CB Insights Founder Anand Sanwal - Data Business & AI

Anand Sanwal is the founder of CB Insights, a business analytics platform and global database of intel on technology companies. Anand was CEO of CB Insights for 14 years and recently stepped into the role of executive chairman. 

In this episode, Auren and Anand discuss: 

• Tech and venture capital narratives vs data
• AI integration and workflows
• Future of data businesses
• CEO transitions at 1 to N companies

World of DaaS is brought to you by SafeGraph & Flex Capital. For more episodes, visit worldofdaas.buzzsprout.com, and follow us @WorldOfDaaS. 

You can find Auren Hoffman on X at @auren and Anand Sanwal on X at @asanwal. 

Editing and post-production work for this episode was provided by The Podcast Consultant (https://thepodcastconsultant.com)

Transcript

Auren Hoffman: 0:02

Welcome to World of DaaS, a show for data enthusiasts. I'm your host, Auren Hoffman, CEO of Safegraph and GPFlex Capital. For more conversations, videos and transcripts visit safegraphcom. Slash podcasts Anand Sanwal is the founder of CB Insights, a business analytics platform and global database of intelligence on technology companies. Anand was the CEO of CB Insights for 14 years and recently stepped into the role of executive chairman. Anand, welcome to World of DaaS. Thanks for having me, Auren Big fan. Now. I'm real excited. Now. Tech is a very narrative-driven world. What are some media narratives or widespread ideas about tech or venture capital that aren't backed up by data?

Anand Sanwal: 0:49

You're totally right. It is very narrative-driven and very pundit-driven. I would say A few things that I think have surprised me about the tech world as we've built CB Insights. One is I think it's an incredibly what I'll call mimetic market. One is I think it's an incredibly what I'll call mimetic market. I think I had this preconception of investors being this sort of swashbuckling group of contrarians when I started and what you actually find is there's maybe 5% to 10% who are in that category, and then everybody else just follows.

Auren Hoffman: 1:21

It seems all the deals are consensus. Especially recently, everything's super consensus. It's like you either get like 10 term sheets or zero term sheets.

Anand Sanwal: 1:29

But even within a sector we know that there tends to be a winner-take-all or winner-takes-most in many of these markets. But we'll still have something like 20 global payroll companies get funded and not funded at $1, $2 million levels, many of them in the $50 million plus $100 million plus levels.

Auren Hoffman: 1:48

Because things are so consensus, it almost leads to worse outcomes.

Anand Sanwal: 1:52

Yeah, once somebody who is proven invests in that area, it becomes the signal that hey, they've done the research, we should go and find our version of that thing. So I've been surprised by that. I guess I was surprised. Now I'm no longer surprised. I think the other things that are not talked about a lot are that the time to exit is just getting longer and longer, and it actually now doesn't marry well with fund life cycles, and so I think that's going to be probably already creating, and will create, interesting dynamics when a founding team is 10 years into building their company and the investors are looking to harvest it and there's this dichotomy or divergence of interest. Maybe at times, but the time to exit keeps getting longer and longer.

Auren Hoffman: 2:39

There does seem already a fair big cottage industry of funds buying stakes from other funds, and so it seems like that is already stepping in for some of those things right.

Anand Sanwal: 2:49

The secondary markets have tried that really hasn't taken off, I would say, but there's certainly a lot of attempts to do it. But I think that is one area. And then the other thing, just related to the exits, is that there's this selection bias in the exits that we see discussed and celebrated. Most of the exits in tech if you exit so this is putting aside those that don't die are sub 50 million, sub 100 million. So those really big outcomes are actually a lot more rare than we might imagine. But I think the narrative tends to be one of we celebrate and discuss the big winners and then everybody else who might have made life-changing money for themselves and the early teams don't get discussed a lot, but they tend to be a lot smaller on that spectrum.

Auren Hoffman: 3:34

It seems like the YCC deals these days are $25 million pre or post. If the more common exit is $50 or $100 million, it seems very hard for the investors to make any money there.

Anand Sanwal: 3:46

I think this is where founders and investors have a different profile, in the sense of the investors looking for that power law returns. They just care about that one thing If they get that one, because they have this diversified portfolio. However, if you're a founder or an early member of a team, you only have one bet, and so you may be that one that gets that $10 billion exit, but it's probably if you're playing the odds. You're probably not going to be that, and so, as you think about ownership and as you think about different protections you might allow an investor to have as part of a future funding round. These are just things I think to keep in mind. But, coming in, I just had this preconception of oh, this is just this market that just generates home runs and grand slams all the time, and they're actually just not as frequent as we'd like to believe or as much as we might like to hope.

Auren Hoffman: 4:37

How data savvy is the average venture capital investor.

Anand Sanwal: 4:41

The average venture capital firm by definition. Definition of being average is not very data savvy. That's probably why they're average. I think what you'll see with average firms is they'll look at something like App Store downloads or web traffic and they'll be like, ooh, graph goes up, that's a good company. And they'll call that data science and then they'll come up with some narrative about how they have a tool that helps them do this. It's good marketing to help separate LPs from their capital and then for those average VCs like those LPs will never see any of that money back. I think we have the benefit at CBI of working with what I would regard as many of the top venture investors, and I'd say it's not VC as an asset class has expanded, so it's the traditional VCs, but you've got corporates, you've got hedge funds, you've got sovereigns, you've got mutual funds. It's a more eclectic group, but I think some of them are using it in really data and intelligent ways, but I'd put them in the above average category.

Auren Hoffman: 5:36

What's an anecdote of an investor that's used data in a super intelligent way.

Anand Sanwal: 5:40

I'll give you a couple of examples. If I break down how the VC process works, the circle of life is see the deal, assess the deal, win the deal, nurture the deal. I think a lot of the focus tends to be on seeing the deal and that tends to be let me find the signal and if the line chart goes up, that's a good company. So I'll give you a couple of examples of seeing the deal. The one I like the most, which one of our customers I know does very religiously, and the term we use internally, or the term I use, is lineage tracing.

Anand Sanwal: 6:11

So what they do is they're obsessed with all the deals they didn't see that they wish they had seen. So what they will do is they'll use us and they'll look at their peers who did a deal let's call it a seed or series A deal and they'll just do a comparison of what companies their peers had done that they had not had in their own CRM. And then what they'll do is go back and look at where did that company come from? Did it come out of this accelerator? Was it funded by this angel investor? And they'll use CBI to figure that out. And then what they'll try to do is go build relationships in those areas so that that never happens again. And so I think they're a top tier fund and they're still humble enough to say, hey, we miss stuff and it makes us mad when we miss stuff. And so they go back and do this all the time.

Auren Hoffman: 6:53

You can't be in the deal if you don't even see it in the first place.

Anand Sanwal: 6:56

Elite firms sometimes might feel like I see the best stuff. All right, I thought this was a really rigorous way of making sure you'd never miss. So this kind of going back and figuring out the lineage I think is really interesting. I'd say the other ways that people use data to see the deal. One is from a thought leadership perspective. So, yeah, matt Turk at FirstMark he came out with this crazy AI landscape and used CBI data for that. But I think that's a really interesting way of being this lighthouse for founders and companies in a space to say, hey, listen, we know this space really well, we really care about it. And then using data and thought leadership to attract the right types of founders for you.

Anand Sanwal: 7:37

And then, when we see customers doing stuff with our data like that, we love it, obviously, and so then we'll distribute it to our newsletter, which goes out to 600,000 or so folks. Hopefully we can amplify that. So I think that's another thought leadership way. Just content marketing, yeah, but using data to say, hey, we care about this market and we've spent time to understand it. And here's some things we've seen, and I'd say the final would be using what I would call non-traditional signals or mashing up traditional signals with non-traditional signals, looking at things like commercial transactions. So which companies have a lot of vendor partnership, customer signings, headcount growth coupled with, maybe when they last raised, we have some scores around like management quality. I think folks will use that data. They'll maybe combine it with existing data they have and maybe they have their own model. Those are, I think, interesting ways of seeing the deal.

Auren Hoffman: 8:31

What type of data on a company do you wish you had, but it's just too hard to get? Or something like gosh oh God, if we could only have this kind of data. Or people have asked you for this thing, but it's just incredibly difficult to get Probably the one that folks ask about is market share.

Anand Sanwal: 8:46

So within a space, let's say, there's 10 players, who's rising, who's falling? Yeah, and how is that trending over time? Sort of like a measure of velocity and tractions.

Auren Hoffman: 8:57

Yeah, back in the day, if you're selling like an ATS to a company, you could crawl all the companies and see if it's on the website, but if it's an internal dev tool or something, there's no way to crawl that. There's no way to know, except maybe you check for new connections on LinkedIn. They made a connection to Macy's. Maybe they're selling to Macy's now or something.

Anand Sanwal: 9:16

Yeah, so we have commercial transaction data which says they're selling to Macy's, but what we don't know is how big of a deal was that? How do you have that data? Yeah, so we do two ways. We call them analyst briefings, so companies will submit information to us about their customers and then we'll validate that with the customers. And then we also will crawl media, their websites, and identify press releases talking about XYZ just signed with McDonald's and it's a POC or it's a full blown deal or it's a partnership. So we collect all of that data.

Anand Sanwal: 9:48

What we try to do when data like market share is really hard to get, we try to say, okay, what are proxies for it? That's a proxy for it. Right, If we see that, coupled with headcount growth, coupled with top tier investors we do interviews with software buyers. So if we see that the CSAT scores in those interviews are high, those are all indications of something that's good. And in those interviews we'll often ask, hey, who else did you evaluate? And so we get a sense for at least who's winning in the deals. Right, Again, it's not exactly market share. When you take all of these things and combine them, I think you get a pretty complete view of what might be happening with the company and with the market.

Auren Hoffman: 10:27

It's interesting because a lot of these vendors go together, so people who buy this vendor tend to buy this other vendor, even though these vendors may not have any type of partnership or any type of thing. If you look at like a DNA strand of a company, if you just looked at all the vendors, every company is unique but they can slightly rhyme with one another and I imagine it could be quite predictive if you could see company A is doing really well. Okay, company B is usually one year later selling to the same companies as company A, then that might be also a good time to invest in that.

Anand Sanwal: 10:59

Yeah, absolutely what we've actually seen. There is really predictive of M&A, and so what we actually find is some strategics that just have a tendency to try before they buy, and so you'll see a partnership.

Auren Hoffman: 11:13

Salesforce, for instance, almost always partners before they buy.

Anand Sanwal: 11:17

Exactly, yeah, salesforce, there's a bunch of healthcare pharma companies that do that as well, and so we'll see that pretty regularly where you'll see a number of partnerships, sometimes with one strategic within a category. So maybe they're trying a few different players out. So you now are starting to figure out okay, they're likely to make a move at least in this category, and then sometimes what you'll see is one of them went from proof of concept to full blown deal and then that probably becomes your favorite for future acquisitions. There's a bunch of things to your point around, things that get bundled together that you could look at. I'd say our customers tend to look at that information from the perspective of predicting future M&A and predicting competitive strategy a lot. So, yeah, it's really useful from that perspective as well.

Auren Hoffman: 12:07

Now, if you think of a data company, there's one thing okay, there's data companies that just sell data, high quality data and then there's other data companies that will build an application on top of that data so that more people can access it. You guys do both both sell data and you've got an application. Do you have some belief? Maybe one is better than the other, or how do you think about that? If you were advising other so many people listening to this today that are building data businesses, how do you advise them to get started?

Anand Sanwal: 12:33

Yeah. So I don't know if it's an either. Or I think some folks want pure data and they want to assimilate that into their own models and applications and this would be like a little bit more of a sophisticated customer Definitely more sophisticated. And then some people want the data to be more processed and they want maybe to serve up the insight on top of it. I'd say, generally, what we found is people will start at the more processed level in the application, in the terminal we'll call, and then they'll move to maybe more of the API product.

Anand Sanwal: 13:03

I think having both is really, if you can get somebody on both SKUs, what you've done is you're now clearly part of their workflow, your retention is going to go way up, your NRR is going to go way up. So I think if you can have both, that one-two punch, that's great. Not everybody's going to be ready for integration into a system, but I would say what we have been finding is increasingly, in the beginning of CBI, the customer was a person on the terminal doing something. Increasingly, the customer is other software that is just ingesting the data and doing things with it that obviously people are programming. But yeah, I think there's a lot of opportunity to do both.

Anand Sanwal: 13:44

I think with AI it probably changes that even more the insights on top of the information, because generative AI especially does this amazing job of just this phenomenal compression engine. I think the terminal sort of will change from I call it like a vertical search engine at times. Sometimes these data companies will resemble that to more of an answer engine, and so it'll proactively serve up things that you care about, because it knows things about you based on your company and things you've done in the past in the platform, and so I think that's a lot of the future of how data is going to be interacted with and consumed, is it's going to be served up to you and it's going to feel like you have your own personal data analyst who's just always on and always able to do just phenomenal things.

Auren Hoffman: 14:27

When we started Safecraft, I had thought that companies were going to grow their data science and data management capabilities much faster. If you look at just a number of real buyers that really spend a decent amount of money buying data today, it's a little bit more than it was five years ago. I predicted it would have been 10x more and it it's a little bit more than it was five years ago. I predicted it would have been 10x more and it's grown a little bit. Why do you think it hasn't grown as fast as at least people like me predicted back five years ago?

Anand Sanwal: 14:54

I'd agree with you. It's definitely grown slower than might have expected. I think it's grown in certain pockets like hedge funds use it.

Auren Hoffman: 15:02

I don't know. There's less than 100 hedge funds that really buy any serious data. If you think of Bridgewater, which is one of the biggest hedge funds in the world, the amount of data they buy is minuscule. It's so small as a percentage of their AUM. It's not even like a rounding error. Most of these you're talking about 0.72, citadel you're talking about this very small number of hedge funds that really buy any real amount of data.

Anand Sanwal: 15:22

Yeah, I'd agree. I think part of the challenge with these efforts is that data science capabilities and roles in these large organizations, I think generally is regarded as a back office function, and so I think part of it is just really great. Data science talent wants to be closer to the product and the customers that it impacts, and so in large orgs part of it is I just don't know if they're going to be able to get the quality of talent to really do data science well. And then I think the other thing in large organizations and I'm probably a little bit more enterprise oriented in this comment is that I think these data science efforts often feel like skunk works, so they buy it and it's not like a recurring revenue thing or something.

Anand Sanwal: 16:02

Yeah, and they're hoping it generates some sort of alpha, and if it doesn't, when the economy turns or cost cutting has to come into play, it ends up being the first thing that's killed. Part of it is that it is often driven by somebody who is data curious and conversant in data, and so they just want to play with information and data they haven't really syndicated internally. Here's the value of what I'm going to do. They're just throwing out. We're going to have a data science capability, but having a function is not value you need to articulate. Here's what it's going to help us do, and it's going to help us grow revenue, cut costs, reduce risk, and if you're not talking in that language, then you're just this nice little fun science experiment and science experiments they don't last generally.

Auren Hoffman: 16:49

Yeah, we saw that in 2022, where it just started separating, where we had customers who were clearly getting value from the data, and then there's companies are like oh yeah, this is interesting, and half the team we sold to got fired or something, and those were the turns that happened. In some ways, that's our bad, because we should have done a better job of understanding that customer and really trying to make sure that they got the right value from it, or really helping them get value faster from those types of things.

Anand Sanwal: 17:16

The point you raised at the end. There is really important because those prospective customers are often sometimes the most fun to speak with. They're doing cool stuff. It feels like they're speaking the same language as you and so you can actually burn a lot of cycles as a team and as a scale-up or a startup trying to get them on board.

Auren Hoffman: 17:33

They might need a little tweak or something.

Anand Sanwal: 17:36

Yeah, and so you end up taking your eye off the ball a little bit because you're like this could be such a cool application and if they get it in it could grow to 10,000 people in the enterprise and it's very easy to get ahead of yourself. I know we've made that mistake.

Auren Hoffman: 17:50

I'm 100% guilty there as well. Yeah, I'm really interested in the idea of data exhaust businesses. They haven't really taken off the way I would have thought they would. Either You've got a company it does whatever it sells widgets, but it's generating this interesting data that might not be core of the business and then maybe selling it to somebody else who could use it. How do you think about that?

Anand Sanwal: 18:14

Yeah, so I love this idea. I call them sawdust data. So I think somebody could build a business here of going to organizations who have this sort of sawdust and building businesses and licensing that data and creating businesses off of it. I think for the company that is creating the sawdust, I think the challenges are a few fold. One is oftentimes they just don't recognize that they have interesting data exhaust, that they're creating the sawdust that has value. Part of it is just a recognition of. They're not recognizing that value. That is driven, I think, mostly by the fact that it's not core. They're running a business and so this data exhaust thing it's just not something they're thinking about. And when they do think about it, I think oftentimes when I talk to folks they're oh, that's interesting, but is that a distraction and is that going to be a lot of brain damage to do that thing?

Auren Hoffman: 18:57

I've had the opposite experience. Sometimes they think it's way more valuable. Oh, this is going to be worth $500 million and it's worth $5 million and so it'd be a nice. They could license it for 5 million bucks, make 4.8 million in profit which is really nice and just manage it. But then they have this, their eyes get so much bigger and then it never happens because they think they have to invest too much in it or it never really goes anywhere.

Anand Sanwal: 19:23

When I've spoken to folks. I've spoken to folks at relatively large enterprises who I think have interesting data and for them they're always thinking about is this a needle mover? And so, in the context of the size of their business, yeah, it almost never is a needle mover.

Anand Sanwal: 19:37

Yeah, and I'd say the final one is reputational concerns is that if what they're licensing or putting into a product, it might be totally mundane, it might be totally safe, but there's this worry that even if there's a perception from customers that somehow their data is going to some clandestine thing, that's happening. And so you got this challenge of, hey, this might not even be a big opportunity and now I'm going to take reputational risk. That might even be, just might be bad perception. Is the brain damage of it worth doing? But I do think there's an opportunity here. But somebody just needs to go in and really alleviate those concerns and I think there's a big business to be built here incubating these sawdust data businesses for sure.

Auren Hoffman: 20:19

Now, if you think of companies that sell data, there are not very many of those that have become unicorns in the last 20 years. There's probably 2000 SaaS companies that have become unicorns in the last 20 years, and I don't know. There's Zoom Info and maybe one or two other data companies that have hit that unicorn status. Why is it so hard to build a really big data business?

Anand Sanwal: 20:41

I might disagree a little bit on the underlying premise because I think if you look at S&P, Faxed Wait, but all those were built like 50 years ago. But I would say not all of them, but some of them have continued to innovate.

Auren Hoffman: 20:54

Yeah, yeah, they're incredible companies and many of them have regulatory capture and other types of reasons why they're so good. Okay, so if you're talking about newer companies, If you're a VC and you only invest in data in the last 20 years, you really missed out. You could have done so much better investing in SaaS.

Anand Sanwal: 21:10

I think what ends up happening is that the data is Maybe. In a lot of these businesses, the data is not the core product. The data they have is an ingredient to solving some other problem, and so they're not positioned as data companies. But yeah, I'd agree If you think about pure play data companies. There haven't been major exits and major outcomes in the recent past. I do think data as a vector for differentiation is going to increase, and I think GAI in particular is going to allow a lot of interesting data businesses to be built. The thing that I'm most interested in, or most excited about is I actually think, with the things you can do with GAI, you can actually build really big data businesses with very few people going forward.

Auren Hoffman: 21:53

Because if you have to link data or merge data or crawl data, all those things are probably things that an LLM are quite good at.

Anand Sanwal: 22:00

Yeah, I've been playing around. Some of the data labeling and extraction stuff that you can do with LLMs is remarkable.

Auren Hoffman: 22:07

So much of data companies are just like an ETL and just like data munging and stuff.

Anand Sanwal: 22:12

Yeah, and so all of that sort of dirty work that you often had to do that required people to train and label data. Now you can do that really quickly. You just need domain knowledge, probably to know what questions to ask and how to get to that data. So I actually think we're at a unique moment right now where there's going to be some really large data businesses built with very small teams. I think that's a really interesting thing that's about to happen.

Auren Hoffman: 22:36

But in some ways then you don't really need much capital. Or maybe you need it for making proprietary data deals or licensing data or something, but you don't need that much capital because the main costs in the past were having really talented engineers.

Anand Sanwal: 22:47

You might raise capital to blow out, go to market once you have fit and even there I think, with PLG and APIs and other things, you might be able to get quite far. Obviously, enterprises want that high touch, but, yeah, you might be able to get quite far without a lot of that other legacy. What legacy companies have generally have to build? Now how?

Auren Hoffman: 23:07

else should data companies be thinking strategically about the rise of AI?

Anand Sanwal: 23:11

I think the biggest one, in my opinion, is around this sort of autonomous research analyst. Getting the data embedded in workflow is really important, and you're going to know what the customer cares about and then you're going to serve that up to them in the format and in the place that they find most valuable. So, instead of having to come to the terminal, if you live in your email or you live in Slack or you live in your CRM, it'll get served up to you there, and it won't be just served up to you as columns and rows. It'll be served up to you If you like it. Like that, you can get it, but it'll be served up to you in a narrative form, and so I really think this idea of your own personal data analyst is really increasingly. That's going to be where the world is going.

Anand Sanwal: 23:53

I think right now, what everybody's doing is coming out with their own chatbot. That's like a bridge to the final destination. I think the ultimate destination is you just feel like somebody is sending you the personalized thing that you need to do so for our customers. As an example, it's been eye-opening A lot of our customers when they say they're in competitive intel. A lot of their work is compiling information and sharing it with the rest of the organization. Think of it like an internal newsletter. Hey, I work at Big Automaker. I need to know what Tesla and Ford and other competitors are doing.

Anand Sanwal: 24:25

Another data munging that goes on. You spend so much time data munging and you spend very little time actually analyzing what's going on. I think that data munging work can be done proactively and on your behalf and just sent to you, and then you become less of a creator and more of a curator and editor. Just sent to you, and then you become less of a creator and more of a curator and editor, and then you're off analyzing and doing things that I think humans are obviously more qualified and more capable of doing. So, yeah, I think that's where differentiated data companies are going to be going in the next five, 10 years.

Auren Hoffman: 24:56

If you think of just the average information worker, it seems like so much of our time is just clicking on things. It's moving from here to here and going to this application and clicking around. If you want to go into your CRM, it's 15 clicks to do anything, it seems. All those things. At least you could have an agent in the background helping you get that information faster, doing those things for you, et cetera, and that will hopefully get us having a lot more time to actually either be creative or do our own analysis or do other types of stuff. Or do you think that's a world that's happening and you think that's going quick, or do you think that's more okay, it's in my fantasy.

Anand Sanwal: 25:36

Yeah, I think that area is certainly already happening automating the existing workflow. I think what's as interesting as that is doing things that make you bionic, that you couldn't do before.

Auren Hoffman: 25:48

Instead of just having a team of humans that could have done it. Now you do things that no human could ever do.

Anand Sanwal: 25:53

Yeah, we're about to put quarterly earnings filings and presentations on our platform, and what somebody might do previously as an analyst might look through a competitor's quarterly results and do some summary of it. What's really hard is to say, hey, actually now tell me over the last 12 quarters what parts of their strategy have stayed consistent and what has changed. They just couldn't do that, or it would be like a multi-week effort. And so there's new knowledge that actually gets created by having access to all of this information and combining it and organizing it appropriately. And I think that's what's. There's new insights here that I think just people haven't seen before. And then to your earlier point there's just a lot of sort of those mundane clicks around and let me go copy paste this information into the spreadsheet. That stuff's certainly going to go away, but I think the new intelligence is actually coming fast and I think that's going to be really compelling and really valuable to organizations and to people as well.

Auren Hoffman: 26:50

Obviously, you've got all these earnings calls, you've got expert networks like TGIS or something like that. You have just tons and tons of data, transcripts, people talking about certain things and reading all those takes a really long time, and then being able to sift through it all these things seem like perfect for some sort of AI agent to help you do all those things.

Anand Sanwal: 27:10

Yeah, I think the key there is you have to be very clear, upfront, about how you're going to organize even something like a transcript. So when we do a software buyer interview, it's very structured. So we always ask about how much did you pay? What's your intent to renew?

Auren Hoffman: 27:25

Because you're doing it. Rather than having someone else do it, you're doing it yourself.

Anand Sanwal: 27:29

Yeah, but the reason our interviews are structured is because then with GAI, it makes it easier for it to pull out the requisite information and then you can actually create structured data products around that, and it reduces the risk of hallucination, because I think that's the other big thing is again for us. When you're selling into an enterprise, you want to make sure somebody doesn't go into a meeting and get pants. They have to feel good about the information that they're putting in front of their board or in front of their boss, and so in some sense, having even your unstructured information be semi-structured makes that a lot easier versus it just being a free-for-all interview style. But yeah, I think there's a huge opportunity in unstructured information and extracting structured signals out of it, and what do you think about these?

Auren Hoffman: 28:12

future of these? I call them all expert networks, whether it's CB Insights obviously Gartner does certain things TGIS, glg, alpha Insights, et cetera. Where do you think all of these are going? Kegas, glg, alpha Insights, et cetera. Where do you think all these are going?

Anand Sanwal: 28:22

The service they provide is very important, and some of it, I think, is people perceive it as being valuable. If I read it online, I feel like everybody's got it. If I talk to an expert, I feel like only I had it, so there's a perception of uniqueness that I think is really valuable there. I think the model of having transcripts and then being able to have scale with those transcripts is ultimately a better business than coordinating and facilitating one-on-one conversations. It just becomes an asset that you can build one, sell multiple times and then I think, if you structure those transcripts appropriately, you can actually create all sorts of derivative assets off of them. That I think is really valuable. So I think the idea of conversations with experts or software buyers or what have you I think is very valuable. I think how you go to market with it, there's certain ways to make it more valuable. I think.

Anand Sanwal: 29:15

The other thing is I do believe there's going to be vertically specific transcript services, and so I think within healthcare, if you want to understand the efficacy of a drug and you want to talk to doctors, I believe there will be or maybe there already is, and I haven't seen it a service that interviews doctors and all those transcripts are available and so you can now see what a thousand doctors feel about a particular drug and what they're seeing. I think if you pick high value verticals where people are making big decisions off of that information, I think there'll be a lot more of those versus these general purpose systems, because I think what you need there is you need density of coverage. Having five interviews with doctors and then having 10 geopolitics experts who talk about the Ukraine and seven who talk about what's going on with Snowflake, you don't have density. I think if you can go really deep on particular verticals, then-.

Auren Hoffman: 30:07

Especially if you structure it and you're always asking these doctors about the other stuff as well, because if you're only asking about one drug, then it's hard. But if you get them on the phone and you ask them about 15 different drugs and 15 different medical procedures and other things, then you can get that learning much faster.

Anand Sanwal: 30:26

Yeah, and I think you then also get out of this rat race of recruiting experts all the time, because you build affinity in a relationship with, let's use, the doctor or, in our case, a software buyer.

Anand Sanwal: 30:31

We know if you buy Salesforce, you probably bought a bunch of other sales tools, and so we can go back to you and talk to you about your experience with all those. And we can go back to you when you told us you were going to renew Salesforce to see did you actually renew and what did the price look like in that following renewal. So there's a bunch of things that you can do by going deep that, I think, reduce some of the other. I'm sure you get them all the time on LinkedIn, these sort of spam messages from these networks about hey, can you talk to us about this company? And it's a relevant cost that they have to incur to recruit experts constantly, and so if you can reduce the amount of expert recruitment that you have to do and just go deep with a particular set of doctors or software buyers or what have you, I think there's an immense amount of value to be built there.

Auren Hoffman: 31:17

I've got this scam I want to create, where it understands your voice and then it does all these expert calls for you on the side, and so you could have five expert calls every day, and you're not even having to do anything, you just earn all this passive income on the side.

Anand Sanwal: 31:33

So you've just taken it to a whole different level. Yeah, I think if you got, what is it? 11 labs or whatever? Get one of those AI avatars. Train them on your voice.

Auren Hoffman: 31:42

And then such a smart LLM you literally can talk about anything. So, yeah, I can talk about software, but I also could talk about the deep neuroscience that's happening. You're just a smart guy. Everyone's asking you all this and you just make you all this passive income on the side. You're just on the beach.

Anand Sanwal: 31:56

Like you're Oren 2.0, who's just out there doing expert calls all day. Yeah, just make it all this big for you. Yeah, I think GLG was supposed to go public and they're going to have to put that into their risk factors now when they issue their next S1.

Auren Hoffman: 32:09

So recently I went on one of these things and looked up SafeGraph and I saw these two buyers who talked about their experience being SafeGraph customers, both of which were at companies that I would love to have as clients but were not our clients. I feel like at some point we've certainly talked to them, tried to sell to them, but they've talked like they've been our client for years. They said all these nice things about us, which is very nice, but they clearly were not. We have inside knowledge. We know who our clients are. Either these expert networks were changing the name of the company instead of like Coke to Pepsi or something like that, to keep them more anonymous, or there's a lot of fraud on these things. I don't know if you have any ideas about it.

Anand Sanwal: 32:51

We haven't seen that and part of it is we do a lot of vetting upfront, we ask a bunch of questions and we kick you out of the process. There's a whole Reddit about how to get expert network survey calls and calls. There is people who are trying to hack this. They're asking questions like hey, who's the least strict? Or hey, I found this hack to get around this. Or hey, here's a survey that pays $150 if you take it and you just got to answer their qualifying questions this way and you'll get put through.

Auren Hoffman: 33:22

Or you could just earn $1,000 a week or a few thousand dollars a week on the side, people will try to find easy money when it's available.

Anand Sanwal: 33:29

I think it tends to be more on the market research survey side of things versus the call side, I would say from what I've observed, because as soon as you have to get on the phone you can get exposed a lot more quickly and then I think you get paid based on the amount of time you're on the call. So I think people tend to be trying to find ways to get those surveys where they get paid whatever 150 bucks for just checking some boxes. Generally there's a little mini ecosystem of people who are trying to game those surveys and make a little bit of extra coin for sure.

Auren Hoffman: 33:58

I love learning about these scams. It's very fun. You actually just stepped down as CEO after 14 years. I'd love to walk through that thought process. How do you know it's time to step away as the leader? Obviously you're still chairman, but how do you know when to do that and what was going through your head? How do you plan for it, etc.

Anand Sanwal: 34:16

I try to be pretty structured in decisions, especially big ones like this, and so my framework was simple but, I think, effective. On one axis I thought about what do I like to do and want to do and people have different frameworks of that, the zone of excellence versus zone of competence, type stuff and then on the other axis was what the company needs. So if I break that down, I think on that what I like to do. I'm a product founder, I like the new and novel, and the company had scaled to a point where it needed systems and repeatability.

Auren Hoffman: 34:48

It's more one to end than zero to one.

Anand Sanwal: 34:51

Yeah. And as I thought about, okay, how do we get to 300 million? It wasn't clear that, as I thought about what would be required of me, it wasn't clear to me that I would enjoy that. I like being outside the organism, I like talking to customers and I like thinking about product and thinking about content. So on that axis it was like, hey, listen, what the company needs or what I like to do on, maybe on the low end of that axis. And then the other one was what does the company need? And I think there's these different adages. I told the one I like different wars need different generals. I think there's these different adages. I told the one I like different wars need different generals.

Anand Sanwal: 35:27

I think founders often ask that question about everybody else on their team. So they'll often say, oh, so-and-so is great VP of sales from zero to one, from one to 5 million or five to 10, are they the right person? I think it's the right thing to do to also ask that of yourself. And so I asked that of myself and said, hey, listen, I think if I assess myself on that same dimension, I don't think I'm what the company needs. It was one of these things where I thought, okay, I know the things I need to work on. Should I spend my time trying to shore up my weaknesses or should I lean into my strengths? And then, when you go back to what I like to do, I was like I really don't want to spend my time trying to shore up my weaknesses, and I was probably low on what the company needs access as well. So it was in the perfect quadrant for, hey, it's time to make a change, and so it was something I thought about for a while.

Auren Hoffman: 36:13

How do you figure out what that company needed? Because then it's really okay, I got to replace myself with someone. Obviously, there's probably some great internal people. There's probably some amazing external people. How do you then make that decision?

Anand Sanwal: 36:26

Yeah. So once I knew that I thought it made sense for me and the organization. Then it was building out the profile, I think, of what this potential CEO would look like, and so I had some hard criteria and some things that I would need to assess in our conversations. So one was I wanted somebody if I think of CB assess in our conversations. So one was I wanted somebody.

Anand Sanwal: 36:43

If I think of CB Insights as being at stage N, I wanted somebody who had experience growing a company to the N plus one or N plus two stage. I didn't want somebody who was N plus 20, because they just don't have muscle memory for what we need. So that was one. Somebody who has a track record of success in a challenger company. So the challenger piece is really important to me, which is, I think, what I underestimated when I left American Express to start CB Insights, was people used to return my calls at American Express all the time and I used to think it was because of me and it was actually just the logo on the business card. So somebody who had a lot of success at a high flyer company that everybody knows about.

Auren Hoffman: 37:23

Right, they're at Google. Everyone's going to call you back when you're at Google. Yeah.

Anand Sanwal: 37:27

So Google wouldn't work because it's N plus 50 from us Plus it's fun to sell when you're a monopoly, so that didn't work. And then, as I talked to folks, I just had certain things that I was looking for. When I got late stages with a few folks one as I talked to references I wanted to see extreme followership, not people who are followers in the. They don't think way.

Auren Hoffman: 37:48

Oh, like where the executives go with them to the next thing and stuff.

Anand Sanwal: 37:52

Yeah, where their references were like I'd love to come, I'll be there tomorrow, that was a big one. And then the other one was I really wanted somebody who had good instincts. I always would ask this question you come in, what's the first 90 days? Look like. And if somebody told me their first 90 days was they'll do a listening tour, it was over immediately, because that's the classic answer, such a pat answer, and it's kind of nonsense because it's okay, you're going to listen for 90 days. So basically nothing's happening in year one.

Auren Hoffman: 38:19

By that point they already know a lot about the business. They know about some of the other execs, they know about lots of things. So I don't know everything, but here's three things I do know that I would do, yeah.

Anand Sanwal: 38:29

I'd share data with folks prior in the interview process. I wanted to see that they had hypotheses and that they were mostly right, like that, their instincts seem to be driving them in the right direction and so, yeah, I looked a lot for that in the hiring process and then once the person was hired and landed in the role, then it was all about transitioning and prepping them for the role. Then it was all about transitioning and prepping them for the role?

Auren Hoffman: 38:53

Did you care if they were a CEO before, or were you willing to say, okay, I'm willing to step into a CEO role even if they've never been one before?

Anand Sanwal: 39:01

Yeah, it didn't have to be that. They had to have been a CEO. They had to have led a big go-to-market organization. That was the number one criteria. The ideal profile was somebody who had built a startup or done a startup also then scaled the company to the N plus one, n plus two stage from us. And yeah, I think we're fortunate in that we found that.

Anand Sanwal: 39:22

And then there's the hiring spent many months with Manlio, who we ultimately hired, and then, once he landed, then it was all about figuring out how to ensure that he's prepped. But actually, interestingly, I think that I was prepped because 14 years in a business, there's probably a lot of things I could have done wrong in the transition, and so I spent a lot of time talking to founders who had transitioned out. I talked to some just killer investors who invested in the best companies in the world and I asked them when you've done sort of this founder to exec chair transition? I just asked them what have you seen done that made it go really badly? And so I just zeroed in on those stupid things. I'm just not going to do those things. And I heard some recurring themes and it was okay. I'll just need to be vigilant.

Auren Hoffman: 40:10

Because one thing I can imagine is just there's a muscle memory of all the other people going to you and asking your thoughts on it and stuff, and at some point you're like, hey, you shouldn't go to me. You got to either solve yourself or go to the CEO. Now I'm only here for these very small number of things. You don't want to be getting involved where the CEO should be getting involved.

Anand Sanwal: 40:28

The reality there is I'm a ghost, manlio is a killer, he's the CEO.

Anand Sanwal: 40:34

He and I chat regularly, but I like to think of it as I'm an invited guest now, and so if he needs support or help or wants me to weigh in on something and talk to somebody on the team about something some product thing or whatever research thing he'll be giving them a heads up that hey, anand's going to reach out, but I'm not slacking with anybody communicating with anybody, unless he and I have chatted about it before, or sometimes customers still just know me and they'll send me something.

Anand Sanwal: 41:01

So then I'll forward it on to the CS team or whatever and I'll just CC him. So he's aware, but it was really important to make sure that it was very clear to me, to him, to the organization, that he's in charge and I'm here. I'm obviously on the board, so I'm here to be helpful, but it's only when, at his direction, I'm curious what grade he'd give me. But I think I've done a pretty good job on this transition. We have a really great relationship and we talk regularly and so we're pretty open about things if they're maybe not going great, but yeah, it's been great. It was like a big change, but I think it was absolutely the right thing for everybody on the team and for CBI and for myself.

Auren Hoffman: 41:39

Now you're prolific on Twitter. I actually follow you at a Sunwall on Twitter and what I love is that you don't really take yourself too seriously. What's your personal brand strategy?

Anand Sanwal: 41:50

So I don't have a personal brand strategy. To be really honest, I hate the whole concept of personal branding. I think personal brands are an output, not an input.

Auren Hoffman: 42:00

If you do good work or say interesting things, or just keep doing this, whatever you do, because you like something.

Anand Sanwal: 42:07

And I've had this debate with folks on our team at CBI who are younger than I am about this whole thing and I just feel differently. I think ultimately on Twitter, our newsletter or LinkedIn or wherever I might post, I try to write like I talk, my friends who've known me for a long time will read some stuff and they're just oh yeah, I could hear you saying that. And then I think the bigger things with the social channels that's maybe not as obvious is that I use them as an open mic, the way comedians try out material, not to say, I'm a comedian but I'm going to try a research idea out, or I'm going to try some new idea out. If it works there, then it's like okay, that should go into the CBI newsletter or that should be a research product or that's a vein that we should tap into. I'll write some stuff there and sometimes I've even heard from members of the CBI team yo, that was a little bit extra and I'm like I'm just trying stuff. Sometimes it's going to miss and that's okay. It gets subsumed in the feed so quickly anyways, like people forget and I'm not going to do anything really crazy. That's going to cancel.

Anand Sanwal: 43:07

It is an area where I can just do quick experimentation and sometimes I'll even try just subject lines for the newsletter. I'm like I'll try a post and then I'll delete it and then I'll retry it and sometimes the second one works better different format. So yeah, I think it's just a great way to see what resonates and what doesn't. And the personal brand thing. If that is an outcome of it, that's okay. The other thing I don't like about personal brand thing is I think to do it really well, you actually have to be this sort of one-dimensional character. I like a lot of different things and I like to bullshit about a bunch of different things and I don't want to be like, oh, people will think that's not part of my brand. Again, that is great for folks that are trying to build that, but that's just not what I'm in it for but that's just not what I'm in it for Two last questions I want to ask everybody.

Anand Sanwal: 43:56

What is the conspiracy theory that you believe? Okay, I love this question that you ask it. When I listen to the podcast, I think this is always really telling. You can grade how far out there this is on the conspiracy spectrum.

Anand Sanwal: 44:03

My fundamental belief is that I don't think the education system in the USA is actually focused on teaching kids how to be literate and numerate and how to think. I think it's actually primarily focused on teaching conformity and compliance and what to think, and I think, ultimately, success is getting you to sit in a classroom for seven hours a day so that you can eventually go to college and then eventually go to work and sit for seven hours in a cubicle. I don't think it's about numeracy and literacy, which I think is a shame, and I think I've been digging into a lot of the data around the outcomes that our kids are getting, and I'm a public school product and it's sad what we're doing to kids in this country, and it's not because they don't want to learn. I think it's just because the education system is architected in a way that it's about conformity and compliance and teaching you how to think.

Auren Hoffman: 44:56

There's certain things, that there's lots of different ways to learn stuff. Duolingo teaches language pretty well. If you just made sure the kids, instead of taking whatever Spanish or French or something like that, if you just watched them, just made sure they were actually doing Duolingo during that class, they'd probably learn a lot more than they would from the teacher there. And so there's probably like a lot of little things you could do. If your real goal was increasing learning which I don't know that it is, but there's a lot of ways you could increase learning a lot faster.

Anand Sanwal: 45:24

We've forgotten who the customer is, and so I think there's things that work, just as an example, the model of cram exam erase, which is the prevailing model to learn stuff, where you basically just purge it from memory after you take the test. There's no long-term retention there, everybody recognizes that, and there's proven methods of quizzing students regularly, for mastery leads to long-term retention. It's hard to do, and so I think a lot of times we try to make the class exciting and interesting, because I think teachers want it to be exciting and interesting because it keeps this exciting and interesting for them. There's actually education science here that says these things work to create, and then I think the other big thing with conformity and compliance is that we have a system for, even at its best, it creates kids who are really good at reading maps, in the sense of they know how to navigate and check a bunch of boxes that help you get into the right high school and then help you get into the right college and that help you get into the right job.

Anand Sanwal: 46:25

It's actually not set up for building kids who are explorers, the kids who are actually going to build a map. It does nothing for them. So, even at its best, it still is about creating conformity and just having you check a bunch of boxes. And I'm a big entrepreneur. I think entrepreneurship is a phenomenal art, and so I think we should be creating mechanisms for more kids to become explorers who don't just read a map, but who develop a map. I can go on forever on this topic, but yeah, I do think the education system in the USA is pretty broken. I imagine it might be similar elsewhere.

Auren Hoffman: 46:57

I'm in the middle of a book called Bad Therapy. The premise is that so many kids today are in therapy. But therapy it's something that can help people very acutely and especially for adults that need help and can understand when the therapist is pushing them the wrong way. Help and can understand when the therapist is pushing them the wrong way, etc. But for most of kids they glean too much on what the therapist says, and everyone knows therapy can both do well and cause harm, just like a surgeon can do well and cause harm at the same time. And you don't just go for surgery when you don't need it and you need a lot of thought just because a doctor suggests surgery. You need a lot of thought sometimes to decide whether to do it or not, and so it's really diving into how so many kids are on it and where in many cases it's actually doing more harm than good.

Anand Sanwal: 47:44

Yeah, that's really interesting. Yeah, we should chat separately at some point. I'll give you another book, though it's one called the Case Against Education, if you've read that by Brian Kaplan. Yeah, brian Kaplan, he's great. Yeah, it's definitely led to the most arguments between my wife and I.

Auren Hoffman: 47:58

Brian lives near me. I've had him over for dinner before and just a super fantastic, interesting guy, All right. Last question we ask all of our guests what conventional wisdom or advice do you think is generally bad advice?

Anand Sanwal: 48:09

So I'll go back to the company building side of things. I think the thing that I heard a lot and maybe this was a Zerp era phenomena was hire good people and get out of their way. I think this is abjectly horrible advice. The job of a manager it's in the name. You have to manage and that means you need to have mastery of details, you need to inspect things. I think it often people immediately go to oh, you're micromanaging, and there's a whole spectrum before you get to micromanagement. But I think the hire good people and get out of their way. Often what that led to was abdication instead of delegation, and I don't think organizations get better I'm not sure teams get better by doing that. It just seemed like an easy way for inexperienced managers to not actually have to help their teams and help their people get better.

Auren Hoffman: 48:57

It's interesting because that's also the advice I think a lot of venture capitalists give that advice hire good people and get out of the way. And I think if you're running a venture capital firm, I think actually maybe that actually makes sense because you're hiring a very small number of super elite people and it's very decentralized and there isn't a lot of coordination, and so if you had a six-person venture capital firm, you hired another partner. It makes perfect sense to do that. It actually isn't good advice to actually go to a software company or something.

Anand Sanwal: 49:25

I think if you look at some of the legendary founders of recent or even of old Brian Chesky or Jobs or Travis Kalanick or Elon Musk they care about the details and they understand those details and, again, they probably and hopefully many of those cases also do a good job of empowering the right people and hiring the right people. But yeah, I think under the veil of micromanagement is bad. We've probably let a lot of bad management pervade our systems and our organizations and hopefully people stop taking that advice.

Auren Hoffman: 49:57

All right. Thank you, anas Anwal, for joining us on World of DAS. I follow you at AsanwalAllen on Twitter, as I mentioned earlier. It's very fun to follow you there, so I definitely encourage our listeners to engage you there. This has been a ton of fun.

Anand Sanwal: 50:11

Awesome.

Auren Hoffman: 50:11

Thanks, Arne interesting to get you there. This has been a ton of fun Awesome. Thanks, arne. If you're a super data nerd, go to worldofdascom that's D-A-A-S. Worldofdascom and sign up for our weekly data as a service roundup newsletter. Thanks for listening. If you enjoyed the show, consider reading this podcast and leaving a review For more. World of DAS and DAS is D-A-A-S. You can subscribe on Spotify or Apple Podcasts or anywhere you get your podcasts and also check out YouTube for videos. You can find me at Twitter at at Oren. That's A-U-R-E-N. Oren, and we'd love to hear from you. World of DAS is brought to you by Safegraph. Safegraph is geospatial data for physical. We'd love to hear from you.