From Insights to Investments: Strategies for Success as a Data Driven Investor

All right. Well, thank you everyone for joining us. I hope everyone had a good lunch break. My name's Paul Ross. I'm CMO here at Affinity, and I'm joined by Mar from Pair Ventures, who I'm really honored to spend some time with. The topic that we're going to be talking about here, and really I'm looking to have an open conversation and would love to also have any interaction from folks in the audience, is to talk about the kind of idea, really building on what Ray was talking about earlier around the data-driven investor. Just to introduce Mar, I'm going to do a very high level, and then you can ... as well. So as I said, Mar's a managing partner at Pair VC. She's a founding managing partner at Pair. Just a little background is that after earning a PhD in electrical engineering from Stanford, Mar developed groundbreaking techniques for optimizing the design of analog semiconductors. That means she's done something real versus the rest of us who really haven't done anything real. Since then, she's accrued over 30 years of founder experience, co-founding three startups in mobile and e-commerce, enterprise software, and semiconductor industries, and has registered 14 separate PAs. Wow. Currently a lecturer at Stanford University, teaching Lean Launchpad, one of the premier entrepreneurship classes at Stanford. Prior to that, she was a consulting professor at Stanford, teaching introductory courses to analog circuit design for nearly a decade, again, doing something real. She's been recognized by MIT Technology Review as a top innovator under 35, named as a champion of innovation by Fast Company, awarded the Digital Automation Conference's Marie R. Pistilli Achievement Award, and recognized on the Forbes Midas list for three years, running from 21 to 23. So Mar, we'll talk about Pair in a moment. I'd love to get everyone to hear an introduction to where you guys are focused on and what you do. But Ray talked a lot about the concept of the data-driven VC this morning in his keynote. And it's kind of interesting, I'm obviously an operator rather than something on the VC side. I spent most of my career in data and analytics. And when I look at the VC industry and private capital investing in general, data-driven is not necessarily the biggest thing, I would say, in terms of some of the decisions that get made. We were talking a lot about FOMO earlier. And when I talk to VCs, I kind of get two reactions. One is obviously we're data-driven, so all of our decisions are data-driven. Or you know, you're right, we need to get better at kind of making these decisions. So that's really the conversation I'd love to have today, is just your perspective on using data to drive where you do. But if we can start, just tell us a little bit more about Pair. Yeah, let's get started. So good afternoon, thank you for having me. My pleasure. Pair is a pre-seed and senior investment fund. So we actually, I don't know what other venture folks have been here today or the audience, but if you're a seed investor, you know that you're seeing a lot of founders before you make an investment. And you can compare to the folks that are doing growth, where they may write one check every couple of years, and they can see a company for a long time, and they're only seeing a very limited amount of target companies. We literally have to be very, very efficient at getting to the top of the funnel, right? And getting that very quickly. And if you're not able to do that, then you're not seeing enough volume to make the right decision, right? For example, if we are going, we were just talking about AI, where every feature AI, there's maybe 50 companies. The only way to actually get really good at it is to see the most companies that are doing that and pick the best team, right? So whereas if you're a Series C investor, you already know who's the winner. You don't have to pick those guys out to dinner. You can win them, right? So it's a different mass that goes behind us. So, you know, I also maybe tell a little bit the story of Pear. As Paul was mentioning, my background is as an engineer, and I'm actually really happy that semiconductors are back in fashion. Yeah, pretty much so. Yeah. For a decade, I used to say, I'm in the chip industry, and people were like, oh, cool. Now they ask me about NVIDIA, what's next? Anyway, so I came from that side. I was an engineer, and my background is really on writing this really super large-scale optimization software for circuits, and my partner, Peshman, you know, comes from actually a completely different side. He's totally like a gut feeling kind of person, and he started in investing by accident. He actually came from Iran with nothing. He got a job at a yogurt shop, at a car wash, and finally at a rug store in downtown Palo Alto, and he's a retail guy, literally selling rugs, and he's really smart. So he realized, oh, people buying rugs are either VCs or founders. I should get into that business. I don't want to sell the rugs. I want to buy the rugs. So he told the owner we should do a fund, and he said, what fund? You don't have any money. You know, this guy literally has no money. Let's just take it from my salary. Don't pay me. So that's how he got started. So the two of us are very different, right? I come from literally writing this super complex optimization software, and he's this guy selling rugs. So it's like the, you know, thinking back, both sides of the brain, and I think it really helped us from day one because we make decisions together. For me, it has to be very analytical and very data-driven from day one, and for him, it's really more of that thinking fast mentality, right? So that's how Pair was built, and maybe a small detail in Ashumi's here, actually, we were the first check into Affinity a long time ago. I think it was 2014, but, you know, we were so fortunate to be one of the first customers, I guess, design partners, I don't know what you want to call it, because we were able to really start Pair from day one with that mentality, and, you know, now we have a team inside of Pair that actually run Affinity and so on, but it's, you know, it came a little bit because of my background. We were able to do that from day one. That's cool. I forgot to mention that Pair's actually, and Mar, you're one of my favorite people that's on our cap table. The reason being... I hope you don't say that to everyone. No, I definitely don't say that to everyone. Except maybe to their face, I would say that to everyone. But you're one of my favorite people on our cap table because the day we did our Series C, you, on LinkedIn, posted a picture of Ray and Ashumi in the first months, and they were so young. It was awesome. It was so cute. Their fury eyes, so cute. Yeah, yeah. He's aged, though. Yeah. But that's mostly the love for Pair, really, yeah. So you've mentioned in some of the other talks that you've given and your approach that, and you mentioned there in terms of your own analytical process about using data, and it's a guide for you. There's lots of different approaches to doing that, and there's a lot of different aspects of it. So how's your approach evolved, I guess, in terms of using data to drive your investing, and how have you set that up internally in Pair? Big, long question. I think the reason I say use data to guide your investment decision is the extreme of thinking that data can do everything is that you can just build some AI model, and I'm sure there's a company doing this, to make decisions, right? If you could just build a model where you input enough variables and enough data, and it would just tell you whether you can invest or not invest, that's what you should build. There's a couple of firms, at least one firm that does that, but they never lead. They always follow. So it's unclear how good that model it is. But most firms don't use data to make decisions. It's more to guide, sometimes on the decision process, sometimes on the sourcing process, and sometimes on both the winning and the support process, right? What I mean by guide, I'll explain what I mean by guiding or sourcing. Our life is sourcing. It's like sales. Anybody that does sales here, you have to source, and sourcing, the hardest thing is where do I put my effort, right? You only have so many hours in a day or so many emails or whatever. So that's number one. The second is you have to make a decision, and you can use that data to at least assist you to, hey, if I invest in this type of company, this is what I've heard, what I've lived before, and how I can complement that team with something else. Winning is actually really important, something that people don't talk about, but the best companies in this world where there's so much money have multiple offers. How do you win that company is actually trying to answer the question, what does that company need, and can I provide it in the next two days? Because that's how long I have to actually win that deal, right? And actually, I just had a conversation today in the Uber getting here and trying to get into a deal and telling the founder, what do you need? He's like, it's a DevOps company. I'm like, oh, I need to be top of mind for a lot of developers, and I really want to hire a researcher in AI. And the next thing I did is tell my data person, go find anybody in our network that has these two things, and we're going to connect them, right? And I think that's perhaps not big, big data, but it's really important that we have that at the tip of our fingers, and we can close that company. And then the supporting, which I think is super critical. Every company is really hard, and I think our companies need three things from us. Boom, start one. As soon as you give them money, they want to hire people. So how do you help them recruit is super critical, and actually having a network that you can query and match efficiently to hiring positions is very important, that's number one. Second is customer introductions. Any sort of warm introduction that you can do, especially in the world where every SDR will be a bot, it's actually going to be more important. I really believe that. If you're getting a thousand emails a day, tailored for you. Perfectly tailored, yeah, sure. And the third thing is really fundraising. So I spend a lot of my time helping our companies raise the next round, right? And for anybody in the audience that's raised money, it's really like a game of chess. How do you play this? And having the right list of investors tailored for a very specific company, and in which order when you see each investor, so you get the right feedback, so you don't go see your best A-plus target day one, is really important. And having data stored as to what happened when other companies in my portfolio dealt with that fund, and knowing within even Pair, who should make the introduction, what other founder do we have that's connected to that fund? It's really every little detail, and actually make one of those deals happen. There's a lot of data involved in all of that. One of the things that when we were chatting before the talk was, I feel like a lot of how you're dealing with things looks a lot almost like my business, where I'm focused on the pipeline of people and things. You told me a mind-blowing number about the number of people you're tracking. I think, I honestly, somebody gave me an updated number. Really? But we have over 120,000 people that we track, which is a lot of people, and we've been in business for 10 years, so it's a long time, but just, you know, we run our pre-seed companies through a cohort-based program that inputs applications. We'll get over 10,000 applications twice a year, so just even collecting what we get inbound, and how do you manage that? Absolutely. And then one of the things that was interesting to me was also your, how do you make the decisions about who you think is going to be good for you to invest in? Obviously, you're matching your investment thesis or whatever else to certain people. How are you making decisions around dealing with 120,000 people about who you actually pull into your process? It's so hard. Well, the first thing, though, is you have to track people. I mean, you have to not, you know, wirelessly track them, but, you know, you have to at least know who they are, and not just, you know, who they are, but I think tracking the interactions, every single interaction you've had with that person tells you a story about that person. There's this famous VC blog, but test founders are not dots, they're lines, so you meet them day one, and you meet them day two, and day three, and you're like, well, what happened in between? Right? Did they get better? Or did they do what they said they were going to do? How quickly did they respond to my email? There's so many signals that you can use as a proxy for how great is somebody executing, basically. Right? So anyway, so at least the first thing you have to do is almost a, and that's how we use Affinity. It's kind of the source of truth, where we track everything, and I think, I mean, it's great. We have all that data there. So the second is actually, you know, we have our own score. I think every venture firm has some form of scoring, and the IP of each venture firm is what is that score, and I can tell you that everybody changes their metric every year based on exactly, you know, what they missed, right? But it's really important, because you are trying to figure out, I don't see the scoring necessarily as we can't invest in this founder, because we are so early. We are so early that many times a company is not complete, meaning they may be missing somebody in the founding team, or it's the first time a founder is an operator, and we need to augment them. It's more to, you know, when we run almost like a clustering model, we'd actually do a clustering algorithm on our data founders, and we're like, okay, you belong to this type of company, this type of company, this type of company, and then we can observe them with those eyes and make decisions that way, right, which I think is the most important. If you don't evaluate the same way two technical founders that just graduated versus two technical founders that worked at Google versus one technical founder, one business founder versus a second-time founder, it's not the same. The score that cannot be, it doesn't mean, you know, you have to almost have scoring per cluster in a way, the way I thought about it. You must be investing in some strong data science capabilities within your team. What does that team look like? We have three people, so we have three people on that team. I would say that pair with SeedFund, but even for SeedFund, we're a little, we have a bunch of people both in the platform teams and in the data team, and the data team is really, it's part of the investment team, so it's not part of a platform team. I think that because of our commitment to how critical it is, what, you know, that information tells us every day, and we have two partners actually overviewing it, so there's a lot of investment in that. I mean, remember, our model is, you know, we have a lot of founders, and we have to get back to all of them, we have to figure it out, and I'm sure we make mistakes, but that's our model, you can't do it all the way. How do you think about the other data? Obviously, you have your affinity data to capture a lot of those interactions, and my guess is you're also pulling in a lot of other data types, which you still have to go into the detail of, but how are you thinking about where you're investing in data beyond the kind of interaction stuff that you have? Yeah, there's a lot of data, you know, there's some data that comes, obviously, LinkedIn data and external data, then you can track GitHubs and so on, you can grab a lot of that external data, and then you can also evaluate the data the founder is providing you, as I was saying, and I think that's also really important, right, and how does it track to, I think that is the key, you know, LinkedIn only tells you some things, and it's actually really interesting, because venture, that's why I think building a full model for making a decision is hard, it's, we have the J-curve model, we're not a standard distribution, and you'll see 100 people, only one is worth investing, so it's, and actually, everybody, there's so many variables, and everybody's different, there's not enough data points, even in a decade-old company, a fund, to really, you know, be able to make a decision, but, you know, you can take a look at even like the way they communicate the value of their company, the way they talk about their customer, and I think that's what we're trying to qualify when we, you know, store it in Affinity, which it takes a little more effort. Yeah, that makes total sense. Maybe with AI, people will put a deck, and they'll say they were really good, I don't know. Yeah, as if, you know, Ray was talking this morning about the arrival of the quant in public investing, that didn't solve all of that either, so it's unlikely we're going to solve it on the public side. Well, we don't have any performance data, right? If you're a CBA fund, and all of these guys have this, you know, they'll have a system that distracts how many employees you have on LinkedIn, how many, whatever, likes you have somewhere, or followers, or where you're number one on product hunt, they have all this data. So actually, my companies always get pinged at the same time by all the associate Series A funds, so I'm like, oh, you hit some level, but if you don't have any of that, then, you know, that's why I can't track that investing when there is zero, so you don't even track this, what do I know about this founder, what did they give me, and how did they talk to me, and et cetera. It's such a different risk profile, like, how do you de-risk that? What's your approach, is it literally the interactions that you have with them, is that that picture over time? A lot of praying and like, it's a hard, it's hard, I mean, that's why I think I have job safety, which is really good, because the signals are just so, you know, they're not as straightforward as for a Series A or a Series B, right, or C. I mean, a Series C, you don't even need a signal, give me your financial, I'll let you know, you know, that's all you need, you don't need anything else. But, you know, for us, yeah, it's like I said, it's more qualitative, although I think you can, like I said, the most important is to measure that, keep that qualitative measure in affinity, and tag the founder the right way, so you remember, you know, who's there. What do I do to forget, because I always say, oh, I met a great founder, and actually, you know, I meet a great founder, and maybe I give my vote a score of 10, Tejman meets a great founder, and then I give him a score of 20, and if I just hire a new person, they say great founder, they might get a one, right, because not everybody's opinion is the same. You're building your experience into the model. Exactly. It's a really interesting challenge as well, to actually get that to be operational. Yes, but you need it. If everybody's a great founder, then there's no point in typing great founders as a label. That's fair. It's interesting, you were talking about applying this data-driven approach across your entire lifecycle with a company, a firm. How have you been able to instill that model into the operations of your entire team? Because there's a lot of people, a pair, lots of different roles and different things, supporting people at different stages. How's that been embedded into your organization? Well, I know you're an operator, so I'm sure you have a dashboard and some KPIs to target. Same with us, right? I mean, I think our KPIs, every quarter we have to send out how our fund's doing. But like in any company, almost that's a lagging indicator of how you're doing. So we have dashboards, and I make my companies do this, so it's not just eating our own dog food. Every CEO should have a dashboard, and every SVP, VP should have a dashboard, and they should see it every morning. How am I doing? How is my team doing, right? So I think everybody at Pair has that, and it's a culture where it encourages, I think, transparency and also collaboration in a weird way, because everybody's aware of the problems, or if I missed a number, okay, how do we solve it as a team? Because they impact all of us at the end of the day, right? So for me, some of the things that we track is at the end of every month is, okay, how many of the top Series A's we did not see? That tells you a sense of coverage, for example, right? Or how many did we see and we passed? That tells you, okay, our decision process is broken. You know, I have a lagging indicator that will, you know, results in some sense, tells you how the fund is going to do. We have a platform team and they have a different dashboard. So the talent team is, okay, how many companies are you working with? How long does it take you to fill in a position? How many positions are we filling? And we actually have a recruiting agency inside of Pair. So those metrics are obviously very, very different. Yes, very much so. I'd love to get a sense, like, so what you're describing is having a very kind of analytical metrics-driven approach across your business, but obviously still injecting your expertise. For the audience here, how many of you have similar kind of analytical measurement happening within your firm? That's interesting because this is actually, I have a prejudice coming from the BI and analytics background. This is the VC is one of the, it's amazing how many billions of dollars have gone from VC into analytics companies and how little is actually used. Yeah, we are the least analytical, but I think there's a certain pride in being a great picker randomly. I'm like, okay, that, like, okay, especially I would, I would say, I know, I know if you're a guy, you're just, it's especially great. You know, I'm such a great picker. I'm like, I'll take that. That's fair. I, you know, it's really hard. Yeah. Why do you think that? And even I think the really, really good firm, you know, if you can actually check their LinkedIn, how many engineers do they employ? It's correlated. You know, there's a big engineering investment in the top firms because they know that, you know, we used to be this cottage industry where literally everybody saw everything. I took forever to do a deal. Uh, so it was no need to do anything. It was a small number of, we see a small number of founders, long time to close. Okay. I mean, I can take my time and so on, but now it's like, everybody's a sounder, everybody's an investor and yes, we're in a downturn. Um, but it's going to go back to normal because, or back to whatever the growth, normal growth, because just technology taking over the world, right? Your definition of normal there is good. Yeah. Don't worry. It's coming back. Maybe not today, but, and I think if you don't invest early on on this and it's going to be really hard to keep up in the next decade. Yeah. That's really fair. So I'm, as I think about, like, as you talked about your, you now have a quite very mature day operation when your, your business as a whole across all these you were to go back five, 10 years and think about what would you have done differently to make sure that you got there quicker or better, or like, what would be the thing that would be different about what you would do? Collect more information earlier. Wow. Uh, you can never collect, you know, uh, too much information. That's awesome. Yes. Uh, anyway, you don't know what you're going to do later, but at least you have it. That's, that's, that's a core of analytics is, is, is, is you can't go back and get the day layer that you don't like right now. Uh, I'm so grateful for your time. Awesome. Thank you, Paul. Thank you for inviting me. Thank you so much for coming. Appreciate it. Um,