The Data Stack Show - 167: Data-Driven Investing and Company Building with Ben Miller of Fundrise

Episode Date: December 6, 2023

Highlights from this week’s conversation include:Ben’s background in real estate (3:27)Why Fundrise was Started (4:37)Democratizing Investment Opportunities (6:35)Investment Thesis for Venture (11:55)Challenges with Data and Technology (12:34)Importance of Data Model Abstraction (20:03)Data Infrastructure and Investments (23:22)Evolution of Data Engineering (25:12)Closing the Tooling Gap (34:23)The user base segmentation (36:28)The emotional reality of investment decisions (40:50)Data inputs for real estate investment (47:07)The work of data infrastructure (48:28)The limitations of underwriting analysis (49:36)Improving accuracy with data infrastructure (52:43)The Data Stack Show is a weekly podcast powered by RudderStack, the CDP for developers. Each week we’ll talk to data engineers, analysts, and data scientists about their experience around building and maintaining data infrastructure, delivering data and data products, and driving better outcomes across their businesses with data.RudderStack helps businesses make the most out of their customer data while ensuring data privacy and security. To learn more about RudderStack visit rudderstack.com.

Welcome to the Data Stack Show. Each week we explore the world of data by talking to the people shaping its future. You'll learn about new data technology and trends and how data teams and processes are run at top companies. The Data Stack Show is brought to you by Rudderstack, the CDP for developers. You can learn more at rudderstack.com. Costas, today we're talking with Ben Miller, who's the CEO and one of the founders of Fundrise, which is a really interesting company. It democratizes investment opportunities for assets that are typically off-limits or very

difficult to get access to for your average person, right? So let's think about, you know, a big corporate real estate development on one of the spectrum, but fascinatingly, venture capital investments in tech on the other end of the spectrum. And you can do both on Fundrise. And there's more. And I'm really excited to talk with Ben, one, because they're an extremely data-driven company. And he's a CEO, so he's very technical, but he doesn't have a day-to-day role in the data space. But they're extremely data-driven. And one thing that I'm personally extremely curious about is with a portfolio of assets that diverse, you're dealing with pretty different types of data, pretty different types of data modeling,

very different data formats, right? If you think about the real estate space, probably still a lot of PDF documents and other things like that, right? So I'm interested to ask about that. I'm also interested in the combination. I mean, how do you run a platform that allows you to invest in an apartment building and data breaks, which is crazy. So that's what I'm interested in chatting with him about. How about you? Yeah. I mean, we've talked with many people all these years about architectures, technologies, like how we use data, like how we process data, you know, like we tend like to focus more on like the, let's say like the technology side of things and what is like really exciting going on there.

And we take for granted like the connection with how like value is like created out of data. But that's what like I find like find so fascinating about the episode that we have with Ben recorded, that we literally have one of the best examples here of how data directly connects and working with data

directly connects with value creation. And of course, the financial sector in general is a very good example of that. But here we have a person who can take us like through more unconventional uses of data as like the business also it's like unconventional in terms of like how like offers access to investment and yeah that's what i'm

really excited about i really would like to hear more about how we start like from the raw data and we end up like in creating creating really tangible value that we can see as dollar signs in our bank accounts, right? So that's what really excites me about this episode. All right. Well, let's dig in and talk with Ben. Let's do it. Ben, welcome to the DataSec Show. We have a lot to cover and we're going to talk about data in multiple different contexts, but give us your background. Okay. Well, so I'm currently CEO and co-founder of Fundrise, but I got here over a

long period of entrepreneuring. So I started in real estate, private equity, venture capital, worked for a tech startup that went from zero to a hundred people back to zero in 36 months. Wow. I've worked in, yeah, real estate development, doing urban mixed use, like huge skyscraper type buildings in the middle of like cities. And that went through the 2008 financial crisis and then Fundrise, you know, we started it in 2012. We have 275 people at the company. We have 2 million users, I think 20,000 apartments that are owned through the platform.

Wow. And yeah, so basically, I feel like I've aged. I feel like I've lived 50 years. Yeah, it sounds like it. So tell us, why did you start Fundrise? I mean, you obviously had a background in real estate, but Fundrise is not just about real estate. So tell us about Fundrise and why you started it. Well, so anybody who went through the 2008 financial crisis came away very skeptical of mainstream finance like it's kind of lost of any confidence in the financial system yep and so i wanted to kind of build essentially an alternative to it that you know my dream was to basically build an alternative to mainstream finance and that mean meant like creating a way for people to invest in things that are

like real estate or like private tech or credit, things that basically super high net worth or institutional investors do. And they're sort of less correlated or less connected to the stock market and the bond market, which are what normally people invest in. Yep. Yep. Okay. So I just have a question around the mechanics of this. So when you think about private investors or high net worth investors, and you can correct me if I'm wrong here, but generally there's sort of a, and maybe I'm thinking about this more from like a deal flow standpoint, right? Like people fight over deals and you're trying to make a good investment. You have money to throw it at.

And so there's, you know, it seems like that sort of a closed door network where deal flow happens a lot through word of mouth or other methodologies, right? Where the public stock exchange is sort of on the other end of the spectrum, where you can go on E-Trade and you can buy whatever you want. How do you bridge that gap? Because it would seem to me, and just to be a little bit more pointed, and if this is too spicy, let me know, but people who are getting good deal flow tend to try to protect their system, right? And so you democratizing that, how does that work? Yeah, that's a great question. Well, so partly I came from that system so I can go into that system and get a good deal. So that used to be my job. And so that's part of it.

And then there's a story that those people tell about how they have this deal flow and they're proprietary about it. But that's mostly story. The reality is if you come in with a checkbook, it turns out that you can- It turns out that money talks. Yeah. Surprise. Yeah. And we've done that now across multiple asset classes.

We bought billions of dollars in real estate. We've walked in the room and done hundreds of millions of dollars in very sophisticated credit instruments, aspect securities. And we were buying and buying in a, this is 144A transactions, we had to have a hundred million dollars in net worth. And it's like, know talking about rarefied yeah yeah i mean that's yeah most people most people are rare you know yeah bloomberg terminals and stuff so and we've done that and we've also done venture where which people told me it was the most exclusive and yet we you know we own we've invested in dbt labs and databricks and Canva and ServiceTitan and Vanta. And so we, so, so yeah, I mean, it's, it takes skills and it's not like, but it's not as rarefied ultimately as it, as they want you

to believe it is. Man, I love it. That's sort of a great, sort of a great story of like democratization. We're also just sort of the spirit of like if you show up with a checkbook i love that analogy actually it's not an analogy the reality of that because that's sort of what you're you know enabling people to do by proxy we have a lot of things to talk about from a data standpoint but of course our listeners know i can't not pick apart the business model a little bit so So, you know, and look, I'm not, this is not a show about investment expertise and I'm not an expert investor, but I know enough about asset classes to know that real estate and venture are sort of at like really

extreme expense, you know, pretty extreme ends of the spectrum. Right. And so if you're investing in commercial real estate, you're generally looking at sort of decades long, almost an annuity type approach when you're thinking about the asset, like far more durable, much longer cycles. And then on the venture side, you said that you work for a company that went from zero to 100 employees to zero employees in 36 months, right? Now, of course, if you're looking at the Databricks of the world, of course, they've reached a level of stability. But can you reconcile that for me and the listeners just to try to understand? Generally, investors make money by specializing in a particular asset class. And you seem to have achieved success by democratizing access

to multiple different types of asset classes. Yeah. God, that's such a fun question. Well, so there's a great saying that I don't know if you've heard, but basically says, in venture, you make 100 investments, you lose all your money on 99 and you make them all back on one. Sure. That's how the unit, that's the economics of that. Power law. Yeah.

And in real estate, you make a hundred investments, you make money on 99 and you lose it all on one. I hadn't heard the back half of that, but probably because i haven't been involved in a ton of commercial real estate because there's so much leverage there's so much leverage in real estate there's so much debt so so yeah there's a lot of bad them that are really opposites and you know if i hadn't worked for a tech company and they went you know up and down and if i hadn't been spent basically another 10 12 years building company funderize like i you know i don't know if i'd have the background in tech but after probably

15 years of tech and venture i have a pretty good sense of like of at least mid to stay mid to late stage venture yep or you might call it growth equity. Sure. And one of the ways we generally have been investing, at least on the venture side, is that we use the tools. We have 100 software engineers at Fundrise. Wow. 100 software engineers.

Yeah. Yes. It's actually funny a lot of companies we got to them because we were wow customer and we're like this is such a good technology we got like dbt right oh my god we have to listen to this company we have to listen to this company this company is like a one-of- It's really, I mean, I think it's going to end up being the central nexus of the revolution that's happening in data. And we just like hunting them down. Yeah. Oh man, I love that. I love that. Okay. So that is... Okay. So would you say, just to dig in on that a little bit more, do you have an investment thesis for venture as part of Fundrise and let's just zero on technology like a DBT? Or is your investment thesis like on the ground value that you can...

You essentially have a lab as an investor. I'm just trying to put my investor hat on. Like you have a lab of a hundred software engineers who are managing and we'll get to the data part. I promise to our listeners, we'll get to the data part, but I can't imagine that. I mean, first party data from apps, third party data, I'm sure it's insane. I'm sure you have like actually pretty severe, like challenges. And so you essentially have a lab where you can like try a bunch of different technologies. And

so you almost don't necessarily have to have a thesis as much as your team can say, like, if we had to pick a winner, this would be it because we're doing it on the ground? Yeah. It's funny. I have this view and I repeat it constantly internally, which is when trying to analyze something, you want to analyze it from the bottom up and then from the top down. And we need to do both. And you actually need to iterate on both because often when you do one and the other, you find they don't actually match. And you're like, well, the bottom up and the top down shouldn't give you different answers. And as data people, you start to, you know what I mean.

And so we have top down because basically one of the sort of, I think mythologies in investment is that there's all this alpha, but I actually think it's mostly beta, which basically means that the macro, you know, if you're investing in cloud over the last decade you know you probably did pretty well investing yeah in like you know crypto is enabling technology basically didn't work like so so like as good as a company picker as you might have been like this the macro is such a big driver of returns like you just bought

fang for 10 years you're oh, you look so smart. Right? Yeah, yeah. So, and so that's like, I think it's very true in real estate. Like if you just had

bought apartments in, you know, the Sunbelt, like everybody moved there and everybody lived there and inflation drove up tons of rents. So,

so that's like half of my belief. The other half is a bottom up and the bottom up is like with the software, we're using it. We're like, right now we're in the process of picking a, I don't know what you'd call it, but basically a reverse ETL. So we're looking at all the companies and I'm like, wow, sorry not to throw any shade

here, but like, wow, Segment really is in trouble. Everybody talks about Segment being so great. It seems like, oh man, Segment's likely like, doesn't have a future. And so I just believe you have to do both. And that's how we do it. And so I have a strong, I mean, our thesis has been modern data infrastructure.

We've done a little bit of Prop tech to investments because not much, because we know a lot about property and technology, but our top down thesis has just been the modern data infrastructure. And now an AI combined with modern data infrastructure is just, it's a, it's a wave, like as big as the computer was and so then we're not venture and like having to pick three people and a dog right we're like we can that's really hard but if you pick like you know our investments like like five fan we want to invest in five fan that doesn't take there's not like that much imagination in that yeah yeah and Yep. And so, and I, and if I were, you know, I'm going to pitch you, I'm going to say, hey, you're in the data industry.

Like, shouldn't you be able to invest in Fivetran and Retool and dbt and Databricks? Like, yeah. Yep. Yep. Super interesting. It's refreshing to hear. And again, this is sort of like total armchair investor, but the logical conclusion with those two things would be that you would invest in property

technology, right? Management, all that sort of stuff. But it's actually great to hear that you're just sort of following the things that you are finding actual value in and not trying to like wedge something in with these two, like sort of core competencies, which is great. One question on segment, because segment in many ways is like a really great product, but you're evaluating reverse ETL. Like why do you, why do you have that perspective? And I'll direct the question a little bit in terms of, do you think, because they're a pretty large business, actually.

Do you think the innovation has slowed down just because they got acquired and it's much more difficult to innovate from a product standpoint? Oh, okay. Well, so our head of chief product officer and one of our senior engineers has been looking at it. And so I could tell our

story our data story is we basically just probably live the story many companies did but we what we don't what we want is very different like so we looked we're looking at high touch census and rudder stack so it's and and so i'm not sure i can do it justice to basically explain to people who are really technical but the difference of basically like we want to have control of our own data we don't want to shove our data somewhere else yeah right and that's probably the main problem yeah i think what that's and and that's why we you know we're looking at the companies that allow that yep yep sort of more of like a warehouse native like the sort of you have control on your you're sort of building an infrastructure around your snowflake or databricks environment right

exactly and be able to like create audiences and shove more data you know like you use iterable we can shove an iterable we use zendesk for customer service we have tableau and looker and we just we basically need to be able to take the data and we want control the way we want and shove it to where we want and not be constrained by like you know sure some antiquated approach yeah no i mean i think i was actually a guest on a completely different podcast recently and we were sort of talking about how abstraction is the way of the future right because the big challenge is that if you get into a situation where you're beholden to, let's put it this way, anytime that you're storing data in a third-party cloud system, you're beholden to what I believe,

and actually I'm interested in your opinion on this. There's sort of two major problems that I see with that. One is that they have to choose a data model or a schema, right? You can't store the data without having an opinion on that. And in most environments, you don't have control over that unless it's your own data warehouse or data lake, in which case you can define your own data schema that matches your business logic to a T, right? But if it's in another system, they have to decide how to store their data. And it rarely matches. That's a huge pain point. The other big pain point, which is maybe not equally as painful from a fundamental standpoint, but from a practical standpoint, is that you're subject to a gatekeeper right as opposed to having access to the raw data those are two like major things that i'm with you and that like if you're not building

towards that or thinking about that it's going to be hard because companies just are becoming they have way less of an appetite for dealing with that It's just not acceptable anymore. Yeah. And like, so now we're getting into the data, which is like, I've become obsessed with. So the data model, that abstraction is so essential to being able to have the right downstream business conclusions

and business like abilities. And so like, actually i'll start with like even though we built the our web apps and our mobile apps first we had like scale on that side you know before we got to scale on the real estate side it's the real estate side where we really got i got more sophisticated about data. We went about creating a data model that we thought reflected all of how real estate works. So it wasn't constrained to... And this is a true thing for companies too. So you have operational data, you have accounting data, you have market data like so you have operational data you have accounting data you have data

and you have a kind of sometimes you have various financial data and all those things sit in different places yep and and are like pretty different in and of themselves right like oh yeah they're different i mean there's the sort of pipes challenge of getting it and joining it and cleaning and all that stuff but there's also just like then you have to have a, but it goes back to the data model. You have to have a sense of how they all relate to each other in a way that's as abstracted as possible so that you basically don't, I mean, the mistakes we've made over the past 10 years is we baked into the wrong place opinions. Can you just give an example of like,

what is an opinion that you baked in? And specifically, where did you bake it in? And why was that problematic? Like at what point in the pipeline did you bake it in? I'll give you a couple. I'll give you one that's sort of more data and one that's more business. So when we first started, we have like, you know, thousands of apartments and every apartment

has millions, basically millions of transactions. Like, you know, you paid 15 cents for water that day. And so we have all of that data. And the utility. Right. Because everything. Wow. Yeah.

Okay. So you know, you had to replace the unit level. So you have like almost a building level and then a unit level. So we're talking about like, and then you have a tenant level, and then... Yeah, it's so funny. Holy cow, yeah. Yeah, this is like what I was saying to you earlier, like basically everything in the world is data. Yeah.

Like, right? The whole universe is data and just and so if you're in an industry and you care about outcomes basically you want to figure out how you can model that subsect of the universe so that you can start to get good at at predicting what's going to happen and understand what went wrong. And so in a property, right, when you have, there's, you know, they're like a city, you know, a thousand people might live there, all sorts of things happening. And you're trying to basically get good outcomes at the, and that's a combination of people feeling good and people paying the rent and people having everything fixed. And then next door, they may be building a whole foods

or they shut down the street because there's traffic, because they're going to modify the road. All these things happening in the built environment. And that built environment also has things happening virtually. You find an apartment by looking on the internet.

That leasing traffic comes through the internet. There's just all these flows of data around a nexus, which ultimately is a property. And if you can get really good at capturing all that data and modeling that data, you can actually just make better investments and you can build infrastructure that ultimately I think everybody wants to use. And when you say infrastructure, you're talking about physical...

Well, I meant data infrastructure. Okay, data infrastructure. But it's both. I mean, you can get to both. But ultimately, physical infrastructure comes from a data decision. An insight that you get from information. Yeah.

And when you say data infrastructure that people like to use, there are multiple potential audiences that you could mean there. So like, who are you thinking about? Were you thinking about like people who are data professionals? Are you thinking about the tools that the actual tenant uses to streamline their, you know, sort of engagement with you as a landlord? Right. Yeah. So, okay. When I listen to your show, right, you usually have people who are technically building tools.

And what's happening in the data infrastructure industry is that there's like this tooling or technology being built that really, I mean, like they're, to me, they're like revolutionary. I mean, they're just from where it was. Yeah. You talked about DBT, like as a shining example. Sure. But I mean, if you go back, I mean, let me just digress for a bit and I'll come back to what you're asking because it's almost like you have to answer it from the top down. And so 10 years ago, when we built Funrise, we have 100, we currently have 100 engineers,

software engineers. One is a data engineer. No way. Are you serious? Because you have what you said, what, 200 to 300 employees, right? A couple hundred employees, 100 software engineers, and one data engineer. That's shocking. Yeah, because the organization is like data engineering or data.

It's like a separate segment. And mostly we have application engineering, and we basically process billions of dollars and we push information to our web application and our iOS. A whole team that just maintains performance of your real-time performance of your portfolio

and we do a billion data points, processing and stuff like that. Those are data products like yeah you're delivering a data product inside of an app essentially yeah but they're not you know the idea of like a separate you know we have a cloud ops team right yeah an identity team but we didn't but there we didn't have data engineering as its own resourced department. Yeah, interesting.

Just one quick perspective on that, and tell me if I'm off here. It's interesting because you're delivering a bunch of data products, which is almost like a software subsidiary in a way of your apps, right? You sort of have like an end user and you're delivering different features and products within the experience that they have, which is interesting, right? Like centralized data engineering isn't necessary to deliver a data product to an end user, right? Data is just an input that sort of feeds like a user experience. And that feeds a user experience. That's fascinating, actually.

That's the layering there. Yeah. Well, I could be all wet on this one, but I just think the evolution we're going through is happening writ large at other companies. Just like if you go back to 10 years ago, we had a monolithic code base, and we ended up with microservices. And so that's sort of what's happening in the data industries. The first phase is you build out pipes with Fivetran, get DBT in place. And then now we're going to do the reverse CL. So you're kind of working your way through yeah like the natural evolution and so like but what where i sit is that the application layer or the business layer where we're going to use the tools

that are being invented by companies like databricks and dbt to build applications for people who are not data professionals yep right. Now, is that the end consumer? Or when you say not data professionals, define that, because that could be someone who is on the Fundrise app and they want to make an investment and I don't care,

or that person needs the data to make an informed decision, or you're sort of presenting them with the ability to make an informed decision? Right. Well, so there's lots of different ways to do that. We have two. We're definitely one is essentially it's just an internal user who can like make better

marketing campaigns and decisions about products and just have data. And we are the way we did it before we had our database and we just dumped it directly into Tableau in Tableau's memory. Like, we're just like, here you go. And then we had like, by splitting that up and using the logic out of Tableau, like we can like, so there's like things that we didn't, that we're like kind of learning. But then, but I'm like the, I go to real estate and I, when I think about real estate, I think about finance. Real estate, commercial real estate, apartment buildings, and industrial buildings, that is a form of finance mostly. Yeah. And so I gathered the stat before this podcast.

So there are 100,000 data analysts in the country. In the United States. In the United States. That's their job title, data analyst, no matter who they work for. I think it said I think the sentence is 93,000 so okay and there are 6.5 million

people in the financial services industry working what so and I think that those two professions converge yep I think that those two professions converge. Yep.

And in the same, like basically in the same kind of pattern, which is that for us, I think for most people to get the data, we had to originally go to engineers. Like, can you run this data custom report? And then we could have got to a place where now like a data analyst can write SQL or, you know, and get the data for us. Yep. And then I think we'll,

we want to get to a place where then like a person like me or a person in the marketing department can get the data. Yeah. Without knowing SQL. And then, you know, what's going to happen is that like, so then that's that person who's like, me is like an Excel user, like a self-spreadsheet, basically how they think about data. And then what's going to happen is that it's going to be done by AI with natural language. Yeah, I agree with that.

So that progression is happening. And that progression is the progression of all technologies, which is that it becomes mass user. So it gets really cheap, and it gets really easy. And so the data infrastructure industry is sort of in the mid-phase, is now graduating from it being deeply technical to being sort of a business person starting to get at it. Yep. And that's what I want to do for real estate and finance. That's where we're going for finance.

So a financial professional should basically know in five years, they shouldn't be using Excel, which they rarely. It's so clunky. Yeah. Okay. So a couple of thoughts there. This is a conviction that I've held for quite some time. So I think the statistic that you mentioned is, I think it's true, but I think it's actually sort of a gross mislabeling problem. And let me tell you what I mean by that. My guess would be just based on hearing your background, that you have built things in Excel that on paper are essentially software, right? I mean, you probably like essentially built

software in Excel, right? Whether that's, you know, the ability to sort of model something or analyze data you're using, you know, sort of like macros and, you know, and some of the best software developers that I've worked with actually came out of an analyst background, quant background, and are just Excel masters, right? And that gives them a fundamental understanding of the relationship of data and how to express things through logic, which is really interesting. The gap has been a tooling problem, which I think you highlighted really well, right? Like you have an analyst who's really good at SQL and they can wrangle Tableau, right? And so they're like labeled an analyst. And then you have someone who is trying to build multiple scenarios across multiple

connected Excel files to try to predict, you know, to predict margin on a large commercial real estate investment that has thousands of inputs into the model. It's like, well, where's the skill set difference? There's certainly flavor differences, but you're just talking about different tools. You're talking about people who are sort of producing similar types of work. They're just using very different tools. And I agree that it's so exciting that's starting to converge where it's like, well, those should not be like separate

and really Excel like, you know, should sort of fade into the background. So I agree like That statistic is fascinating, but I also think that's a tooling gap and it's starting to close. I think it's probably maybe earlier than you. You sort of said we're in the middle phase. I think we're still in pre-middle phase,

pretty early, but maybe that's just my perspective. Well, I guess my goal is to try to move us to the next phase, to build a product or products that start to leverage the more technical products like DBT or Databricks, both of which we invested in. Oh, wow. Okay. So, Fundrise, you gave people the ability to invest in both of which we invested in, in ART. Oh, wow. Okay, so Fundrise, you gave people the ability

to invest in both of those companies. Yes. Both Databricks, dbt, we've invested in ART through one of the ART technology strategy. And both of those companies are enabling technologies that if you can basically get the right data into their tooling

and then get that into the right UI so that's basically, I mean there's a lot more than that happening, right? But that's where you can start to basically democratize access to the innovations. Yeah. Okay. I have a funny question about, this is a quick sidebar, but then I want to ask about third party data because the real estate aspect of that is really fascinating to me. I'm just thinking about a Fundrise user, and I'm thinking about investing in an apartment building, and I'm thinking about investing in

Databricks, right? And I'm thinking about the number of people I know who have an appetite to invest in both of those things, because traditionally, when you think about the way that people want to allocate capital, like an individual who wants to allocate capital, you're going to play to your strengths, right? And so you have people who are like, they invest primarily in real estate, right? And then the other end is venture or whatever you want to call it, right? Angel investment. You have a large appetite for risk and you have, you know, your sort of portfolio like

affords for you to sort of make a wide investment in a number of things that are like asymmetrical. And so you have a larger volume of investments, but, you know, and you're willing to lose because one of them might be asymmetrical in terms of returns. It seems like Fundrise is a place where you're presenting an opportunity for someone to sort of do both things, but is that the actual user? How is your user base segmented? I'm just thinking about myself, actually. It's like, well, I actually would kind of be interested in making both of those investments for different reasons as part of my portfolio. But maybe your average investor isn't thinking about investing in commercial real estate and Databricks, right? And your traditional commercial real estate investor, not to stereotype, but may not know why Databricks' technology is you know creating so much value for their company. Yeah

that's why we need high touch for one of these companies so we can create better audiences. Yeah. We got to create the audience in Databricks you know whatever anyways. Yeah. And because we want to build more personalized

experiences. Yeah. But in terms of just your gut sense of your users, it's such an interesting dynamic of extremes of the spectrum on this single platform. Yeah. I mean, we have a lot of investors. And so what happens is there's a lot of use cases. I don't like personas because there's just too many.

When you have a million and a half million investors and people invest, this is one of the things. They invest. What they're doing in 2021, the middle of the pandemic, when print money is being printed, and what they're doing right now is so different. Oh, yeah. Even the same persona is a different persona. Yes. Or they're just,

so their behavior is much less consistent than they believe or any kind of like, you know, archetype modeling would believe. Interesting. And I think that like, if you put the, I mean, right now, archetype modeling would believe. And I think that like if you put the, I mean right now everybody's talking about AI and a year ago no one was talking about AI. Yeah. And so like, you know, we've invested

in some pretty good AI companies too, which we haven't announced yet. And if I were to start talking about like like, you know, RAG start talking about RAG and why this company is so central and OpenAI uses them and people a year ago would have just been totally not interested. And not persona driven, right?

I mean, this goes a bit of the top down, like the macro, high interest rates, oh, I'm making 5% of my savings. Maybe I'm not going to invest in anything. There's so many things affecting that investor that it's not as driven by the persona or psychographic demographic stuff. The problem with the analyst is they really want to believe that it's like analyzable. Yeah.

Yeah. And they seek signal desperately, which is like Peter P hacking. That's just their, that's their flaw. Yeah. And, and so like,

that's like usually my experience with seeing data is that it's either dead obvious in the data or there's no signal and you have to look for it. You're basically p-hacking. Yeah. Unless you have a pretty stable data set. Yeah. But what in this world these days is stable? That's a great... Okay. This is maybe a little bit more of a personal question for you, but I can't resist. And if you don't want to answer, that's totally fine. But you have millions of investors on the platform. A lot of people believe that investment is an inherently emotional decision and we convince ourselves that it's a quantitative decision. What's your take on the individual investor? And the reason I ask that is because I'm trying, and really that's a mirror question

for me, because I think about investing when there were interest rates were zero, pandemic, and am I a different person now? I am a different person. A lot of people would say, well, you shouldn't have an emotional response to this, right? Like you have a plan, you stick to it. But I made decisions based on, you know, my perception of things. And a lot of that is emotional. Yeah. I mean, my conclusion is that there's the emotional reality first, and the facts are gathered to fit the emotional reality and that's how people are and doesn't and it doesn't matter if you're an

individual or you're a institutional investor like the facts are second and the emotions are first yep and that's like you know why the passive investing movement has been, I think, really constructive. And a little bit of our investment philosophy is like, we're not trying to be smarter than Sequoia or Blackstone. We're trying to basically just create access and index what hopefully are obvious macro drivers of good companies. But it is frustrating because in 2021, people were shoving money at us. And I'm like, no, just got to just sit and hang out. So hold on, hold your horses.

And now it's like people are much more reticent, much more concerned. Everything's going down. Everything's going to keep going down, I think, generally, broadly at least. And that's a great time to be investing. Prices are much lower than they were and everybody's much more reluctant. And so it's hard. I mean, investing, stock markets are a mass psychosis business. I mean, it's just massive. People look at the markets and think that reflects quantitative analysis,

but it mostly reflects a psychology sentiment. Yeah, yeah. And you can't really get outside your zeitgeist. And so it's one of the hardest parts of my business. Now, I would think though, and I don't want to read too much into it, but one thing that's interesting to me about having a platform like yours is that if you think about an individual making investments, it's not like they're getting data back from the systems that they use.

Again, like let's just think about someone buying a stock on E-Trade or even someone who makes an angel investment in a technology company, right? They're like, as humans, we will automatically bias ourselves to confirm that the decisions that we made were good, especially when it's our personal money, right?

It's very difficult to break out of that. What's really interesting to me about the dynamic that you provide is that you can provide data back that doesn't care about how you feel, right? And so you, in some ways, can potentially stem a little bit of the confirmation bias that we all struggle with by providing sort of an objective perspective with data is that true i think that the way we think about it is slightly different but that we try to provide more content, like more, like, like if you are,

if you were an investor and you're looking at our app, it would, it's a newsfeed. It looks, and you're getting like investment strategy memos about the investments. Like, Oh, we invested in this company.

Why is this company exciting? From Fundrise. Like you. from us. From us. And I, there's a reason why people don't invest in things is they often don't have confidence. And they build confidence by building knowledge.

But the world, I mean, the world today, the last four years, it's just been an absolute roller coaster. Yeah. And, you know, clearly if you look back, like it'll be fine you make your way through it, it goes up, it goes down mostly if you do

the right stuff it does pretty well but that volatility people hate that and that's just a challenge and it goes back to the volatility is just an emotional the emotional reaction to the volatility things are going to go down.

I think maybe it's if you go through enough cycles you get less affected by that up and down. Yeah, I think my big takeaway there is data itself is not education.

I think that's really wise. Okay, I want to get to some specific data questions here. This has been such a fun conversation and I have a million more questions, but can we talk about the data inputs for real estate investment? Because, you know, deal with as a company, what you expose to your users, and I'm sure I'm going to be way wide of the mark here, so keep me honest, but let's think about simple, reasonably stable data, which would be maximum number of tenants, capacity, average rent, average ongoing maintenance costs, things that you have a...

You probably have a lot of historical data that serves as a proxy where you can probably get that within a pretty accurate margin in terms of predicting what the cost basis is going to be from that standpoint. So you have that side of it. But the value of real estate is on some level highly subjective, is influenced significantly by market conditions, changes in the environment, you mentioned construction, the ability to incorporate new technology into some sort of large building, which is a capital investment. There are all these variables that feel much more subjective as inputs into an investment model. What does that look like for you? What does that look like for your end user? And then of course you have, I would assume, an immense amount of third-party data that you're sort of putting in this model is just publicly available, right? I mean, prices,

tax information, all those sorts of things. Yeah. I mean, this is the opportunity and we're in the process of attacking it. But the way it's done today is by hand, manual. Really? You go gather data. Mostly you buy the data. There's a handful of companies that are the primary suppliers. You go get that data and then you manually input it into a spreadsheet. That is brutal.

And there's like, I mean, we have a hundred real estate people at the company and that's just like they spend their days taking stuff out of PDFs and out of spreadsheets and putting them into other spreadsheets and trying to figure

out what... It's a layer and a bunch of assumptions as you said. and putting them into other spreadsheets and i'm like trying to figure out kind of what it's a layer in a bunch of assumptions as you said and then figure out what you should do and i this is like kind of the you said this earlier this is like the big insight mostly the work is getting the information and then getting that information sort of organized and cleaned and it's not actually once you have you have it joined and cleaned and aggregated and everything else, usually the conclusion is pretty straightforward. But the work, the hours are actually going to that front end part. What if that was done with data infrastructure rather than by a bunch of people doing it by hand?

But the entire, whether it's real estate or venture or private equity or whatever, go down the list of all the financial industry, all lending, the lending, if you're, you know, I work with all these banks, like banks all do it by hand. And so that's crazy, but that's how, that's basically where the industry is. And then the part that like, like the part that I wanted to do this because we've done thousands of underwrites, thousands, and I'm like, if you took all those underwrites and back-tested them, you would see that they're no more predictive

than random. Yeah, no more predictive than a venture capitalist call on who's going to win. Yeah. Because the reality is nobody predicted the pandemic in 2019. No one predicted inflation in 2020. No one predicted the Ukraine invasion in 2021. No one predicted... And so no one predicted that climate change would basically wreck Florida. I mean, kind of predicted Florida, but you wouldn't actually think that insurance costs. But it happened in Texas with inflation. Our rents went up 20% in a month.

Typically, that would happen over a decade yeah and so that was unpredictable and so there's just all these things that have been buffeting the models and i think and so i just and so the industry is obsessed with this analysis and it's because i believe that it's really a narrative it's a sales tool yeah it's actually not really data driven it's really a narrative. It's a sales tool. Yeah. It's actually not really data driven. It's not really data driven. It's like,

what do I need to, what do I have to put on this piece of paper that gets us to yes. And that's like how the investment business works. Yep. And I believe you can get to a place it's going to happen because once you have good data infrastructure, right, you can then start attacking it with ML and AI and you can start to get to a place where it's going to happen because once you have good data infrastructure right you can then start attacking it with yeah ml and ai and you can start to get to a place where it's where you know you

it's pretty not objective because ultimately yeah who knows what the next shock is going to be but it's not like you know you can pretend that this person's a genius because there's so much chance in it yeah yeah. Yeah. I think about that as like, there's a mental model I love called like sort of the, like beware of the fat tails, you know, like the, you could be tracking things so closely. Right. And then like a depression hits. Right. And it's really unexpected. Right. Or to, to your point, I mean, there's sort of all these things that, you know, the pandemic, right. That's like a very fat tail thing where it's like, okay, well, in the bell curve, like most things work, but there's an outlier in the fat tail that hasn't really happened or manifested in the way that it

occurs in history. And so it's a novel experience. And that's just part of the fat tail, right? Where it's like, the tail is not actually close to the X axis. It's like pretty far. And so there's a lot of things that could happen in there that are impossible to predict. And so do you sort of view data infrastructure as like getting the tail a little bit closer to the X axis so that we can have more accuracy or like, how do you, right there, does that show up? Right.

I mean, so basically part of the reason that we're so bad or how do you... Right, that's exactly right. So basically, part of the reason that we're so bad at tail risk is psychological because we have a bias to extrapolate the present into the future. And so we have a sort of expectation that the future will be like a bell curve. So that's partly a human dynamic.

And then it's partly a tooling problem. Excel is horrendous at modeling the kind of multivariable kind of outcomes. It's a very linear structure. And so if you get to a place where on the output, you have better tooling and on the input, you have not just people, but AI essentially that's basically helping. You're like, hey, this is where the data is saying, you're saying this, but let's just, a fat tail event happens every four years. How much of a bell curve is this? So I think that this is what's going to happen in the financial industry. It's going to happen. I'm hoping to be part of it. And that's basically,

again, the application of the tooling that you normally cover is being able to get this more sophisticated approach to data into the hands of the business analyst and the business person who today basically is making bad, poor decisions as a result. Yep. Yep. All right. Well, we're at the buzzer, as we like to say. But I'll end with, again, sort of another little bit of a personal question. Okay, so if you had to completely switch careers and you couldn't be involved in real estate and you couldn't be involved in an invest,

you know, sort of the investment world, what would you do? I really would be a teacher. What would you teach? Oh man, history, maybe, you know, like, know like it doesn't almost doesn't matter science i'd love to teach astrophysics or something like like physics would be fun but i i feel like i just i get a lot of joy in figuring things out and then teaching it and so like i guess like my fallback is to go be a teacher somewhere yeah i hope you have like a some gigantic exit and then you know you go teach kids in high school who are like who

are you and you're like well you know i built a huge company and sold it but i'm gonna teach i'm gonna teach you about like how our government works that'd be so fun that would be i i don't teach i just don't want to teach mbas yeah yeah i'll teach anybody but it's not you know like because and bring like uh real i i want to teach entrepreneur because i don't think that's like teachable but i think yeah i agree with that yeah but i think that there's like stories and lessons and stuff from it that I would want to teach to people who are not MBAs. Yeah. Yeah.

And I would argue this is a whole other episode, but I would argue that like physics and history are probably like much more closely related subjects than a lot of people would like take on face value, right? Like, wow. Okay. Wait a second. You have to tell me at some point what you mean.

Yeah, yeah. We'll have you back on and we can talk about it. All right. Well, it's been such a good show. Learned so much. Congrats on the success of the company. Good luck wrangling the data, pulling out of PDFs. And we'll have you back on sometime soon.

Beautiful. Thanks for having me. We hope you enjoyed this episode of the Data Stack Show. Be sure to subscribe on your favorite podcast app to get notified about new episodes every week. We'd also love your feedback. You can email me, ericdodds, at eric at datastackshow.com. That's E-R-I-C at datastackshow.com.

The show is brought to you by Rudderstack, the CDP for developers. Learn how to build a CDP on your data warehouse at rutterstack.com.