We don't know how to meet new founders. There's only so many Elons in the world, and that's the limiting factor. You either have a superpower or you don't. You need to be in the top one to 15 basis points on some dimension, or the chance you're going to build the next Nvidia really is zero. Everybody's a VC, everybody's an angel investor, everybody's a seed investor, chasing after the 5, 10, 15 people a year that have a non-zero probability of rearranging the planet to their will.
Winding up here, that is Keith Rabois. It's his second time on the podcast, so I'll keep this brief. Today he's a managing partner at Coastal Ventures and CEO of OpenStore. Keith is a bit of a legend in the startup world. He's an early investor in names like Airbnb, Palantir, DoorDash, Airtable, Ramp, and YouTube, and has operated at places like PayPal, LinkedIn, Square, and as the co-founder of Opendoor. This makes him a pretty opinionated encyclopedia of knowledge on how to run a company, with the rare experience of having been on both sides of the table. I found myself referencing Keith quite a bit, but realized there wasn't an interview that tried to concentrate many of the most important of those practical lessons in one place, so we sat down and gave it a shot. We spoke about the Hollywood model of startups, picking co-founders, identifying, attracting, and assessing talent, raising capital, building a board, delegating, metrics, and knowing what to measure. This is the Lao's Here podcast. We choose to go to the moon. We deliver right to the desktop. We're going to reinvent the phone.
The real, real higher-level question I think is the bottleneck to innovation and startups being created in the first place. So in your opinion, which part of the startup value chain is the constraint from us 10xing the number of exceptional startups?
It's really straightforward, really simple. Capital. Yeah, capital. There's too much capital chasing too few founders. The scarce resource is world-class founders. People have the potential to be a world-class founder. The idea that you can take a company from scratch, overcome all the inertia, and reinvent an entire industry or change the planet is borderline ridiculous actually for most people. And so there's a scarcity of people who are kind of crazy enough to actually embrace that challenge and also have the Venn diagram overlap with the skill set, talent, superpowers to pull that off and create the iconic companies. So it's a little bit like major league baseball. There's only so many people in the entire planet that can throw a fastball over 90 mph, and there's only so many a fraction of them that are left-handed. And if you're not left-handed as a predominant athlete and you can't throw a fastball over 90 mph, your chance of actually playing in the major leagues as a pitcher rounds to about zero. On the entire planet of 4 billion people, we still have not figured out really how to train somebody who isn't throwing at least 90, 88, 89 mph how to get over 90, despite all the advances in science and cameras and technology. And that's true of founders. We don't know how to meet new founders. There's only so many Elons in the world, and that's the limiting factor. Everybody's a VC, everybody's an angel investor, everybody's a seed investor, chasing after the 5, 10, 15 people a year that have a non-zero probability of rearranging the planet to their will. So if you're a founder and looking for the things that move that needle from being zero to nonzero, you talk... I don't think you can. I really don't. You either have a superpower or you don't. You can have a superpower and waste it. Lots of waste talent, God's gifts, intelligence, you know that happens all the time. Happens in sports too. There's athletes who have potential who get injured, there's athletes who have potential who use drugs and get distracted. But if you don't have a superpower, your chance of creating this company is literally zero. Now there are ways of finding out whether you might have this superpower or magnifying or amplifying the superpower, but you really need some superpower. You need to be in the top one to 15 basis points on some dimension, or the chance you're going to build the next Nvidia really is zero.
Okay, before we go off the paper here, you've done a pretty good job of summarizing your philosophy on where to build. This is a quote: 'Find large, highly fragmented industry with low NPS, vertically integrated solution to simplify value product.' Break this down. How do you identify those pockets and why integrate vertically?
Yeah, so fundamentally this is my pinned tweet, so anybody wants to consult it, just go to my X profile, you'll find it right on top. The basic, I wouldn't say this is an exclusive formulation, I just think it's a common denominator of many, many, many very successful startups. Vertical integration. Vertical integration is the key to building a trillion-dollar business typically. There are some exceptions, I'd say Meta, Google are not vertically integrated, but most are more vertically integrated. And the reason why is it's very difficult to get vertically integrated, to build it takes more capital, more time, more talent. But when you build vertical integration, nobody can compete with you for usually 10, 20, 30 years. Let's talk about smartphones. The only profitable smartphone manufacturer in the world at any scale is Apple, and it's because they're vertically integrated. Nobody can do anything about it. Amazon can't do anything about it, Meta can't do anything about it, Google can't do anything about it, because they don't really know how to do vertical integration. Vertical integration means they've integrated the chips, the software, and the hardware to be completely... not even symmetrical actually, create accumulating advantages by integrating the combination. So the chips are designed for the use case, the software is designed to take advantage of some of the dimensions of the hardware, the hardware is designed around the limits and advantages of the software. Those are all reinforcing, and it's going to take 30 years before someone creates a competitive device. Probably what happens is devices go away before someone figures out how to compete with Apple successfully and build a profitable device.
In a lot of ways, the first weeks, the first month, the first six months of a company's existence are the very, very most important in determining its trajectory. And in this context, you talk a lot about the Hollywood model of startups, in excruciating time and detail identifying the key solvable variables that will affect your company's trajectory, then recruiting co-founders that burn down those core few key risks of a company, like casting for a movie. Is this fundamentally at odds with the common sort of Silicon Valley advice of essentially talking to users, throwing stuff at a wall, and seeing what sticks?
Yeah, 100%. So it's a bit like the canonical question: is mathematics discovered or invented? I don't believe startups are discovered. I believe they're invented. You forge it through willpower. So it's like a movie. You have a vision of how the world could be better, how an industry can be better, how a product can be better, and then you execute across that vision. And that requires producing the movie, which involves capital, casting, and then creating trailers, which is basically your marketing, and selling tickets, which is really your revenue. And so just like a movie, you start with a vision of how what could be, and then you turn that into reality, and then you hopefully turn that into both a critically acclaimed movie and a blockbuster.
When starting a company, how should you think about the mantra 'why now' versus micro timing? So what I mean by that is Amazon was started only a year after the World Wide Web was released into the public domain, but Google is the 11th search engine. How do you reconcile?
Well, I don't think 'why now' means like are you first, second, third, fourth, or fifth. 'Why now' is what enabling technologies, what enabling sea change has occurred in the world? What's changed? Because if you think about this concept of inertia, inertia is your worst foe as a startup. As a founder, the world just isn't going to suddenly embrace you, and you need to create momentum literally in a physics sense, so inertia into momentum. If you can identify a crack in the world where things are in a sort of volatile or things are in transition, there's often an opportunity to ride a wave of momentum that you don't have to create from scratch. Creating your own momentum from scratch is either very painful or very expensive. So if you can identify someone else's wave to ride, a little bit what happened with Nvidia. They were smart enough, they were really doing gaming chips, smart enough to see somewhere in their data that these AI consumption use cases were starting to become important, and then double, triple, quadruple, 10x out. And everybody else, the Intels, the AMDs, completely missed it. So they were propelled along, among a lot of other things, but that wave of AI starting and noticing that wave versus everybody else ignoring it with blinders enabled Nvidia to be the company that we see today. So the 'why now' can be the beginning of the AI wave. The 'why now' can be people are going to have a computing device in their pocket, think about the mobile wave. But to try to identify where the worlds that are kind of in transition and then trying to capitalize on that, you can create startups from scratch through willpower and create your own momentum, but it's a hell of a lot easier if the world is changing, in a flux, and you're taking advantage of capitalizing on that.
Yeah. You talk a lot about integrating vertically, obviously, but how should you know whether or not to go multi-product? Is that something that should be baked in from the outset or something that is responded to the market?
So I think it's opportunistic more often than intentional. Meaning you start with a product, it resonates, you have product-market fit, you're doing really well successfully as a startup, and you see there's other value propositions that your customers are demanding and they're logical extensions of what you already do. That's the best way to build multiple products. The ways that are less attractive but sometimes happen in SaaS businesses is the economics of your business aren't very good, they're mediocre, so your payback time on sales and marketing against your revenue can be acceptable but not exciting. And so that drives from a top-down perspective the CEO to say, 'I want to have multiple products to kind of expand lifetime value.' That's a lot less exciting than when it's fairly organic and the customer is basically saying, 'Please help me solve this other problem. You've been excellent at what you do. I've got this problem over here, and there's some synergies between what you've already created and what you're expanding into.'
Got it. There's a question of how a lot of this sounds in theory versus how it actually plays out in practice. You back a lot of really successful companies at the earliest stages. Of the most successful, call it top 10, how many of them roughly scaled and executed against the same key risks and on the same idea that they identified at the outset?
I'd say 70%. It also depends upon when you're going to demarc and start measuring from. From time of raising seed capital, it's probably 80 to 90%. From time the founder started the company, it might be more 60. But by the time I intercepted many of these companies as the early stage or first institutional investor, angel investor, it's probably 80 to 90%. But if you dialed back before they actually pitched me, it might be as low as 60. The ones that have succeeded in evolving, I think it's more like the term people talk about pivoting, but they forget the derivation. Like in basketball, you're only allowed to pivot if you keep one foot on the ground. You can't move both feet. So a pivot where you have a foundational principle and you're editing your direction based upon that foundational advantage can be very attractive. So for example, you can talk about the origins of PayPal. PayPal was 'we're going to be money from Palm Pilot to Palm Pilot company.' And the foundation of that though is we needed a unique identifier per person per user, and that was an email alias. So the email alias was the critical ingredient. And then as the Palm Pilot world wasn't really exploding as much as forecast, and the use cases back in 1999 and '98 around beaming money to each other weren't so obvious, people did recognize that the email alias was a more universal, more approachable, and more broad product extension. And that was the foundation of PayPal, was the email alias. And so it wasn't like completely abandoning the heritage of the company, right?
Okay, let's talk about starting a team. You said the biggest lesson you've learned from the No Costas is that the team you build is the company. I mean, you know, like any good advice or wisdom, it's something that you know is true, but when it's expressed, it becomes that much more impactful. That said, team building and ensuring that the foundation of a team is as strong as possible, you've said it's like working with concrete. When it's liquid, you can shape it, and once it solidifies, you have to take a jackhammer to it to fix it, which is very noisy and expensive. In the case of co-founders, that fix can be fatal. What are the least obvious but most important traits to have in co-founders, making sure that your founding DNA is correct from the start?
I think there's two principles in identifying co-founders and co-founder matchmaking. One is simply a complement. Everybody who is a successful founder, as we discussed, has a superpower, has a superpower or more, but typically they also have things that they're either weak at or don't like to do. So having a complement, and the best possible complement may vary by your business objectives, the market you're in. Sometimes it's a technical complement, sometimes it's a sales complement, sometimes it's a marketing complement, a product complement. To identify your strengths and what are the gaps in your DNA. Then secondly, I think it's important to have alignment around first principles. So you don't want to be debating 'are we in office versus remote?' That's not constructive. You cannot have that debate every month. But the application of principles with a founder who's insightful, that is very helpful. So someone who can help clarify your thinking, makes your thinking better. You can actually usually tell even in a coffee whether someone could help you clarify your own thinking. When I meet somebody new that I'm potentially recruiting for some role, in that coffee conversation, they usually ask very sharp questions that make me pause and think about the problem set and the challenge I've been working on in a different light. And if they can't do that in the first hour, usually they can't do it after the first six months.
Peter Thiel taught you that as a startup, you need to recruit undiscovered talent because if talent is easily legible, they usually get hired away by someone more well-capitalized. I think it's Eric Lyman at Ramp who calls this the Moneyball strategy. What are specific reasons someone can be mispriced? What shapes or archetypes of individual should you be keeping an eye out for?
Yeah, so fundamentally, when you start a startup, you're like one to two people sitting in a proverbial garage, and you have to compete for talent with OpenAI or Google or Meta or Apple or Nvidia. And so you have to have some advantage, some comparative advantage. The mispriced talent is almost surely the most appropriate advantage you can sort of create and forge. It's difficult, but at a conceptual level, you're looking for talent that these large companies just don't know how to process accurately. So lack of data points. I like the 'legible' concept. If someone's very young and early in their career, they may not have enough data points to read. So you find a 19-year-old. Does Google know how to assess a 19-year-old? Are there enough data points? Maybe not. That's great. Second thing is, most large companies, not all, but most become sort of homogeneous in their hiring process. And so people who have disruptive characteristics break the mold. Those people can be very, very, very high leverage. So if you identify people who have like a screw loose somewhere, a standard homogeneous hiring process where everybody has an effective veto will tend to eliminate those people too. So you can identify those types of people. You can sometimes find geographic talent. That's become more difficult the last 10 years than 20 or 30 years ago, where there's places that large companies aren't looking. I think that's a lot less common and available now. But 20 years ago, you probably could create geographic arbitrage. But you should be able to isolate at a minimum, as a founder, 'Why do I have some alpha here that other people are going to miss?' It could be also historical experience. Like for example, if you've known someone for 15 years, perhaps you have data points that aren't going to show up on a LinkedIn profile, that aren't going to be processed in a standard hiring process. So that can be a competitive advantage too. It's hard to scale that. Almost nobody has relationships over X period of time with enough people to scale a company, but selectively that could be very strategic.
Okay, let's talk about how to hire and sort of rock down the list here, starting with attract. I think sort of an underrated aspect of this entire process is that Silicon Valley itself has become a little more consensus, right? Like it's become sort of a common creed, like a Harvard grad was before. It used to be really the misfits.
Yeah, when I started my career, or at least when I graduated from Stanford, only the misfits who couldn't get real jobs started companies. That was literally true when I left Stanford. And then after 'The Social Network,' I think that's how many people demarc the change, it became a mainstream career occupation that was aspirational for quote-unquote normal people. Another way to restate this is last June I interviewed Max Levchin, founder of PayPal, on stage, and this video is available publicly, and he described me as the most normal person that worked at PayPal. So to show you, we had 254 people in Mountain View, and I was the most normal one. That was kind of the type of the set of characters that went into startups. I don't think I'd be the most normal person at most startups today.
Okay, so fast forward to today, right? With more options, noise, more average people, you need to get pretty good at attracting great talent. So what are practical ways to cut through the noise and not just attract people but attract the right kinds of people too?
So I think there's two principles. The canonical advice is you have to have a mission that's compelling. And I think there are some companies that can create such a differentiated mission. I think Anduril is a really good example of this. Dialing this back: 'We're going to make America great, we're going to protect democracy, we're going to use technology to insulate democracies from authoritarian and totalitarian regimes and terrorists.' So that was contrarian and differentiated in 2016, 2017. Less so now, but that was a magnet, an attraction for a certain type of people. Palantir had that a decade earlier or maybe even 15 years earlier. And so I think that those kind of missions can be compelling and create the right signal-to-noise ratio of candidate to candidate. If you don't have that much of a differentiation, then I think you need to use other techniques. And those are more challenging actually in a world where everybody's getting funded in a hot market. Much more difficult challenge for founders than when the market collapses. When the startup market is difficult, it's easier to cut through the noise as a new company. So I think then you have to have a compelling answer to an old Peter Thiel question, which is: 'How are you going to recruit your 20th employee?' You can recruit your first 10 through personal connections, through will, but after that you need a really strong answer why this person who has incredible talent is going to join you versus all the other options. Now again, back to the point about recruiting people that have some alpha, if the people are going to get excluded from a lot of other options, you can close. So that's great. You have to create momentum after the first 10 or 20 anyway. Momentum does attract talent. It doesn't always attract the right talent, and you have to filter. I think cultural values can work. So if you have really just like you have a crisp vision that is differentiated, if you have non-mundane, non-trivial, non-banal cultural values, that can be a signal to the right kind of people. So if you say 'We're going to do X' and that's very different than Y, that can bring in the right people too. So having very strong cultural principles that are differentiated from the rest of the planet does create a nice signal-to-noise ratio.
So moving from attracting to assessing. Looking back at your time at PayPal, you realize you were actually pretty bad at assessing prospective hires, but you were much better at getting to read on friends and existing employees. Turns out you weren't bad at assessing talent, you just were bad at assessing strangers, right? Yes. So when you don't have years to get to know people, which is in most cases if you're hiring, what are some practical ways to get data points on someone quickly, and specifically what should you be looking for?
It's really difficult. You may have to invest the time and energy to get data points. It's like as a VC, the same problem. I have the same problem I confront every single day. So I meet a founder for the first time, and then 20 to 40 minutes I have to decide whether we're going to seriously consider investing. And if I have only those 20 to 40 minutes, that's a pretty limited data set to draw a line about the potential slope of this person and this team over the next decade. So you think about it from math again, geometry. You can draw an infinite number of lines through one data point. If I have any more experience with this person or this team, drawing those lines becomes much more talkable and much more accurate. If I have a lot of data points with this person, I should make that call every time. So for example, let's be very concrete. When my friend Max Levchin, actually Max R., and his co-founders were raising money for Fair in 2015, many people misassessed both the opportunity and the team. Max had worked for me for two and a half years. I played soccer with him for probably four or five. And so I just need to make that call perfectly. I have so many more data points and so much more context that I should never be a VC if I can't make that call every time. Now when I have to make a call competing with like Alfred Lin and Sequoia, where we have the same data points, that's going to be challenging to have more alpha and be more right than him. So for example, he did make the wrong call in the seed. He didn't invest because he listened to the general wisdom about the team in the market. Later he realized he was wrong, and Sequoia is on the cap table and we're very excited about having them as partners and building the company. But it's really difficult when you only have a limited data set. And the same thing, you know, when I was considering leading the seed round for DoorDash, there were 17 different startups trying to do some version of food delivery. And people always ask me this question: 'Well, how'd you know the DoorDash house?' I was like, 'Well, Tony was an intern at Square. I knew Tony.' And I was like, 'Well, if we're going to bet on anybody, if anybody can solve this problem, which is a really hard math problem actually to get deliveries per hour to the threshold where you can be profitable, it's really a math problem, a routing problem on drivers and dashers. And it's like, Tony's brilliant. He's a math major undergrad. He should be able to solve this or it's not solvable.' And so having an unfair advantage of having worked with Tony and knowing his background led me to make that call much more accurately than other people.
Got it. Going on to you mention, this is a nice segue. A chronically overlooked aspect of this process of hiring often is doing proper references. Tony notoriously did 20 reference checks for any one candidate. What do most people get wrong about references and how do you do them well?
So I think there's a big debate here. Tony mentioned on stage when I was interviewing him last year that he does 20 references for any critical hire. I recommend about 10. I think references are challenging though because most people... well, let's talk about founder references and references for hires. I think references for hiring are a no-brainer. At a minimum, you should perform 3 to 10, period. References for founders are tricky because most people calling don't know how to assess potential founder. They're thinking through the lens of an employee, and founders are different. And so for example, applying the Fair metaphor, one of the reasons why my friend Alfred and several other investors passed was they did references on Max and they weren't particularly good. Max was not the best employee, trust me. But if you'd asked the question 'Does he have the potential to be an outstanding world-class founder?' the answer is obviously yes. But if you ask the wrong question, like 'Is this person great at their job?' the answer would have been no. So you need to know what you're asking. So for example, one savvy investor I know did actually call Jack Dorsey and asked about Max. And Jack was accurate and said Max actually can be a great founder. But if you just asked general people who worked at Square, you know, his strengths and weaknesses, you'd get a lot of more mediocre answers about his talent as an employee, because he didn't, for example, listen to direction very well.
Okay, just because you found someone talented and you've assessed them properly doesn't mean they don't have other options and doesn't mean they're fully bought in. How do you close a prospective hire effectively?
I think it's Mike at Trello who talks about anti-selling. Mike uses the anti-sell process at Trello, which is basically to really amplify their cultural values, which includes really Olympian work ethic, in-office culture, etc. And it's not really to close the candidate, it's to more filter the candidate, sort of the discussion we had previously, make sure that they're hiring the right people. To really close, strongly highlighting the differences from the other companies to make sure the individual is opting in to the right... I mean, hiring is a matchmaking process, and so you want to hire the right people for your company. And so that's the anti part. I think for closing, the best way to close is typically I start with asking the candidate what they're optimizing for. Like, what are they trying to get out of their opportunity? And so because of this matchmaking, depending on what their goals are, it may or may not be the right fit. But at a minimum, I can address their goals and explain and articulate how this particular company, this particular role helps advance their own goals. Second, I think the other thing you want to do is talk about the challenges. Here's the muscles you're going to build. Here's the excitement you personally and the company. I think that's what motivates the best people. Challenges and building muscles, intellectual muscles, challenges motivate people. I used to say when I was young, 'Challenges propel me.' And so I think that's true of talented, ambitious people. And so you really want to highlight the challenges, and I think that can be very effective. And then finally, on closing, a little technique that Jack Dorsey taught me at Square. He would always ask candidates that he wanted to hire, last kind of question in an interview context would be: 'Is there anything that would stop you from accepting an offer?' And then because he would attempt to extract what the blockers' concerns were, like what are you concerned about, he could then directly address them. I think sometimes there's concerns and they're sort of suppressed. And so sometimes candidates reject offers and you don't even really understand why. And then later you learn and you're like, 'Oh, that's a stupid reason. That was dumb. I could have answered that in a split second.' So I think Jack being transparent and clear and provoking actually is very effective.
Got it. Next point in the step here is promoting. Brian at Airbnb doesn't believe in general managers. Whoever is the best at the specific thing manages that. That's an old PayPal technique that Peter really adopted, probably created. In his words, 'A manager who doesn't know how to do a job is like a cavalry general who can't ride a horse.' That often means promoting internally. Your framework for understanding when to promote comes down to whether an employee's personal growth rate is outpacing the growth rate of the company. Exactly what are good signs someone is scaling, that their growth is outpacing that of the company, or even vice versa?
