting people at a multi year time horizon is important and gonna work tat so let me double click on a couple of those points so first of all you actually explicitly allow any employee to veto a new hire that is true and well inevitably it sort of you have to circumscribe it a little bit as you get larger just because not everyone is involved in every hire and but but in the beginning obviously way more costly to make a bad hire than to the pass on a good one and so vetoes are really important and we also sort of tried to give people as much opportunity kind of as possible to spot a characteristic that they might want to veto and so this kind of comes back to the you work with him for a week beforehand or something like this I think that I think this is the case for literally all of our first maybe twelve or thirteen people we worked with for at least a week before we made them an offer and so people had I mean they had lunches dinners everything with them and in many cases we had them stay with us for the first week that's not always an option but we really tried to self-assemble as much data as possible such as it folks could could spot anything didn't like and then about 20% of your company founded a company previously yes actually I think it was almost all of the first ten people and I had started companies and III I think yes every one of the first ten and and this kind of comes back to the kind of being known and hitting the ground running we had sort of a sense for the the like what we wanted to build on a one to two year time horizon and the question was just how can we do that most you know expediently and if we were sort of trying to task manage people or sort of just trying to hold I mean we were not experienced managers herself we sort of wanted to kind of specify the end state and then have sort of self-organizing people who would achieve that as quickly as possible and and again once I think you're doing something hard and sort of important and Marut founders and that by and large their sort of their goal function is something like you know work on hard problems be in a role where I can sort of define the future light cone of the company with the best people but it's for most founders at least that I know the founding bit is not actually critical it's kind of it's a proxy for those other characters there's a cultural ownership characteristics right the other thing you mentioned that a lot of founders don't do but you're sort of unique odd is you will recruit people that aren't that interested in stripe initially many founders and many companies exclude people that initially don't show enthusiasm and passion how do you think about that so am I guess there's a few ways one is it in some way I think it's I think it's almost kind of anti-technology to to insist that they're excited about you in the beginning what I mean by that is you know that there's the this kind of a great observation from Alan Kay that if you're building something that's truly fundamental technology kind of almost by definition you can't really know all of its kind of end consequences right in that if you're working on the transistor you cannot predict Facebook right you just know there's kind of something interesting and important here and so I think that you need people who are sort of locally interested in sort of the short-term direction because I understand that and I mean that I could be able to make breakthroughs right and you need people who are excited to work with with the other people I think that's you know that the most salient part but I think that a if you're working on something that's kind of again your fundamental technology in some way you actually can't know that much about the end of stage I think you always need to describe to them how that's part of what's exciting about it right and so this the some element of that I think also in a world further sort of where there are so many interesting kind of technological vectors I think I think it's a little bit arrogant to assume that sort of yours is immediately and obviously going to seem the most promising instead I think you kind of need to come to understand what do people really want I think normally it's the sort of build a great and enduring culture in as much it is to sort of achieve a particular end State and so I I think there's am I think had I had a characterize it but I I think there's a certain vanity on the part of founders instead of presupposing that all the folks they hire should have this immediate attachment to an end stage that you've spent you know years thinking about and sort of had years that have get yourself excited about and assuming that we'll sort of immediately come to be understood by the folks that you want to hire and then lastly I think people underestimate sort of how once you kind of submerge yourself in a problem it becomes more interesting right and that yeah I know from conversations with you that you sort of weren't immediately enamored with venture capital right and now I think after having spent you know so many months in it you now get the depth of it and sort of the intellectuals awesome here and so the intellectual attraction but you know when you hear BBC said of on day one you don't immediately get the full resonance no that's true I wasn't interested in payments even when I joined PayPal we had and idea that I would spend a lot of time doing payments you also I think blogged about a while ago the distribution of equity among the first ten employees I will talk about that yep and this is actually in part advice that I got from Sam Altman and so I think for him it was made just an offhand comment that somebody done stuck with me he said that I guess given the audience I'll have to maybe modify a little bit but something like M be extremely generous with your employees and be less generous with your investors and I think he used a more aggressive term than less generous but we're out the coast of Summit and so I you know I think it's one of those things for when you think about it it's almost self-evident right in that if you are fortunate enough to sort of succeed in some major way just kind of picture yourself there on IPO day or acquisition day or whatever the case might be and you know you're definitely not going to regret sort of a marginal couple of additional percentage points that you've given those airlie employees and the people who kind of built the company with you and and and so I think it's not really so much a kind of a necessarily sort of a financially motivated decision so much as something kind of more personal right in the juice you really want those people to do exceptionally well I think there are kind of some benefits to it for you know they themselves will sort of identify more as being founders of the company as it's not quite of a binary state it's more sort of a continuum and and so we for our first ten people at stripe we we gave away more than more than ten percent of the equity which you know this kind of sparse data on this but I think is a typically high and and all through the company we've generally tried to be and actually now we're sort of putting you know proper comp frameworks in place and things like that we've we've had to generally be not exactly sort of stingy but but noticeably sort of less aggressive than other companies on salary we sort of aim for no more than fifty percent of kind of median for the role we're just not going to compete there we don't want the people who for whom this would be kind of the the the deciding factor and to really consistently over-index on equity we got a comp consultant to sort of analyze our cap table at some point and her observation or primary observation was like good god you've given away so much equity to the employees right and so you know that that's true and then on the other hand I think are again the data here is sparse but at our tents that are retention of those people who've now been with us for like four plus years and is is much higher than normal right so a couple times you alluded to the culture and importance of culture and recruiting other observe that you have a pretty unique culture how would you describe it I think these things are you know it's always in in some way so kind of ineffable and hard to put your finger on that I'm not sure I can be really sort of trusted to characterize it and you know there's kind of obvious sort of phenotypic things we do that are different to other companies like one of the ones we've kind of talked publicly about is this idea of email transparency and we're basically all kind of essentially all internal communication is public throughout the company so that if you're you know interested in what the legal team is working on are the partnerships team or who knows you you can go in you can sort of get that context and the thinking behind it is that you know we want people to sort of make the best local decisions and ideally without us having to sort of you know manage the overhead of sort of trying to get them to understand the overall framework things would be much more efficient if they can just sort of pull as needed and sort of when they're making a decision on I don't know which bank they're going to integrate they can go and look at the sales pipelines of different countries or whatever the case might be and and I think it's actually been you know quite effective there and so this is I guess transparency motivated by a desire for kind of federated understanding right and I think there's also a way in which you know in general one of the things we've sort of really pushed in the engineering culture is sort of you know the traditional view of engineers in the valley is very much sort of engineers as implementers right you've the product requirements document you've the product manager you've the market research you you you've all the things right then the engineer sort of types out the Java's and we've tried to take a broader view of what an engineering role should be right and actually we only literally on Friday hired our first reflect manager at 250 people and we basically said that the engineer should be doing everything end to end they should be talking to the customers they should be designing the product they should be building the one of us they should be showing that to the customer is going to the office getting the feedback into the complete feedback loop right through to writing the blog post at the end and so this was kind of motivated by just taking a the the the there's a tendency for companies I think to take this kind of simplifying abstraction of people are a particular thing and a particular function and sort of that's that's a simpler kind of awesome with which you can assemble the corporate molecule and but of course people are people and not just kind of their specific functions we sort of tried to allow for that a little bit more and so part of the motivation we have the email transparency is even if I'm an engineer I might be super curious about what the legal team is doing it just like how does a legal team even work right or or what does the PR team do or whatever the case might be right and so it's it's partially a desire to them to kind of say to people's curiosity and and and indeed support it and I guess the other thing I mention of the culture is we've always tried to be yeah yeah these things are all going to be so so sweet generous and kind of particular to characteristics of the early people that I'm not sure this is kind of that much generally you can say but but one thing I think that is generally useful is to just be pretty deliberate about it and it's kind of easy to be deliberate about it in the early days because there's only ten people who talk about what you want and you know you're all sitting around over dinner anyway I think it's a little bit harder to be deliberate and to kind of have it be something malleable as you get larger and something certainly something we've grappled with grappled with and I think one is useful to just sort of get into a habit of talking about it at pretty early so that people can make observations about us what's working well what isn't and so forth but it's sort of seen as kind of a product of the company like your actual products right and then to sort of think about what are the ways in which you're going to how are you going to sort of modify the culture over time just what our kind of your habits and patterns for doing that and so for example it's something we've just started doing is this thing it's kind of a terrible name and but but org hack where basically will assemble every couple months a list of the kind of source from the employees just like the top organizational problems within the company and and then we'll sort of have subgroups of them sort of go away and think about you know ways to improve it or remedy it or whatever they'll come up with kind of recommendations will go trial those will see if they work out or not or whatever right and you know I think that's maybe a you know a useful thing for us to do at our size but sort of whatever the particular implementation might be I think having a deliberate answer is that if when somebody asks you how do you modify the culture of your company knowing what that is I think it's actually kind of pretty important so there's also a tangible example you have a very unique onboarding process can talk about that a bit um yeah so well actually going through the process of modifying it right now so it might be very different in a couple months but and the the the main thing that we sort of really tried to do in the beginning is to sort of not do what seemed to happen at Twitter and and in particular what I mean is we had a bunch wheel join a stripe from Twitter and just kind of described problems where there was a huge amount of siloing between the teams right I remember one guy saying to me that no one had ever left his team for another team in the company the the state was always either joined the team or leave the team right and so sorry I should clarify like when I say don't be like Twitter I mean it kind of specifically in this regard and so we're kind of terrified by I would love to have tourism market cap and we've sort of we were terrified by this and that sort of as as we divide it up into teams and that sort of that didn't lose kind of the the collective and whole of the culture and and so we we started out by doing is that every engineer who joined em would go and ship something on every other engineering team before they went and joined the you know the team to which they were sort of a going to be on long term and then in general we sort of tried to encourage it as much as we could this is switching between teams right and that even if it wasn't kind of globally optimal for the company because the person had a ton of state you know in this area not in that area whatever the case might be we thought there was kind of long term culturally good for there to be a lot of sort of a cross-pollination in this way and so now we're set about the scale where there's enough engineering teams in enough context of that kind of getting a fairly difficult to do and so I think we'll probably end up with something more like Facebook's boot camp and where this kind of still a designated period where you're not on your sort of again the long-term assigned teams that you you sort of gain broader context across the company and though we haven't implemented that just yet the other thing I'll just kind of briefly flag is I think it's actually we didn't really notice it at the time but it's become clear in hindsight I think it's really valuable for there to be a time delay between sort of when somebody starts and when they join somebody's team because it helps correct I think one of those prentices forces in hiring where there's kind of a principal agent problem between the team and the company the team wants is sort of the team is overburdened they've too much to get done they sort of really need more people right and so kind of their incentive is to to hire more people even if potentially they're sort of below the kind of the global average of the company and and then of course on the company standpoint you you don't care anymore than sort of you know a relative basis about the the the you know what's going on instead of a particular team and so I think having the time delay there means that it's much harder to kind of quickly hire people to solve a short-term problem right you're not going to be able to sort of quell the the hair on fire issue today by hiring more people and so by imposing you know we accidentally I guess imposed you know one to two months delay I think now actually that perhaps should be even longer that it's like three months between sort of the you know the requisition request and when you actually got someone Wow alright so each stripe also deferred hiring external managers for a very long time I mean probably 150 200 employees when you serve added your first external hire what was the signal to you that you needed to start changing the management layer um so we weren't one of those companies that believe that we shouldn't have managers or never would like that right I remember one of our engineers edited our jobs page at some point to say stripe didn't have any managers and we like that day we went and reverted it because kind of wanted to wasn't true and that arguably I was already a manager and and to just kind of seemed naive and you know we certainly were going to end up at that point we didn't want attract people who we're coming for that reason but I think was um I think was largely a consequence of hiring founders and sort of people who already knew what they were doing and again could sort of hit the ground running in that I think that if you successfully primarily hire those people they just don't need a lot of management right i I think sort of amount of manage when people require is is almost an inverse proportion to their level of experience right and so that just there wasn't a whole lot to do and I think if you hire a lot of new college grads or sort of people new to the domain or things like this then just in practice third there is a whole lot more and so I mean we eventually hit the point where sort of it was necessary but I think we rel defer it for for a very helpfully long time right in that it meant we sort of didn't need to you know deal with a bunch of management overhead in the early days it also meant that we were able to preserve the kind of people who would make great managers as individual contributors for a long time right in that there's this guy who manages our rml and data teams and he's now a you know a great manager but because we didn't need managers for so long he was actually able to be like an independent contributor on these teams for like several years and that's basically like having you know the lower LeBron James of data infrastructure just like building it for you right it was it was really great and so once you decided you wanted to add some management how do you go about the process of identifying who'd you want recruiting them I didn't you know assessing them so I guess it's kind of two sides to this it's like the management team and there's kind of management middle management throughout the company and the the middle management slash I guess the non management team and it still kinda makes me sad because I'm very happy with the people we've ended up in these roles but by and large we we promoted from within and which I guess is sort of what people you know say is is the better one to do what they don't tell you as much or die I never heard was that that basically means you're sort of snipping out your most productive people throughout the organization right and so effectively we went and fired our like ten most productive engineers and we fired them into management and so I'm still a little bit rueful about it but that's essentially how we went about it one kind of nice side effect of sort of how long we went this was there was a tremendous kind of an degree of kind of moral support for the move in that it maybe if we done it at the sort of the you know normally right time and I maybe with 60 people or something that people would have questioned whether or not we actually do this because we've assumed at this moment for so long people were like quite enthusiastic about it and then just briefly in a management team and the one thing I had sort of quickly flag is in my experience entrepreneurs don't start thinking about it sort of nearly early enough I certainly did not start thinking about it nearly early enough and like I remember somebody asking me even or a 50 people or so like what's your management team like and I mean what it didn't exist and to I hadn't even kind of particularly thought about the question before um whereas in fact you know I mean the management team is kind of the the entity through which you exercise most of your daily influence over the company right and so I think sort of deficiencies in that or any kind of anything short of sort of of optimality is actually a huge cost to the company and I think that sort of in practice I mean you know you you can sort of them take kind of functional cuts and I have founders or executive should be spending the time like on recruiting or in product or whatever but I think there's also just kind of a a general kind of maybe a horizontal cut of looking at this through the lens of the management team right and I think you know it does not sound it all crazy for me to me for a founder to be spending a third of their time on just building that and that might involve recruiting or training or one-on-ones or whatever right but but just kind of getting that into the right state so about I don't know six nine months ago you added a very senior business operations hire how did you think about that process so this came from around that time when I sort of started realizing kind of the oh moment on the matter one team and and so spent about a year just meeting as many people as possible who seemed like not even necessarily with particular roles in mind but just kind of I didn't feel like I had a particularly well calibrated sense for what sort of a really high performing business leader was like and then I felt that maybe after meeting kind of a couple dozen people I could sort of start to see maybe the the deficiencies in our own team and and as that happened sort of after maybe again a year of this process it both became clear that we had to a major deficiency in this area we just kind of hadn't hired senior leadership in terms of sales account management support sort of user operations and things like this and technically that the person who kind of leaving our particular kind of exigencies and needs aside the person who I just thought was most impressive was this woman at Claire Johnson and so it kind of it matched up pretty nicely there but we were pretty terrified about I talked to some people at Facebook and it sounds like when they sort of hired Sheryl that even though Sheryl has been sort of such a spectacular success for them that it was like a pretty rocky kind of early months as first couple months as kind of the adjustment happened on both sides um and so we probably spent I probably spent maybe like 50 hours with Claire before we sort of decided that you know this was the right thing for the company and luckily it has been awesome well we're basically out of time so thank you for joining us thank you having me cool you talk about recruiting and team-building from 50 to 250 if you spoke a lot about 0 to 10 yes and so so the big thing we spent a lot of time on be kind of post 10 between 10 and 50 was building a good engineering brand and I think that a lot of not a lot of companies take this kind of X seriously as sort of an explicit kind of a goal item and and there were a couple things that influenced us one was M there was a very small company out of Boston called K splice they'd built sort of live patching technology for the Linux kernel and we'd hired a couple people from it and and they had they hadn't raised any VC and they were in Boston and they sort of they did a couple of others that have disadvantages and so they decided that they I would I went to school in Boston so I can criticize it but I'm they because I know they were sort of you know weird and difference in a few ways they decided that they sort of needed to compensate for that by taking a building a brand sort of really seriously and they published all these like really great engineering blog posts about I mean things that were only tangentially related to what they did but had a very powerful effect of just showing how smart and competent they were and counted the point of that being the kind of high order bit in who you want to work with was actually extremely effective and you can sort of do all the stuff about convince people I live Linux kernel updating technology is cool kind of later but but they just made sort of really powerfully clear and just how how high the kind of degree of talent was and and we sort of figured that you know if this is a passage in in creativity Inc about the Ed Catmull book about a sort of you know in the do is between certain people and ideas which is actually more important and and he sort of you know generally people sort of say well you know it's kind of mixes the two but he kind of powerfully comes down on the side of no it's just the people that is the answer and sort of if you if you take that as being true then I think kind of you you're I mean recruiting in general and you're recruiting brand in particular sort of more obviously become sort of a top-level kind a goal for the company and so we decided to invest quite a lot of time and effort in in building this no one had ever heard of us before and so we just took it as a goal item and so we did things like we did capture the flag security contests they took you know each we've done three so far that probably took a each one PI took like a man year of engineering time and so these were like pretty significant investments we wrote random blog posts about just like cool engineering things if you go back to our I mean now our blog is kind of had you know product improvements or something if you go back to the early days of our blog there's a lot of sort of random stuff about just things we thought were cool but that have attracted attention instead of these particular areas and anyway so that was the real really the goal we knew that we couldn't sort of invest like the same degree of kind of care and time in you know employees 100 through 110 as we did in in 0 through 10 but we tried to ensure that we sort of had the the reputation in general and that you know such we could perhaps compensate at least a little bit for you know where we am we're for going in terms of assessment ability in terms of increasing the quality of the applicant pool and you know I think if an obviously doesn't just apply to engineering cultures but if you think that you have sort of an outsize capability and some access or in some area whatever I think it's really to your advantage just sort of to pretty aggressively kind of make that kind of broadly known and and and public in that you know as an engineering culture we're just kind of knew that the vast majority of companies just like could not match us and the things they were doing it they can't sort of turn around and start writing sort of equivalently good engineering posts or running accordingly good security CTF that is like literally did not have the engineers for us and so yeah I I think sort of in whatever your kind of competitive domain is doing something like that is probably likely to be pretty effective okay thank you

Channel: Khosla Ventures

Views: 35,870

Rating: 4.9208632 out of 5

Keywords:

Id: fPfYN5gU_C0

Length: 34min 43sec (2083 seconds)

Published: Wed Jul 15 2015