Invest Like the Best with Patrick O'Shaughnessy - Auren Hoffman - A Deep Dive on Data - [Invest Like the Best, EP.320]

Starting point is 00:00:00 This episode is brought to you by Teegas, the modern research platform for leading investors. I'm a longtime user and advocate of Teegis, a company that I've been so consistently impressed with that last fall, my firm, Positive Sum, invested $20 million to support Teegis's mission to expand its product ecosystem to unify and streamline investor research processes. In addition to the library of 55,000 transcripts, Teagis now combines at-cost, on-demand calls with a full suite of financial workflows, whether it's quantitative analysis, company disclosures, management presentations, earnings calls, TECIS has tools for every step of your investment research. They even have over 4,000 fully drivable financial models. Tegas's maniacal focus

Starting point is 00:00:41 on quality as well as its depth, breadth, and recency of content makes it the one-stop end-to-end research platform for investors. Move faster, gather deep research to build conviction, and surface high-quality alpha-driving insights to find your differentiated edge with Tegas. As a listener, you can take the TIGS platform for a free test drive by visiting tegis.co slash Patrick. Before we transition to the episode, I want to highlight the Founders podcast, which is part of our Colossus network. David Senra, who hosts founders, has devoted his life to learning from history's greatest entrepreneurs, and every week he distills the lessons of a different founder.

Starting point is 00:01:16 If you want an entry point, I highly recommend starting with episode 136 on Estée Lauder or episode 288 on Ralph Lauren. I hosted David on Invest Like the Best last summer, and it's hard not to walk away, insanely energized after listening to any episode with him. You can find a link to founders and those episodes in the show notes of this conversation. You can also search all past transcripts on our website, join colossus.com. Hello and welcome everyone. I'm Patrick O'Shaughnessy and this is Invest Like the Best. This show is an open-ended exploration of markets, ideas, stories, and strategies that will help you better invest both your time and your money. Invest like the best is part of the

Starting point is 00:01:56 Colossus family of podcasts, and you can access all our podcasts, including edited transcripts, show notes, and other resources to keep learning at join colossus.com. Patrick O'Shaughnessy is the CEO and founding partner of Positive Sum and the CEO of O'Shaunacy Asset Management. All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum or O'Shaunice asset management. This podcast is for informational purposes only. and should not be relied upon as a basis for investment decisions. Clients of positive sum or Oshonnessy asset management may maintain positions in the

Starting point is 00:02:35 securities discussed in this podcast. My guest this week is Oren Hoffman. Oren is the CEO of SafeGraph, which creates data on physical locations. He also founded LiveRamp, a public data connectivity business. Orrin knows more about data businesses than almost anyone I know, and that is the topic of today's discussion. We look at the business of data from every angle. and we finish with a fun masterclass on how to host a dinner party.

Starting point is 00:03:02 Please enjoy my conversation with Orrin Hoffman. Orrin, we're going to teach the audience everything there is to know about data businesses today. I think you're the world's preeminent expert on the topic, and we're going to not spare any detail. I think your two-by-two matrix for categorizing data businesses is a phenomenal place to start. Everyone loves a good consultancy two-by-two matrix to visualize out there. Maybe you can explain the axes of that matrix, which will all. allow us to explore all different kinds of data companies. Sure.

Starting point is 00:03:34 So the first axis, I think, is really important to understand is, are you focused on what happened in the past or are you focused on what will happen in the future? So I think of this is truth versus religion. If you're focused on something that's happened in the past, this is a fact. This happened. This is what it is. So if you think of like Experian, did you pay back your loan that either happened or didn't happen?

Starting point is 00:03:57 And it's supposed to be true. And then religion is something about the future. So often you'll take those facts, you know, make some sort of prediction about the future. So FICO would be religion where they're predicting some sort of future. So that's the first most important axis. Where do you want to be on that continuum? And then the second is, do you want to be more of a data business or do you want to be an application business that's maybe using data as part of the application?

Starting point is 00:04:25 And then when you're a data business, you're often serving application. and you're helping many, many different applications, but you're really just about facts. You're really just about ones and zeros that you bring in. And then when you're in application, you're much more of a solution. And so you have to figure out where you are on that continuum and every company is somewhere on that continuum. If you're a small company, let's say below 50 million in revenue, you can really only

Starting point is 00:04:50 be a one place, in my opinion on that. As you get bigger, you know, if you're a multi-billion dollar company, then you may have some many different bets along. that continuum. Maybe we can go quadrant by quadrant and give a company example to really drive home the different kinds of data businesses and maybe starting with religion data. So this is, I think, where FICO would fall. Describe that quadrant and the sorts of things that fall there. And what kind of matters? What makes quality products and businesses in that quadrant? I think really, do you solve a particular problem that goes there? If you think of FICO's case,

Starting point is 00:05:24 they're just having one problem, like, if I'm going to give you a loan, what's the likelihood that you're going to pay back that loan. That's a very simple use case that can be there. Now, FICO, usually religion companies, there is tons of competition and nobody even has any high percentage market share at all in these religion companies because, okay, well, you're a Catholic and I'm a Protestant and you're going to go to hell because you don't believe what I do. So everyone has a different way of interpreting what religion is. In Fico's case, they've converted that religion into a currency. Of course, once you become a currency, then it's much easier to to become the dominant player.

Starting point is 00:06:00 In their case, the FICO is packaged everywhere. I love FICO. I think it's just an incredible company because it's such a good currency. And then these companies are always going to take some of these data companies. Data you can see much more when or take most scenarios where sometimes there is one player. Often there's two core players that are delivering the data because data is very, very hard to manufacture and takes a lot of work to go do it. But then once you do it, everybody can buy it.

Starting point is 00:06:28 And unlike software, which also takes a lot of work, software comes with a UI. And so you may have very, very different use cases. Data is just facts. So once you get the data, if you think of if your stock trade, the tickers and the prices, and that goes back over 100 years. So maybe the tick 100 years ago was AT&T. And it was every day you would get a price. And maybe now it's every tenth of a second you get a price or something like that.

Starting point is 00:06:54 But it's great data. You don't need 100 vendors. giving that exact same data. So maybe you're going to have one or two that it's going to give, and then you'll use that going forward. Before we leave that, quadrant, can you say a bit more about what you mean by currency and how companies make the transition from being one provider amongst many

Starting point is 00:07:13 to really being the cognitive referent that FICO has become? Like, everyone knows what a FICO score is. And I'm sure early on, it wasn't preordained that they were going to be the winner. What does currency mean? And how do companies make the transition into being one? I actually don't know. I think it's very, very hard. Most of these religion companies do not become currencies and they become good businesses, but they rarely ever become great businesses because there's so much competition and everyone will have a different kind of idea of an

Starting point is 00:07:43 algorithm. If you want to have some sort of fraud score or some sort of other types of prediction of whether there'll be a climate disaster or something like that, it's very hard to create a sustainable, very, very huge business on that. In FICO's case, people started to incorporate it into loans. And then if you wanted to sell those loans, well, okay, you could say these are the loans for this kind of tranche or FICO score and this is loans for a different type of tranche or FICO score. And so it was a very, very easy way to package these loans if you wanted to resell them.

Starting point is 00:08:15 So ultimately, that's how these things become currencies. And then once they do, didn't these businesses become phenomenal? FICO is a phenomenal business. Sorry, just I'm going to ask so many different definitions because it's like an interesting new topic. What is a great business to you defined? To me, a great business is you could fire the top 100 people in the company and the business is still amazing.

Starting point is 00:08:37 So the best business would be Visa. I'm sure the people who run Visa are amazing and they're great. And yes, it's probably slightly more amazing that those people are there. But if we fired all hundred of them and replaced them with my dead grandmother, it's still probably works really well. Simple definition. What about things like, as you think about a great business, does margin matter? Does KAC matter? Like, are there other variables that you're always honed in on as indicative of a great business, whether it's a point estimate or a rate of change? The best businesses have decreasing KACs over time. So a lot of really great software businesses,

Starting point is 00:09:11 their KACs actually have been going up every year. To me, those are interesting businesses, but they're not great businesses. The best businesses, even if you think of like a great B to CBC business like Facebook or Google or Apple, their CACs are going down every single year. And then the higher the market share you have, the more likely you are, the CACs are going to go down. So Visa MasterCard are incredible businesses that kind of own the market and their CACs go down every single year. But there are plenty of other types of businesses where they have high market share and then therefore you would imagine they have relatively low CACs.

Starting point is 00:09:46 I think FICO is such a great example of a religion company, something about the future. And it as a data point, as a number, just makes so much sense to me. The next quadrant is one that maybe is the most confounding to me personally, which is the application religion company. Maybe give an example of a company there and how it functions. Well, there are these companies, and sometimes they can be both. So once they get big enough, they can be both. But you have these companies that will help you, like I could say, understand fraudulent behavior.

Starting point is 00:10:13 But they're not just giving you a score. they're actually giving you a process or giving you some sort of other types of things. And some of what Varus does would be both an application and a religion. Can you explain Varysk for those that don't know it and how the company set up and what the product does? So Varysk is an incredible company. It started off as a data co-op. They started off as a nonprofit, similar to Visa, also signing up as a nonprofit, which was basically a bunch of insurance companies got together to share data.

Starting point is 00:10:42 and then they would use that data to help initially do things like prevent fraud and the insurance. So things like if you had multiple claims that are similar to different insurance companies, that would be an easy way to catch that and then you could start to reduce fraud. It's an industry good. And where everyone Varus came out, all the insurance companies benefited from Varus existing. No one insurance company could do it because they didn't have enough market share themselves. They couldn't actually share data with the other insurance companies because then you have this scenario where you could be like price fixing and other types of things.

Starting point is 00:11:13 So you need another entity to go do that. So Varus can to be born to be out there. And Ferris, like FICO, is just an incredible company. It's a company that we all should study. This co-op model is so interesting, like studying the history of Visa or something. It's almost like the competitive moat or advantage to these companies gets baked before their standalone for-profit businesses. And then some of the best trades in the world have been people buying the equity of those

Starting point is 00:11:40 companies as they've emerged as real companies. Is there anywhere else you've seen this? Maybe it's just that the N is so small that it's not worth talking too much about, but say a bit more about the power of that co-op thing. Co-ups exist everywhere. Even in your Gmail, you'll have a little co-op for spam filtering, where if you click on spam, that helps me and makes my job better, so they can exist within a product, or they could be, as you mentioned, like the product itself, where you have co-op. And probably most of the companies you're involved with are in some sort of co-op for salaries where they submit their salaries somewhere and they get back salary benchmarking. So any type of benchmarking thing is usually some sort of co-op. And it could start top down.

Starting point is 00:12:17 So that's where like Visa or Vera started where it's very, very top down. And you have a small number of companies getting together and doing that. But it also could start bottoms up. I would love to see a co-op where, hey, you can off your QuickBooks. And then you could see anytime you want to have a new customer or something, you're doing graphics or something. thing for somebody. And you have a choice between General Motors and Ford as a customer. And turns out that General Motors, their average DSOs are 90 days and Ford's are 30. Okay, well, if they're going to give you the same price, you're probably going to go with Ford. So that could be an example of, hey, if we all off, we could get back the DSO type of thing. And that could be a really

Starting point is 00:12:59 interesting co-ops. So you could have these bottom-up co-ops and different people are building those off of, let's say, Spotify merchants or Stripe or there's lots of other types of things where you can build these bottom-up co-ops as well. Do you think there's anything that investors can do or think about around co-ops that's productive or actionable? Besides investing in them? Yeah. It's possible that some of these market makers look a little bit like co-ops.

Starting point is 00:13:23 So like the Citadel's of the world could look a little bit like a co-op. Yeah, that's a fascinating one. Obviously, some high-profile investors into the Citadel securities business last year. So you're not alone in thinking that, I think. Let's go to the next category. I think this is the one that probably. probably the most people will very intuitively understand, which is the truth application company. This could be Bloomberg. This could be Cap IQ, especially the investors listening, will fully get

Starting point is 00:13:48 this. But what your spin on it? What do you think about and look for in the truth application companies in the data sphere? Sometimes they're just taking data, which is harder to use, and then they're building some sort of UI on top of that data. There's a credit card data that A lot of people use the Yodley credit card data set, but it's very, very messy. It's very, very difficult to use. You have to be very, very technical to use that. You have to have essentially a strong engineering team to use the Yodley data. And then there was this company's second measure, which is now part of Bloomberg,

Starting point is 00:14:18 which came along and basically said, hey, I'll just take that data. I'll take that stream and I'll build this beautiful UI on it. I'll allow people to do it. And then any user, you don't have to be super technical, can use it to make some sort of influences on top of the data. So that's where we start to see on that. And sometimes they're developing their own data. So I'm an investor in G2.

Starting point is 00:14:38 I love that company. And that's kind of an application on top of some of their own data that they're creating. What does G2 do? G2 is a company that's kind of like Yelp for software. So you get to rate your software, how good it is, et cetera. And then they built all these different tools on top of that to allow you as a software buyer to get the right software for you. And then the last category to round it out before we start to talk about how to build these things

Starting point is 00:15:04 and what to look for, the different metrics itself. Talk about the more pure data and truth-oriented company, which I think is where SafeGraph probably sits with the company you're building now, but also just the whole category with maybe a few other examples. The data businesses historically have not been great businesses. Well, maybe very historically they have. So most of the data businesses that are huge were started at least 40 years ago. In fact, in the last 20, years, the only data business that's really, really broke out pure data business that has become huge is Zoom info. So there's been like a thousand SaaS companies that have been unicorns and maybe Zoom info, maybe one or two other data companies that have been. So it's been probably a much

Starting point is 00:15:44 better place if you were an entrepreneur or investor to be in SaaS than to just be in a pure play data business. We think that it's possible that that's starting to change now. You're seeing a lot more buyers that are looking to buy data today and they weren't looking to buy data five years ago. And that trend is starting to continue. And then you're also seeing just the capabilities of those buyers go up now that you have things like Snowflake and Data Bricks and maybe don't have to be like the world's best engineer to take in data. So just the number of people who can actually use data and benefit from data has grown as well. It's kind of interesting that you cite that Zoom Info concept that maybe that's the one breakout company. But on the other hand, you've got this

Starting point is 00:16:25 insane proliferation of just the amount of data that's being generated, the number of data scientists that are trained in a certain way, tools like Snowflake for storing it and processing it. Obviously, AI is the big topic of the day. That's just all data. Why is that the case that there's this explosion probably of demand and tooling, but a dearth of great modern companies that have been built in the pure data space? Well, there's a few reasons. First, maybe we can step back and just define, like, really, if you think about data, there's really four nouns of data. There's data about people, there's data about places. So that's what SafeGraph does.

Starting point is 00:17:00 We have data about places. There's data about organizations or companies, and there's data about products. So almost all data companies are about at least one of those four nouns. Sometimes they're crossed with one another. And then often they're crossed either with time or with price as well. So if you think of that ticker data that we talked about, the noun would be an organization and be crossed with time and with price. So you have to kind of understand what you're doing as a data company.

Starting point is 00:17:24 and then you want to be organized ideally around one of those four nouns. So you can start creating these join keys, so you can start joining these things. So if you think of, okay, data on a ticker, okay, there's this great join key. It's called a ticker. And it becomes really easy to start understanding what these things are. And we all use the ticker and it becomes really easy to join almost all things around price or joined on the dollar. That's a great joint key as well.

Starting point is 00:17:48 Almost everything around time is joined around Unix time, which is another great join key. So you want to have these ways of allowing to join that data because the more this data could be joined to, the more valuable it is. Even if you quoted prices in ether instead of in dollars or something like that, it just become a lot more difficult to join, especially if you start to go back in time. It becomes really, really hard to start joining that data, which is why we almost all of us, when we quote prices, we quote them in dollars. And then selling that data is actually really hard. It's really hard because data is rarely the solution, data is just an ingredient. We like to say, we're selling high quality butter to

Starting point is 00:18:27 pastry chefs. Just because you have that high quality butter doesn't mean you're going to end up with like this amazing croissant. You can give me high quality butter and I'll never going to give you an amazing pastry. So you still need that chef and then that butter is maybe a prerequisite from having a good pastry. You can't have that good pastry without that high quality butter. But you need many, many other ingredients. And then you also need the ingenuity and you may need the presentation and you may need some different tools and the right frying pan and other types of things to create this incredible dessert. Yeah, I like that concept. And what it makes me think of is this notion of wanting to own the end demand in any given business. And what you're basically saying is that's not the

Starting point is 00:19:05 case with data businesses. Like, you don't own the demand for croissants. So say a bit more about how the best data businesses think about the end markets. You want it to be a more fragmented end market where the demand is, you know, there's 500 bakeries versus one or two. oligopoly bakeries or something, how do you think about the nature of the end demand pool and how that relates to the opportunity to build a good data business? Sometimes you're selling to a specific industry. So if you have price ticker data for stocks, then you kind of know who you're selling to. You're selling to hedge funds and you're selling a very, very specific industry who needs

Starting point is 00:19:42 that data, who's going to do analysis on that data. Sometimes you're selling across industries. Those data businesses are going to be harder because you're going to have different go-to-market efforts, but sometimes you need to be selling cross industries because there's not that many data buyers in any given industry that's out there. In most places today, they're still not ready to buy external data. So if you think of like an average retailer, most retailers don't even use their own data very well. They're still very, very early on the curve of using their own data until you get to a certain part of the curve where you're actually pretty competent

Starting point is 00:20:17 about using your own data. It doesn't make sense to really bring in any external data at all. Once you get far along on the curve, let's say you're a Starbucks, a Walmart, a Target, you're much more sophisticated retailer. Okay, at that point, yes, it makes sense to go in. But most people aren't even close to that. And if you look at almost every industry, it kind of looks like that. Even in the hedge fund industry, there's really like a hundred hedge funds tops that are really good at using alternative data today. And if you think of real estate, there's like less than five that really can use data. So even these very, very large industries, like they're still very, very early on on their data maturity. So there's sometimes,

Starting point is 00:20:55 it's like, hey, I want to sell data to the real estate industry. Well, like, it's going to be hard. You may have to build an application. So CoStar is an incredible company. And they built applications because it'd just be too hard to just sell like raw data to a lot of folks. I remember from my quantitative investing days, especially when I was building stuff and doing data evaluation and things like this, the ability to take an independent data set and seamlessly join it to our existing security master file was like a key, key thing when evaluating something. Like, if that was hard, we were just like, forget it. It's just going to not be worth the effort. So talk about what good data is and does. Like, what are some features of a good, useful data set?

Starting point is 00:21:36 Well, the most important thing about data, and people often forget about this, but the very most important thing about data is that it's true. If you're selling facts, if you're selling something that's backward looking, it should be true. And now it doesn't mean there's not going to be errors. If you think of the price ticker data, my guess is it's like 99.99% accurate, even going back 100 years, I'm sure there's some fat finger typos and stuff like that and that data, but it's generally good and you can rely on that data going forward. The more data you have, obviously, the harder it is to be true. The more current it is, the harder it is to be true. But your goal as a data company should be to try to get it to be 100% true, even though it never will be because data changes and stuff like

Starting point is 00:22:14 that. And Safegraph, we have data about store hours. So even if we were true, a month ago, that doesn't mean we are true today. They may have changed their store hours. So these are things that are constantly changing. And when you have billions of billions of facts, again, you'll never be perfect. But your whole goal should be to get to truth. Because in the old days, a lot of people were buying data for marketing. And if you're sending postcards to all the Thai restaurants or something like that, and you happen to send it to a few Italian restaurants, well, no big deal. You just waste it a little bit of money. It's not that big of a deal. And so in And marketing, even if your data was like 50% true, is probably really, really good.

Starting point is 00:22:52 But today, if you're building these models on top of it, if it's in some sort of machine learning model and some sort of data science model, if you start timesing 0.9 against each other just a few times, you're going to a really small number really quickly. So you often have to be at least 90% accurate. And many times you have to be 99 plus percent accurate to solve certain problems. What else matters? So accuracy is obviously one thing. What I was referencing is like, I don't know what to call it, like joinability or something,

Starting point is 00:23:21 like the ease with which you can fold it into other systems or data sets. What are some other key characteristics? Like, if all I did of his job was evaluate the quality of a data set, what are some of the checklist items that you would go through? First of all, just can I understand it quickly? Is the schema public? Is there transparency around the data? When they have bugs, how do I even let them know about the bugs and how do they fix the bugs?

Starting point is 00:23:45 if you're selling data, you're often selling two engineering teams. So one thing, at least that we do, is we try to understand, like, how much time are we taking for these engineers? And you want to take as little of their time as possible. So can I pre-process the data? Can I reformat the data in a way that can import it? Can I push it directly to whatever? They're using Snowflake.

Starting point is 00:24:08 Great. I'll just push it into Snowflake. No sense them having to download the data and then import it in. That's a whole other step that they have to do. So maybe there are certain APIs I could build to make it really easy. So how do you make it so that you're not wasting their time? These engineers are extremely expensive and they don't have a lot of time. They've got a lot of other pressures on their job.

Starting point is 00:24:27 So how do you make it super easy for them to consume and use the data? I've gotten to know Henry from Zoom Info over the last year. And one of the things that really stands out from talking to him about it and studying his business, Zoom Info, it's really hard to cobble together a new super valuable data set over time. We'll talk about why that's great in the end state, but I would love to talk about the early stages of these businesses and the strategies that you've seen entrepreneurs take to collect the data in the first place, which I can imagine is often a very expensive in time and dollars exercise. So talk about that stage because it seems pretty critical. It's extremely expensive.

Starting point is 00:25:08 And depending on how you do it, it also could change the way investors might look at your business. there's just one point where you're going to go and collect the data yourself and you're going to go manufacture the data yourself. And in that case, it kind of goes below the line and maybe your business looks a little bit more like a SaaS company. But there's another case where you go out and you like buy some of the data that you need. You might need a whole bunch of raw ingredients that you need to go do something else. In that case, for whatever reason in the accounting world, that goes above the line.

Starting point is 00:25:42 It's a cog, even though it should be a fixed. cost in many cases. It's a fixed cost. So it makes your margins early on look really, really bad. In some cases, in year one, you might have a negative margin. So over time, those businesses should converge. So when that gets to $100 million in revenue, maybe that's just a rounding error. If you're spending $2 million on data, it doesn't really matter. But if you're only doing $5 million in revenue, that $2 million of data just looks like you have a really bad business. So from the investor perspective, they need to understand a little bit more about how these businesses are going. And then if you are an entrepreneur and you're raising money, you need to be able to explain that and you need to be able to show them that these are more temporary margins.

Starting point is 00:26:26 And then these are going to change over time because these are fixed costs and not variable. I like this term that I learned recently from Miles Scrimshaw benchmark of a sugar rush product. You're seeing this a lot in AI right now where a product can just take off and insane speed. but very often like the half-life of the usage is very quick, too. Lensa is a great example of this, at least so far in the data, where everyone does their little AI avatars and then forgets about Lensa. It seems like data businesses are like the opposite of a sugar rush. Like it's brutally.

Starting point is 00:26:55 It's like broccoli. Yeah. Definitely the opposite of a sugar rush. They take a lot longer to build. So they rarely grow as fast as SaaS companies, but they're usually much more stickier long term. They're slower build. they grow slowly, but they can become really, really, really nice businesses over time.

Starting point is 00:27:16 Probably many of us have certain data businesses that we love in our portfolio, so they can get really, really great, and they stay high margin usually for a very, very long time as well. Say a bit about market share. I think normally market share is not something that is high on the minds of investors or even operators, I guess hires better, but it's not like a key distinguishing characteristic of a business. Talk about the role of market share and what it can do for data companies relative to, say, normal SaaS companies. I think it can do for any company, the higher market share is better. Just in SaaS is just so much more competitive. And it's often more competitive because there's some sort of workflow, some sort of UI. It rarely makes sense for anyone to be a

Starting point is 00:27:59 true winner in the space. And usually even the largest SaaS company has 25%ish market share in its market. In the data business, it can make more sense to have just one or two players because it's just data. So you don't have that UI side. You don't have anything else that goes on there. And again, you can have a data plus some sort of service. If you think of Stripe or Twilio, those are companies that have very, very high market share. And then the higher the market share you have, essentially, the lower the KAC, because they're likely going to go with you anyway. So you don't have to spend as much on sales and marketing. Most of these companies that do spend a lot on sales and marketing, could reduce their costs by 80% and barely that would even affect much of their sales at all.

Starting point is 00:28:43 Those are really, really good businesses to have because you're going to go with them anyway. You talked earlier about the role of trust around and truth, the truth of the underlying data set being so important. Say what you've learned about brand building and the importance of brand in any distinct way for this kind of company versus, again, say, a traditional software or some other kind of company. If you think of a company that may be most of your audience would be familiar with like a Twilio or something like that. So Twilio basically has APIs for telephony. Now, if I'm a software developer, do I know that Twilio is better than any of their competitors? I don't, but I don't even know the names of any of their competitors. So I'm

Starting point is 00:29:23 probably going to try Twilio first. Then maybe if they're really having a problem and they're not servicing my needs or they're just so expensive or something like that, then I'll start hunting around on the internet for other people that might do something. something similar. So brand can be a really, really important thing. It's hard to do that in the beginning. So you can't really have much of a brand when you've got like 10 million ARR or something like that. But over time, these brands become very important. I would say the same thing is true for Stripe. The same thing is true is you start to become much more of a middleware player, like Plaid or something. And then of course, Zoom Info, as we mentioned before, they have a great

Starting point is 00:29:57 brand. Is their data really better than their competitors? I don't know. I've never even tried to find out. I just use Zoom info. Talk about the most common struggles that you've seen entrepreneurs in the data category face over and over again and how you think about those struggles? One thing is it's just really hard to sell. So selling data is really, really difficult, and you need a different salesperson often that are selling typical SaaS out there. Your sales cycles could end up being longer, which is never fun. You sometimes don't even know how your product's being used because you might be giving them data and then it's like, okay, well, so now I need to invest in customer success a little bit. So I can like talk to that

Starting point is 00:30:35 customer a few times, at least once or twice a quarter, say like, hey, what are you doing? Because you might not really have a good sense, even with an API, okay, well, they hit me an API, but like, how's that valuable to them? I see that they're increasing their usage. I need to understand their workflow. I need to get in there. Maybe if I just added this one other field that could double the amount of value that they're getting. So all these things, I think, are really, really hard. And they often require really talking to your customer and really understanding what the customer wants. And that's not for everybody.

Starting point is 00:31:08 Not everybody wants to go into some big enterprise and spend a lot of time with them and listen to them and hear them out. And they're all fun, these enterprises may be at various different points in their own journey of their own data maturity. And they're using different things. So you built this great integration with Snowflake. Great. Not everyone uses Snowflake for everything.

Starting point is 00:31:28 And somebody might be using some secure FTP or something. You still might need to build something. if you want to get this organization as a client to help you there. Is Google a data business? Is that how you think about Google? Google's in everything business. So Google competes with everyone about everything. So if you don't compete with Google today, you really just don't exist.

Starting point is 00:31:48 Google is everywhere, but in the end, they really want everything on their platform, and they want everyone using them. It's interesting to think about the broccoli analogy and how different that is from the kind of Sugar Rush type of company and the long build times that these things, take. I'd love to walk through the SafeGraph story in some detail. And I love the idea of you sitting down, none of how long ago it was now, trying to list out the world's biggest problems. You can talk about that exercise and the problem that you alighted upon, because I think you were looking for the intersection of a big problem that you can actually do something about. But first,

Starting point is 00:32:21 talk to me about what it was like to sit down and try to write the world's biggest problems down on a sheet of paper. Well, I think a lot of second time founders do this. They did the first business and it was great and it made some money, et cetera. And now, okay, I want to actually have an impact on the world. And sometimes they'll have a problem that they're particular passionate about, but it doesn't mean that their particular capabilities line up to solve that problem. So you might be passionate about climate change or nuclear proliferation or too much war that's happening or not enough energy abundance. There could be all these things that you're passionate about, but it doesn't mean that you have the ability to have an impact on there. So you also, I think, need to look yourself in the eyes.

Starting point is 00:33:01 okay, what can I really help on? And one of the problems that we saw happening was just the fact that a lot of data was becoming siloed in a very, very small number of organizations, Google, Apple, Tencent, a few other organizations were really hoovering up this incredible amount of data. And they weren't letting anyone else have access to the underlying data. And if data really is the most important thing, if it really is fueling all this innovation, then you're just going to see scenarios where we're just going to have a lot less. innovation that's happening, and you are going to see the rents for those innovations generally

Starting point is 00:33:37 accrue to those small number of companies that have access to the data. And of course, that's not a world anyone wants, not even the people of those companies want that. So the way against that is to have really much more open access to data so that anyone can get access to that data. And then really, it's about being an innovator and not just the fact that, okay, I had to do some sort of BD deal to get access to this type of data. So talk about the early year or two of Safe Graphs history and like literally what you did.

Starting point is 00:34:06 What were the steps that you took to get going? Once you'll figure, okay, I want to be a data business. Then you want to figure, which noun of the data do I want to be? And you can build a data business on any of the four nouns. They both have their pros and cons. So we chose data about physical places as one of the four nouns that we care about. And, you know, if you're a map nerd and if you care about those types of things, then it's kind of a fun thing to do.

Starting point is 00:34:29 I don't think there's any one noun that's better than the other. You have to choose the noun that you are most passionate about. And if you have some sort of take on it, what you want to do. If you're having nouns about people, okay, well, people are the most valuable noun to have data on, but there's also a lot of privacy things that you have to figure out. So like, how am I going to have data about people while still also protecting their privacy? And I think today it's much easier to do that than when we started SafeGraph, even six years ago, because today you have things like comomorphic encryption,

Starting point is 00:34:58 You have better understanding of differential privacy. There's a lot of ways to create synthetic data on top of data. So originally, I wanted to do something in the medical data world. And that just didn't seem feasible when we started SafeGraph. Today, that might be actually a very, very feasible thing to do. Say a bit more about that privacy thing. I don't know the encryption term that you mentioned. So what can we do now?

Starting point is 00:35:20 And how is this changing? The privacy technology has just progressed so much over the last decade. And if you remember just 20 years ago or even more than 20 years ago, AOL, they released all their search history to academia. And they tried to anonymize the people who are doing the searches, but very, very quickly, people were able to de-anonymize some of the folks doing searches. I think someone even searched for, how do I kill my wife and these other kind of like crazy searches that were out there. And obviously that becomes a real problem if you could de-anonymize someone's medical history or something like. like that. Today, we've really come way, way further on the privacy technology and we can create either synthetic data on top of the data. You can start joining data sets without seeing underlying

Starting point is 00:36:08 data. There's all these really, really cool things, some of which take a very, very high compute. So some of them are more feasible than the others, but they're all happening at the same time. And it's really, really, really exciting. And I think we're going to see a world very, very quickly where anyone can query medical data without ever knowing who the data is about, or we could make all the IRS data available for querying and for understanding without being able to see the underlying data, or even joining medical data and IRS data together without anyone actually being able to see that underlying data. All those things are going to be possible very, very soon, and then we're going to be able to answer just incredible, interesting questions about humanity.

Starting point is 00:36:53 Today, we don't even really know that chocolate cake is bad for you. We just don't know anything, and we just don't have the data. Very soon, we should be able to answer interesting questions about relationships and love and all these other things once we start conjoining this data together. From a company-building perspective, it seems like the coolest setup for a system would be a database that maintains itself. So you think about LinkedIn as a database about people, they created a system that incensed everyone to keep it up to date about themselves.

Starting point is 00:37:21 So obviously, it's mostly true. they could be lying, I guess. But they do the work for LinkedIn. They like shape the data set without having to get paid to do so. How do you think about opportunities to do that elsewhere? Like maybe an organization, it behooves them to for free. Well, LinkedIn is a data co-op, essentially. The reason why I submit my data to LinkedIn is because everyone else is submitting the data and I can get this like really valuable thing that comes through. It just happens to become more of a B2C version, but of course they can mine it for B2B use cases as well. And LinkedIn and Zoom Info, in some ways are in the same business, but in some ways, they don't compete at all.

Starting point is 00:37:57 Because LinkedIn can't really be just like a data selling business that's a little bit too crass. And people might lose their trust in LinkedIn. So they just have to build tools on top of LinkedIn, like advertising, recruiting, selling other types of tools on top of LinkedIn to use the application. Whereas Zoom Invo's like, hey, I'm just giving you the data. Here's a bunch of leads for you. Go crazy. You can contact them on LinkedIn if you want, but you can also contact them via email. you can call them, you can send them a letter, you can contact them on Twitter, however you want

Starting point is 00:38:27 to. They're in some ways the same business. They have almost the same type of data, but they don't exactly compete with one another. How have you thought about that for SafeGraft? If you think forward five years, maybe describe the data set as exists today, but if you think about the best or rosiest version of the future, how that data set has changed and what might have precipitated it? SafeGraft has really boring data about physical places. So if you want to know about Italian restaurants near me. That's the type of data that SafeGraph has. We'll have like the store hours of the restaurant and the shape or the square meters of the restaurant and the categories and other types of things about that place. And we cover lots of different places around the world.

Starting point is 00:39:06 Today we cover a little bit more than 40 million places around the world and that we want to update. We want to make sure we can update those places at least every month and make sure we have accurate data about those physical places. But there are roughly the same number of places as people worldwide. And if you start including things like trees, which I do as places, okay, well, now we have way more places than people worldwide. So 40 million is just not enough. I'd love to cover every single place in the world, including underground places and lots of other interesting places. And one lamppost is potentially more valuable than another lamp post if you want to put a cell tower on it, even if they're 10 meters away from one another. So really understanding just the raw number of places. And that's the

Starting point is 00:39:47 first way to think about data is there's rows and columns. So you want to get a as many rows as possible. And then the columns are the different attributes about a place. And today we cover a little over 100, let's say attributes about a place. That's interesting. That's a start, but there's probably 10,000 relevant attributes. So we maybe cover 1% of the total attributes one could cover.

Starting point is 00:40:08 So can you cover as many rows and as many columns as possible? And then you could become much more a go-to shot where people can go to you to get the data that they need. What is your prototypical customer? Like, what is the most common use case for pinging this database to get this kind of information? So I think for us, we're in somewhat of an unfortunate situation that we have to sell to many, many different types of industries because there's no one industry that dominates.

Starting point is 00:40:33 But the biggest industry for us is local search. So I want to search for cafes near me or something like that or anybody who's putting something on a map. We do a ton of stuff in logistics because we do a lot of things about warehouses and power plants and those types of things. So anyone who's moving things around if you're trucking, delivery, and then, of course, retail, site selection, and then there could be an advertising use case. Out of home advertising is big.

Starting point is 00:40:58 And then even people who are just trying to understand the macro economy. How are the store hours changing, restaurants closing or are they opening? Those types of things become really valuable as well. How do you stop people from stealing data? Seems like it would be a huge consideration that's different than a normal company. It can be. If you're selling to enterprises, it's a little bit easier because, you don't have a ton of customers usually, and they're usually very good about those types of

Starting point is 00:41:23 things, and they're often buying data from many different folks, so they have internal controls to be able to do that. But most data companies will solve their data. So they essentially will have some fake data that they deliver to their customers. So think of it as some sort of like watermark that's in your data. And if they start to see that data out in the world, they'll have a sense of what happened. Because also that customer could have had a breach. Maybe they had a security breach or there could be other types of things that happen. So back in the day, if you think of mapmakers a few hundred years ago, they would put fake cities on their maps.

Starting point is 00:41:56 So if somebody copied their map, they'd go, oh, I know, you copied my map. They can show you that it copied. That city doesn't exist. Of course, sometimes these fiction become truth. And there are actually cities who became cities because they were fictional cities on maps. Does it make sense that the frequency with which the data changes? is a relevant variable? The more frequent it changes, the more valuable the data is.

Starting point is 00:42:21 So if there's some sort of temporal component to the data, and sometimes there's a price-related thing as well, then that can be really interesting as well. So if I just want to tell you the facts around the war of 1812, okay, well, those facts aren't really changing that much anymore 200 years later. So maybe that data isn't as valuable and it becomes really part of the common world. And that does happen, even with companies that are selling data, like their data becomes less and less valuable on a per data element basis. So your customer might still be spending more money with you every month or every quarter

Starting point is 00:42:59 or every year, but they're getting a lot more value from it. So you think of data is much more like the AWS business, where the spend might be going up, but the amount of compute that they're getting, the amount of storage, all these other things are also going up pretty dramatically as well. And this is why some of these data businesses become winner-take most, because the cost per data element keeps dropping over time. So it becomes really, really hard for new people to compete against them. Assuming it's a business where it's not like LinkedIn, where the data gets maintained by the users and you don't have to do it proactively, I have this visual of the Google

Starting point is 00:43:34 cars driving around with the cameras to build their street view or something. How do you think about getting a system in place that can grab and maintain data as it changes. If you think of data, there's often like three steps to any type of data business. There's the first step where you're kind of ingest, where you're getting a whole bunch of data from lots and lots of places. And you're only buying it from one, but usually you have thousands and sometimes even hundreds of thousands of sources that you're getting data from. The second, and we can walk through all that ingest phase, then the second phase is actually

Starting point is 00:44:07 reducing that data to actually making it good. So that's where your machine learning might come in. That's where you're really trying to understand. You have two conflicting things. One says this McDonald's is open at 7 a.m. The other one says it's open at 8 a.m. They can't both be true on this Tuesday. So you have to make some sort of choice.

Starting point is 00:44:25 And it might be neither of them are true. But those two data points help inform you understand what's true. And you may have holes in your data. And you could also use machine. The machine learning is really good to fill in small amounts of holes in data as well. so they can fill in some of those holes. And then the last step in any data business is delivering that to your customer. How do you have that delivery experience being credible, including the documentation

Starting point is 00:44:46 and everything else that needs to happen on that front? The number of sources of data is very important. First of all, it's more valuable to you as a data business if you have data coming in from hundreds of thousands of sources than if you have one. You're much more beholden to that one source in the latter category. So the more sources you have, the better doesn't mean that, they're equally as good. I mean, we have sources that come in, which are over 90% of the data that comes in from these sources is bad. Like, the data is wrong. Maybe it used to be right,

Starting point is 00:45:17 but it could be like very, very old, but there's still like some nugget in there somewhere that's valuable. So how do you be able to sort through the band to get through the good and be able to take that out? Because that one nugget could add a lot of value to some of your customers. How do you think about column selection? So you mentioned Rosen, columns, like every business, you have to start somewhere and focus somewhere early on. You might provide a single data point for just restaurants or something. As the business grows, and it's like the equivalent of a product roadmap, how do you decide and stack rank which columns to attack next? I mean, it's not that different from a product roadmap. So you ask your

Starting point is 00:45:55 customers, what do you want? Do you want more rows? Do you want more columns? Which rows do you want? I care about power plans at Vietnam. Okay, interesting. I'm going to go ask a few other customers before I go invest in that, if that's valuable. And it's like, okay, what columns I care about? Well, I care about the emissions from those power plants. Oh, that's just I don't have that calm. That's interesting. Turns out lots of people care about climate change. Maybe I'll go create that column. That's a very interesting thing that's out there. People are going to have very different types of things, and you want to be able to rank it. And of course, you're ranking, okay, first of all, you're looking around for competition. So are there a lot of other people

Starting point is 00:46:29 providing this? Or is it more of a green field? If other people are providing it, is the data good? Is it valuable? My customers are asking me for this for a reason. Maybe they can't get it or they don't think it's good that's out there. How easy is it for me to do this? Is this going to take me millions of dollars of investment? In many cases, the ass do take that. And then you're like, okay, in this environment that we're in right now, you know,

Starting point is 00:46:51 of a high cost of capital, maybe I don't want to invest millions of dollars for this particular thing. Or maybe this is something I can do for hundreds of thousands of dollars and I can get it to market relatively quickly and I can very, very quickly recoup my costs. So those are the types of things that you're constantly evaluating. Are the leaders of these businesses different in personality type or in skill set than in a normal business? Like, is there anything that you would, if you're only investing in companies led by certain kinds of founders, building data businesses, are the traits that you would look for in the founders that matter more here than elsewhere? Well, data businesses are weird.

Starting point is 00:47:26 And they take someone who is a little bit different to run the data business than maybe, I think, some of these typical businesses. often when you see a slide when someone's raising venture capital, there'll be some sort of thing where they're at the center and there's all these spokes of other people that are surrounding them. That is not a data business. The data business is usually like one of those guys on the side and the customer is at the center. And then there's all these other things, including your business, but often many other data businesses that are feeding into that customer. So you have to be okay with kind of being on the side.

Starting point is 00:47:58 One of the things we like to say internally is like, we're archivists. We didn't write the Constitution. You've got Madison and Hamilton. They're on the pedestal. They're the innovator. Where are the archivists maintaining the Constitution? Nobody knows the names of those people. They're still really important.

Starting point is 00:48:14 You have to tell everybody what you're doing is really important. But there still is the true innovator, and that innovator sits on the pedestal for a reason, and you're here to support that innovator. And that means you have to maybe take a little bit of your ego out of it. you have to have some more humility in what you do because you're really just in a supporting role for eventually this innovation happening. If you were teaching a course on this at Stanford or something and had to assign three businesses where you knew all the students would go study those businesses from inception through to the present to learn the most as possible about the world of data,

Starting point is 00:48:50 which three come to mind for you and maybe a little bit about why for each? the Experian FICO or Equifax FICO type of thing would be really, really interesting for people to understand. Most people have a sense of how these impact their lives, and they're both incredible businesses. They're actually very different businesses. They're symbiotic. They rely on one another, but they're incredible different businesses. And I think just some sort of case study of understanding that. Now, in the U.S., we have three major credit bureaus. We have Experian, Equifax, and TransUnion. In most countries, you have either one, which is like a super monopoly or you've got two, which is some sort of duopoly.

Starting point is 00:49:28 U.S. is actually the rare case where you really get to three. It's very hard to see a scenario. We'd have more than three in any place because it's very, very hard to collect that data. And then, of course, the FICO's of the world, in some countries, you may have 20 versions of that. But because FICO is so dominant in the U.S. and also overseas, it's becoming the more global standard as well. and then you can start to understand relationships there too.

Starting point is 00:49:53 If you think about your four nouns, is there any one of the four that you feel like is the most underserved today? Obviously, the world would be best. We had perfect data about everything all the time. Where are we lagging the most, do you think, today in the world of data? Well, certainly that data truth quadrant, we just have the fewest number of companies being started. I think there's good reasons because the market is not as good in that quadrant.

Starting point is 00:50:17 It's very hard to understand. There's not as much venture capital money there, and there's not a lot of outcomes outside of Zoom info that we can point to that have had massive numbers of success that are out there. The other company that we all might want to study a bit more that you mentioned earlier is Bloomberg. Bloomberg's just an incredible company, and they're weirdly getting even more incredible over time. They're really at that point where you could fire the top 20 people, and they still will be amazing. Why do you think it's gotten more better over time? It's a good question because every few months, something gets across my desk if somebody's starting a new Bloomberg competitor or Bloomberg for this or this other types of things. But in some ways, if you think of them, they're kind of like Netflix or something where they have these very, very high fixed costs.

Starting point is 00:51:04 Every time they want to get a new data set, they might spend millions of dollars for this new data set. And then everyone gets that data as already part of the bundle. So it's already there. So it becomes really, really hard. and for you specifically who uses Bloomberg, you might use very, very, very different data sets. There might be like one or two that are really common amongst everybody, but then you might use very, very different data sets than somebody else who's using the same terminal, even who's on the same floor as you in the same financial institution.

Starting point is 00:51:32 So it's very hard. Everyone has a different use case for Bloomberg and be just like in Netflix, just what you're watching Netflix might be very different than when I watch Netflix, but we're still both using Netflix. As you think about your own career in data, is there an unexpected plot twist that happened that when you look back on it turned out to be a gift of sorts? I think one thing is just transparency is really important. I think it's becoming more important in everything that we do.

Starting point is 00:51:57 So not just in data, but anything around business, just how do you be more transparent with your customers? If you're a coffee company, you have to be today way more transparent about your supply chain than you did before. You have to be more transparent about your labor rules. So I think transparency is becoming really important. And in data, it becomes more important because it's very hard to verify every single fact. And so they want to be able to understand, okay, what is going on, what's happening?

Starting point is 00:52:24 And then how is that changing over time? Are you getting better at what you do or you're not getting better? There's a lot of companies that are out there, especially some of these companies that have been around for 40 plus years, that really have just taken their customers for granted. And they're not getting any better. They're not really focusing on. And so the ways that small companies can compete in any scenario is velocity, is your product velocity.

Starting point is 00:52:47 So how quickly is your product changing over time? And usually that's a good barometer for the fact that it's getting better. If your velocity is 10 times faster than the incumbent, even though you might be very, very small compared to them today, you can just start playing it out where at some point you can really start being competitive with them. If you think about failure modes, I think it's safe to say now with SafeGraph, and LiveRamp before it, that you were very good at building data businesses and been doing it for a long time. If you think about the most common ways that other people trying to do the same thing fail,

Starting point is 00:53:21 what are those common failure modes that you've seen most often? The most common way, I think by far, is to try to be in more than one of the four quadrants at once. It's like, I'm a data business and I'm an application business. Starting an application business is super hard. Starting religion versus truth business, all these things are super, super, super hard. So you just add to your chance of failure by trying to do more at once. Yeah, when you get to 300 million revenue, great, add one of these other lines in there. But until that point in time, you really want to be very, very focused on one thing.

Starting point is 00:53:53 Do you think that the most natural way to do this is to start by originating some unique data set and then work towards becoming an application over time? Have you seen it work the other way? It definitely can work both ways. It's much easier to start an application business today than to start a data business. I'm running a data business. I would not wish this often on some of my worst enemy. It is a tough slog out there.

Starting point is 00:54:15 It's literally you get no ice cream and you're just eating broccoli all the time. You have to really be passionate about data to do a data business. And there's just so many more examples of these application businesses that have been so successful. So it's a lot easier to become application. The problem is once you start an application business, it is hard to start a data business because unless you've really been thinking about it all the way through. like Salesforce has an amazing amount of data that everyone input into Salesforce. So I would love to have some sort of data co-op that comes back.

Starting point is 00:54:45 How do I benchmark versus everybody else or update my contacts? I have contacts about the CIO of General Electric. Well, somebody else says that it could update all that stuff. That would be great. The problem is every single one of their contracts that they have says that they won't do this. So if they wanted to re-change that, they would have to go back to their original agreements, legal agreements with every single one of their customers. And of course, if they went to their salespeople and customer success people and say,

Starting point is 00:55:11 hey, change this thing, well, if they did that, that means they might not be focused on upselling them. And so their revenues will go down and a lot of the tough. So they're just never going to be able to go do that. So you have to think about it intentionally before you start and you at least have some sort of sense. I'm going to bake a data co-op into my business, even though I'm an application right now. Obviously, we've covered a ton of different aspects of data businesses already.

Starting point is 00:55:33 thinking back to that idea of a Berkeley course on data, is there any major chapters of that story that we haven't covered so far, do you think, that are really important to know about? You mentioned earlier is really just understanding these join keys, and the join keys can make data really more valuable. Data is essentially more valuable, the more it can be joined to. So the more you can join it, and some joint keys are proprietary. So like Dun & Bradstreet has these proprietary join keys,

Starting point is 00:55:59 but sometimes joint keys are open, and they're open to anyone to use. Again, like the dollar or Unix time or some other type of thing. In our industry, there's something called Place Key that is really great for joining data about addresses and stuff like that. And then it can be open for any single person to use. If there isn't joint keys in your industry or in your particular thing that you're looking at, then thinking about creating one and thinking about open sourcing it could be really,

Starting point is 00:56:24 really valuable. So one of the biggest problems, if you think of the four nouns, the last noun is data about products. It's really hard to join products. For books, there's like an ISBN number, but for most cases, it's very, very hard. And if you start to think about food product, well, because you're hamburger like my hamburger and yours as two patties, it becomes really, really hard to even think, how do I even do that? So if you can create some sort of, and sometimes there's some taxonomy involved, but some way to create these join keys in places where these joint keys don't exist, then you really, truly could unlock the power of your data.

Starting point is 00:56:58 Interesting vector of innovation to be thinking about how can I just make this more accessible. I love that. If you think of, okay, well, some of these joint keys are really created top down. So if you think of the meter, the meter, which is probably one of the most successful joint keys of all time, that was decreed by Napoleon that we were going to use this measure. Sometimes it's helpful to be the emperor of all of Europe to decree your joint key. But sometimes joint keys also can come bottom up as well. Like standards. in the history of computer science are so important and so key. I love on LinkedIn that you've got CEO and Chief Historian as your title.

Starting point is 00:57:35 Let's switch topics as we start to wind down the conversation and talk about great dinner parties. You created dialogue with Peter Thiel, I don't know how many years ago now, but basically something singularly dedicated to like getting interesting people together and showing them through an interesting format like a fantastic time where they had to meet each other and learn a lot and debate and converse. Give us a little crash course on what you've learned across. many years of, I'm sure, having countless meals with really interesting people, and what makes

Starting point is 00:58:03 for a great meal? So I love dinner parties. I'm passionate about dinner parties. It's really outside of work and family, it's probably the most passionate about. I think there's a small number of things that you can do to just have a way better dinner party. The first one is that they should appeal to introverts and extroverts. Sometimes you have these things that just appeal to one or the other, often just to extroverts. So having things, that appeal much more to introverts are good, having one conversation is always better than having multiple conversations.

Starting point is 00:58:35 So if you're just talking to your neighbor, it's not that interesting, which means that you need some sort of place with good acoustics where you can have one conversation. It's hard to do that in like the middle of a restaurant to have dinner parties. And even some of the private rooms or restaurants are really, really loud.

Starting point is 00:58:51 So you may need to have them at your home. And even if you have a tiny home, it's still fine. You can just get a bunch of people around the table. It could be a little card table. And the one thing that is probably the least important thing in dinner parties, in my opinion, is the food. Nobody remembers the food in the end. They remember a lot of other things.

Starting point is 00:59:07 They remember the quality of the conversation, what happened to the conversation, et cetera. But the food isn't as important. And then, of course, you want to have some sort of good conversation about something. So letting people know ahead of time, we're not just having a dinner party. We're going to talk about these things so people can start to prep their mind. Not everyone's so good about coming up with brilliant things to say on the spot. So at least on the drive over, they can start thinking about, oh, this is the type of thing I'm going to talk about at this dinner.

Starting point is 00:59:35 And it could be a very, very narrow topic. Like we're going to get into this really, really interesting piece of the supply chain of semiconductors. Or it could be a very, very broad topic, like, what's the meaning of life or something? All those things can be incredible. How do you appeal to introverts? Well, the more structure, the better. So one conversation is always better. Having a seating assignments where you know where to sit.

Starting point is 01:00:00 I mean, I remember going back to junior high where I had my tray and I'm just walking around and I wasn't like one of the popular kids. I didn't know where to sit and I just didn't want to end up sitting next to the school nurse. That was just my number one goal. So just assigning someone you're sitting here. It's really easy. We're going to have one conversation. Then it's just going to be so much better. These like cocktail things that are out there, cocktail parties are just the death franchise.

Starting point is 01:00:24 introverts. I personally can't stand them. So having some sort of way where, like, you can allow that introvert to thrive is really appealing. What's the spectrum of the number of people you can have to have one conversation? Well, you need a moderator. The better the moderator for the dinner, the more people you could have. But I think it would be really hard to have more than even with the best moderator, let's say maybe the maximum is 14 and maybe the ideal number is eight. What makes a great moderator? It's very hard to be a good moderator, and people who are most interesting are often not the best moderators. So they have to be able to get everything flowing.

Starting point is 01:01:02 They have to make sure that everyone's contributing. They have to be able to cut people off because if someone's going on, then that's no fun for everybody else. They have to make sure that's interesting. Hopefully they can interject some humor into it, because if everyone's laughing, that's usually a good dinner as well, and they can keep everything going. And then I also think all dinner parties should end early. Keep people wanting from it. So if you start a dinner party at 7, everyone, let's say get seated by 7.30,

Starting point is 01:01:28 get everyone out of there by 9.30. If people want to linger on and talk till the wee hours of the morning, great. But people have other things to do. They have to wake up. They've got their crazy workouts at 5 a.m. or something like that. So give people an excuse to go late. Worst dinner parties like people like,

Starting point is 01:01:45 they slowly kind of, you're still trying to go. But people, oh, sorry, I have to excuse me. myself, I've got an early flight tomorrow. So be respectful of everyone's time and let people get out there when they need to. Any other tips or tactics from the God knows how many of these that you've been a part of that you think people should think about? The other thing is just letting people know ahead of time who's coming. Letting everyone know, okay, here's who's coming. Here's a little bit about each person who's coming to the dinner. Patrick O'Shaughnessy is coming and he's this amazing podcast. You didn't listen to this podcast. You definitely should. And then after people go to the dinner,

Starting point is 01:02:18 the next day, send them everyone's contact information. So many times you went to dinner like, oh, that person across the table, they said so many interesting things. First of all, I kind of even forgot their name. I don't even remember that person's name. Now, at least I have their email so I can reach out to them. Oh, wow, that was so interesting. I'm going to be in New York later.

Starting point is 01:02:36 Let's have a chat about it, et cetera. What's the best or most memorable dinner party you've ever been to? Oh, my gosh. The most memorable dinner parties are the ones that I probably can't say. Can't talk about it. Yeah. I think it's also important that they're off the record. If they're off the record, people have the ability to play with ideas a bit.

Starting point is 01:02:54 And in this world that we live in where everything's on Twitter, it's very unsafe to play with an idea. And you want to be able to say, look, this is a safe space. And we're going to assume positive intent. Everyone can play with an idea here. And we're here to help you make your idea better, but not necessarily judge you personally. I asked Cheryl more of this question. And I really liked it.

Starting point is 01:03:13 So I'll ask you, since you can't tell the actual answer. If you had to organize a dinner party with, let's say, up to five people, including you, and let's just say all the guests have to be dead so you don't make anyone feel bad that's alive that you don't choose. How would you form the dinner party? Who would you invite? One of the most impactful people in the world are religious leaders. So if I could bring in somebody like Jesus and Muhammad and have some sort of conversation with

Starting point is 01:03:37 these folks and really try to get to their core and understand these things and this would be incredible. I think actually very soon, maybe someone's going to build an AI, maybe not with the religious leaders, but with other types of people, you could have dinner with Cleopatra and with Julius Caesar. And I just think that would be so cool and so interesting. And of course, you might want to have dinner with a dead relative and really try to understand them or maybe you never reconciled with your father or something like Field of Dreams, and then you could have dinner with your dad.

Starting point is 01:04:06 So I think all these things are going to be really exciting to do in the future. If you had Buddha and Jesus and Muhammad and all the world's major religious, just figures at a table and they all spoke English. And you got to ask one question that they all had to answer. What would you ask? It's funny because your classic question, the kindest thing, maybe not be good for them. That might be the one you don't want to ask them. One thing might be, because these are people that may have had thousands of years to reflect on things, it's just like how a society surprised you in those 2,000 years. What has changed out there? What are you disappointed with, because obviously there's a lot to be disappointed with, but also what are you

Starting point is 01:04:43 enthused about with our society? What has been a positive surprise that has happened? I'd be really interested in how they think about it. And I can imagine that there are answers would just blow us away. What are your answers to those two questions? What are you most disappointed about and enthused about? I do think that people are generally good. There's a small number of really bad apples out there. But I think people are very good and they want to be in I want to be inspired. They want to be able to do the right thing. They want to find people who can help them do that. And we sometimes live in a world where there's just somewhat lack of inspiration. This ask not what your country. You're like, that doesn't exist as much. Maybe it never

Starting point is 01:05:25 existed, but it doesn't seem to exist as much for my generation and your generation. If we could get just inspired a little bit more. And sometimes when you meet these people and you interview so many people in the interview, these people are so passionate about what they do, and you could hear it coming through on the microphone over the podcast about their passion. That is why people listen to your podcast is because of that reason. Everyone wants their getting inspired by the, even if they're doing something completely different, they're like, I'm investing in a random hedge fund or something like that. It's still cool to hear their passion. I find when you meet someone who's really good at their craft, even if it's just completely random type of thing, like they're a great tour guide

Starting point is 01:06:06 for the Louvre or something. And there's millions of tour guides, but they're just so good, and you just get so excited by seeing someone be great at their craft. Yeah, it's a wonderful closing thought. Find as many of those people as possible in your life and you're going to live a good life. You know my traditional closing question, Oran, what is the kindest thing that anyone's ever done for you? Well, first of all, I love the question. I think the question is amazing. And I was thinking about this, sometimes the best example is actually a counter example here. So when I was a teenager, I kind of fell under the influence of a really bad person. And I got caught up in their orbit. And this person was like super highly unethical, but also alluring, especially for a teenager.

Starting point is 01:06:44 And it took me a couple of years and I finally broke away and realized I didn't want to end up like this person. And to this day, I actually often think about this person as what not to become. So in some way, I think the kindest thing someone has done for me, it's the lesson that the least kindest person I've ever met gave me. Yeah, it's a totally unique and very different answer. I'm curious, what you think it was that allowed you to get caught up with that person? Well, I think it's very easy to get caught up with people. Even if you're old like me, it's easy.

Starting point is 01:07:13 But when you're younger and you're more impressionable, it's very easy to get caught up with charismatic people and to kind of fall into their orbits. I think all of us at one point or another got caught up in something. And sometimes it's good. We get caught up in something that's bigger than ourselves and we do something that's better for us, but you can also get caught up with people that maybe aren't so good as well. There's a little tiny Turkish proverb that I've always loved, which is no matter how far you are down the wrong path, turn back.

Starting point is 01:07:40 What caused you to turn back? How did you come to the realization to change? I don't know. I saw this person and didn't want to be like them. And you could see, okay, they were a few years older than me and I could see how I could easily end up being like this. person, that's the path I'm on. And then it's like, okay, well, are you sure you want to be that? And I could see how there is some allure there. They may have money or they may have stature or

Starting point is 01:08:04 et cetera. So you can have some sort of allure. Maybe it was just lucky or maybe just I tripped one day. But I think there's definitely many scenarios that I replayed my life where I wouldn't have made that decision. Well, you talked about the best ever tour guide at the Louvre or something. I think you've always been to me the best tour guide of data and this entire world. And I've so enjoyed doing this finally with you on the record. I appreciate all the lessons, all the nuance. I think people probably haven't thought about this category in such a deep way before. A lot of this is new to them. And if you really think about it, especially as you think through the four nouns, everyone kind of does this in their own product, in their own business, in some way,

Starting point is 01:08:40 shape, or form. And it's just really useful to have some frameworks around thinking through data. So, Oren, thank you so much for your time. Thank you, Patrick. This has been a real pleasure. If you enjoy this episode, check out joincollasus.com. There you'll find every episode of this podcast complete with transcripts, show notes, and resources to keep learning. You can also sign up for our newsletter, Colossus Weekly, where we condense episodes to the big ideas, quotations, and more, as well as share the best content we find on the internet every week.

Invest Like the Best with Patrick O'Shaughnessy - Auren Hoffman - A Deep Dive on Data - [Invest Like the Best, EP.320]

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.