Leading a $1B Revenue Startup AND $1.8B VC Firm – A convo w/ Arjun Sethi (Tribe Capital + Kraken)

Well, you flew up, right?

The bat.

Me too. Yeah, thanks for coming. And nice to meet another fellow operator. That's right. Exactly.

So, I grew up here in the Bay Area. So, I'm a local. My parents were in the startup ecosystem for a bit. I kind of learned from them back and forth on what they were doing. Actually, if I take a step back, there were tons of people in my family here in Silicon Valley who were part of a bunch of startups. They either started it, failed a few, but everyone was always trying to get into something. When I was going to school, we always used to learn about the Gold Rush. And when I was watching people come here to the Valley to be a part of a company or start a company, they were just coming into some sort of Ponzi scheme of getting into a company and maybe some get-rich-quick scheme. So, I actually was quite averse to working at a startup. I just looked at anything that was traditional. It's really funny. I think my dream job at one point was if I could just get a job at Citibank, like it's all set. Anyways, I ended up working with my first company, if you want to call it that. My friend started building aftermarket parts for cars. These were things to make it sound better. Over time it was more getting into the engine itself. Superchargers, turbochargers. I got really excited about it given my proclivity towards a little bit of design and manufacturing. I was coding a little bit, but not a lot. I think building something from scratch, putting it somewhere, and being able to deliver that experience to yourself and other people was pretty magical. And I don't know if you guys are car guys at all, but just putting the turbocharger in one of these cars and letting it rip is like the greatest feeling on earth. So I got into that very deeply. It was the first time my friends and I got together and started a company around it. It's not like we said, 'Let's start a company.' We were building these products. How do we sell it? How do we do online payments? How do we get a bank to set up something? So it's like, okay, incorporation? So when you're between 14 and 18 years old, you start figuring out what all these rules look like. That was the first time I got into thinking about anything. But then I also essentially dropped out of school. This is high school, around 14 years old. Dropping out is a political way of saying I got kicked out because I just stopped attending.

Um, I think they were quite frustrated. But today they laugh about it. At that time, I think they were like, 'What's going to happen to my son?' And my dad said, 'I didn't come to this country to have my son become a mechanic.' We grew up in a traditional Indian family. In some senses we're traditional, in others not. But my sisters were accelerating through high school the traditional way, doing well in class, top grades. And I was the inverse. Failing every class.

Yeah, well, actually I wasn't. I didn't have enough confidence in the beginning to race them myself. I just thought it wasn't me. Over time I had to because it was the only way to prove that our products were the best. So you had to race them. Autocrossing and drag racing professionally. We started saying, 'Okay, we can't find a driver. We don't have enough money. So we'll have to do it.' I have a thin frame. In order to qualify in some races, we started losing weight. It was crazy, just the things we wanted to do to prove we built good products. But anyway, long story short, that was the beginning. Eventually I enlisted in the military, didn't go through the whole process, and ended up going into consulting with different parts of the DOD and DIA, spent time outside the United States, then came back and studied at University of Maryland, College Park. Math and history were the focus. Finished with history. I was older by then, getting 23, 24, and I started thinking, 'Man, I got to get a job at some point.' So, I was freelancing for websites, working with a bunch of different companies, mostly in PHP. Then I got an opportunity to come back home to California with a company called PlaySpan. I don't know if people will remember it.

I came back for that, and then Google was starting to think about Orkut. I started writing white papers around social networking. I made up a lot of stuff, but it was based on certain concepts of clusters and network effects. Then Facebook asked me to do something for six weeks, so I kind of just started going around, meeting people, working on random projects, and eventually landed into social gaming. The rest is history from there.

Sure. So, I always tell people that Tribe is an interpretation and an artifact of our worldview. It's not that Tribe is everything for us, but it's an extension of how we think about the world.

So, the story of how I grew up — I stopped at social gaming. In the social gaming and mobile gaming space, the way to think about it is this was the first time where you could measure and quantify how people were using our products. Tactically, if you were building a game, you'd look at the onboarding flow. How many steps to get someone to click the tutorial and how do you educate them? How do you get them to buy something and have different actions in the game? Sometimes the faster you got them to buy made them engage over 30, 60, 90 days. Sometimes slower got them to engage longer. Every game is different in how you teach people. I played a lot of video games, still do. So it was about, 'How do you build infrastructure to analyze all of this in real time?' Essentially product analytics, early versions of app tracking at every action. But also, how do you get informed and interpret that data to make a decision? What are the criteria? Today I say investment criteria or decision criteria, but at that time it was, 'I'm going to put people to work, fix something, change something, add new content to elongate retention.' If you think about any product today, that's what we're thinking about. There are exceptions — like LifeLock or insurance, you don't want them to ever touch it, just subscribe and don't come back. But if you have a workflow product or what I call 'life flow' product — units of time throughout the day — you're thinking about every point of interaction for the customer. So we say, 'We got to build that out.' To do that, you have to put out tons of frameworks. A lot of those frameworks come from a similar way to think about accounting. Accounting is frameworks, it's benchmarkable. We were always obsessed with how to get customers to do something, how to benchmark our customers versus other customers, like a Netflix or something. The more you build frameworks on top of data sets, in theory you build points of view. But the only points of view we had were myopic in gaming, messaging, or social networking. Over time, my goal was, 'How do I get as much data as possible around how all these companies interact and what are the workflow decisions people make?' You can't work at every company, but you can get access to their data. That was the mission 15, 20 years ago. I met a VC, I'm forgetting the firm, but I pitched this whole idea. I still have the picture — it was right when the first iPhones came out. He said, 'Everyone's doing this. It's going to be a commodity.' I felt deflated. If everyone's going to do it, I should probably not work on it. I started another company, a mediocre result, but it had the same structure: how do we grow our product, how do we measure this? It's not just about growing 3% week over week to have a 5x at the end of the year. It's about the small nuances to hit that compounding rate. Most people today don't think that way. They think about feelings of the product, aha moments — all valuable, but that's the art on top of the science. If you don't have a scientific approach to get there so you can have trial and error, you get into notions where you make cyclical mistakes and don't know what's working. I was talking with an LP the other day. He said, 'Arjun, the hardest thing about what you guys do is we can never put you in a box. Every time we talk about it, what Tribe does is hard to articulate.' He said, 'You guys are just a quant VC.' The problem with that is you say everything is quantitative, but that's not true. There's enough quantitative to benchmark, but then you have to make decisions. You have to interpret the data to think about what you want to do. Some things are easy to analyze — surgical, boom, decision. Other parts are not: how should you design the product, what feelings do you want to instill? Should I focus on males, females, in the US, Caucasian, Asian, Hispanic, African-American? Should I look at emerging markets? Even within emerging markets, there are different tiers of purchasing power, different interactions, what they care about in fintech versus someone at a different purchasing power. Then language barriers. We use words like localization or translation, but they're too generic. I think about the world in terms of cohorts and segmentation all the time. That's very natural to me, but explaining it to someone from a financial background — it's a foreign concept because they want to put everything in a box like GAAP accounting. That's one of the most valuable frameworks, but it's the baseboard. Everything is on top of it.

One of the lines we use is that we're experts at identifying product market fit. And we are continuing to aspire to augmenting product market fit. There's a nuance. Experts in identifying doesn't mean we can make a good decision after that. We can identify something that has product market fit.

We have frameworks to understand what product market fit means. I'll give examples. The three of us, if we have time, we'll go to Pete's Coffee. Every week, a certain number of people come back every day. Those are your returning users. Some bring two people — new customers. You guys don't live here, so you'll churn. Look at any business, any product, brick and mortar to software — those interactions happen all the time. That's the essence of a business. So what I gave you is a small framework. Expand that out as much as possible around every interaction that could be leading or lagging indicators. If you get access to a raw data dump of the transactional log of everything the company does, you can start building off that framework. More importantly, once you do that for one company, 10, 1,000, then 15,000 worldwide, you build benchmarking. So we said, 'Now that we meet more companies and they want insights to compare to other companies in the ecosystem — not just their marketplace, but any company with similar characteristics as a business.' That's most important. People think, 'I'm the best marketplace for freelancers, let me only look at freelancers.' You should look at every company with marketplace dynamics and stack rank there. Then understand how to build your business: do you need extreme distribution, limited or high engagement? A lot of people say for SaaS, you need to hit a million ARR and set. The problem is how did you grow? How did you churn? How much money did it take to get to that million?

Roughly 2013. Even though Stewart was a well-known name, they had good metrics but it was non-consensus. They couldn't get a round up.

That's right. My partner Ted also helped lead that investment while at Social Capital. So it was me, Ted, and Chamath. Right at that moment, the first growth and data science team, and a lot of these frameworks built off that were at Social Capital. They were doing it manually at the time. A lot of manual processes were also used at Facebook and my companies. Fast forward to Tribe: Tribe is a manifestation of where we want to spend our time and capital. The company that ingests all this data is called Terminus. We work with a host of people in the industry to get these data sets.

Terminus today has tons of customers. It is a tool for something different for those customers. We call it a quality of earnings tool, quality of revenue tool, quality of metrics tool. It can be for a debt provider, sovereign wealth fund, private equity, late-stage growth equity. Hedge funds use it too. That's on the capital allocation side. Then we have companies, public and private, that are customers as well. They use that same framework to make decisions about which products or features are working, how the whole company works — from spend, return on invested capital, fully loaded CAC. It's a full spectrum. I call it audit for underwriting the risk of the asset at any given time, and then deciding what to do. For a seed to seed-plus company, it would be a 50 to 80-page report about where your company sits today, so you can make a decision. The more data you have, the easier it is to assess and underwrite risk faster. For earlier stage, there's less history, but there are patterns around how people use the product. We consider six months of data for any company, even without revenue — just people interacting with software, agents, or transactions — enough to say it has product market fit. We ask for a data dump, maybe financials, depending.

Scientifically, it tells us if the company has product market fit. Now, it tells you if it's growing or not. But you put art on top of science: 'If I put more capital at risk, can it grow profitably? Can it grow?' The job of venture capitalists is to identify opportunities that grow at high velocity. You want a high multiple return on capital. That's the math. That comes down to the art: you have the fundamentals, speed, benchmark. Do you align yourself to the risk outcome? The more data, in theory, less IRR. But everything I'm saying sounds logical and should be standard, yet 99% of the industry doesn't do this. They flip it. Nothing in this conversation is about the team, how they think about product. Those are extremely valuable, but we do that after underwriting. Once we do that, we build a relationship around how you want to build the company. If we align, we're more likely to write a check and double down, even if others don't believe. We have alignment on fundamentals and how you run the company. Not every company or founder wants that. We won't always align, but the goal is to reduce the delta between what we know and what they know, quickly, to help build the business.

Depends on your definition of gems. Everyone looks for the biggest outcome. For Tribe, we try to put the least amount of money into the best asset to have the highest velocity and multiple. And we minimize loss ratio. We're okay with singles and doubles as long as they have product market fit. Instead of zeros, we have singles, doubles, and then home runs. That gives a net 5x return profile, versus needing maximum ownership. The reason is no VC wants to say they're a bad picker. They claim venture capital is hard, a black box, all art. 'I look in their eyes and know they're a winner.' Masa says that. My belief is that's wrong, but you can get lucky. We can't do that. I can't look in someone's eyes. No VC, even the greatest, in private will say, 'I had no idea it would be this big.' So I think no one knows anything. There's herd mentality. Our perspective as operators was: if I'm an operator using data to make informed decisions, why can't I do that from day zero? When we look at companies, we see patterns of extreme growth and product market fit. It doesn't always work out, but we decide what risk to underwrite. We see a company and say, 'This looks exactly like another company in a different space.' If we didn't build these frameworks, we wouldn't have identified Slack, Zoom, Plaid. We saw them but didn't act at our previous firm. At Tribe, we got better at saying no and identifying patterns in Mexico, Argentina, Indonesia, India. Interesting doesn't mean do it — study it. Then decide if we want to build from scratch, invest as minority, concentrate positions, or run the company. That's what we did with Kraken.

Also, most VCs net net negative value because they don't understand operating skill sets. Most VCs couldn't start a company. I have a different perspective: if you're a thoughtful capital allocator and know enough about operating, you have a running head start. That doesn't mean you'll execute better, but you have a head start. I think of Fred Wilson at Etsy. There are examples. Then I think some things are a snapshot in time. I go back from post-World War II to today. All the anecdotes about how to run a company, how people report, scaling are repeated. But think about the tools we have: speed of communication, analysis, number of contacts daily — very different from 5, 10, 20 years ago. Advice I get: 'You need to run your company this way.' I ask why. Invariably: 'It made these companies successful.' But those companies were built ages ago. It's not that they aren't valuable, but could be an outdated playbook. The base foundation of culture and communication is important, but we have 10 tools to do things faster. Pre-COVID, remote was bad. During COVID, remote was best. Now, remote is terrible. Reality: some work for some companies, some don't. It depends on location and tools. I go to emerging markets. No reason for remote work because work is a safe place and they can't afford tools. Now you have baseline understanding. As you move up the stack, remote can work. VCs, bankers, BD people all use tools. So you ask: what is the tool built for, what's the workflow, how to augment? Can a VC run a company? Absolutely. What is a VC? No identity politics. The job is to deploy capital into best opportunities for high velocity returns. If you get warrants because you operate the company, great. If you start the company and get ownership, great. If you can assess talent and bring product from idea to delivery, why not? Our first incubation at Tribe was in crypto. We saw kids with cool ideas.

Yes, exactly. We see a pattern, study it, and if it's viable, we incubate. That's how we built Capital and others.

Structured in the way in which we thought would create some sort of intrinsic value in the ecosystem. So we said, 'We'll build it because of what we're seeing worldwide and in these patterns.'

No, no, this is a these are I'm incubating products. And so we said, 'Okay, well, why don't we find the team? We'll have more skin in the game. We'll have more economics for the amount of capital that we're deploying. And guess what? It's our idea and we're deploying the capital and we're finding the team, we're hiring them, and these first set of folks that we're building in are coming in as co-founders. If there is an emergent lead there, they'll be the lead, and we'll sort of slowly step back if that makes sense. If there's no emergent lead, maybe we're the lead. If there's another team that's a lead then we'll figure it out. That's how we kind of thought about it. Now, the best time to start a lot of these things is when capital is scarce. So, we didn't have a playbook to say we want to be an incubation house. We just said, 'Hey, based on the way in which the market is interacting today and we have a long-term belief on this ecosystem, we think we should create these protocols, write these white papers, and deliver this experience.' We had lots of failures. We've had a ton of success. And that's great. And then what happens along the way is that we are considered leaders in that ecosystem. A lot of founders want to work with us and they want to partner with us and then they ask us to invest in their company. So, it becomes this nice cyclical notion. It's not vicious in any way. It's just that founders will want to trust other founders and if those founders are coming to you and saying like, 'Okay, I'll also deploy capital while I'm helping you.' That just becomes much more of a natural conversation versus come pitch me at the tabletop and I will tell you if I think you're worthy. It's fascinating because the more you roll up your sleeves and do the thing, everything changes so fast. If you've been just allocating capital for decades, you have to work very hard to actually be current and valuable. Or differentiated.

We are probably at number nine or 10. I think four or five really matter today. Okay.

Well, I get excited about all of them. You love all your children the same. The way I think about it is it's a snapshot in time. Right? And so we'll talk Capital. So, we're co-investors there. Yep. The pitch from them, they had already started the company, right? So, I wasn't a co-founder there.

Yeah, and Eddie. Eddie's the third co-founder.