Ep. 41 Hero S2 Ep41 Andrei Hagiu

Decoding Platform Companies: Andrei Hagiu on Platform Strategy and Building Your Business


37 min

Ep. 41 Hero S2 Ep41 Andrei Hagiu

37 min

“I skate to where the puck is going, not where it has been.” – Wayne Gretzky When Andrei Hagiu wrote a Harvard Business Review case study on AppDirect, he used this quote from Canadian hockey legend Wayne Gretzky to show the mindset each founder should have before starting their company. Andrei is an expert in platform strategy and started a career in academics as an Associate Professor at Harvard Business School, as well as at MIT Sloan. Andrei’s research and teaching are entirely focused on platform businesses (e.g. Airbnb, Amazon.com, Facebook, etc.) and their unique strategic challenges. Today, Andrei is an Associate Professor of Information Systems at Boston University’s Questrom School of Business. In today’s episode, Andrei talks about the difference between virality and network effect, how to create a successful, multi-sided platform business, and the future of Web3’s decentralized platforms.

Read transcript

"You want to make it very easy for individual users to adopt it, right? That's the key because it matters to you to have a lot of users, and then it's easier from there to actually become a platform later on."

Quick takes on...

The difference between virality and network effect

"Virality just means...if you do a marketing campaign and your customers tell other customers about your product, it's of course a very useful tool. But that doesn't mean that your product all of a sudden has network effect. Network effect is basically my willingness to pay us higher for this particular platform or product if there are other customers on this platform."

Creating a healthy relationship between buyers and sellers

"Make it easier for you to transact with them, which in itself is an insurance against leakage, right? If all you're doing is making it easier for buyers and sellers to find each other, why would they come back to you once they found each other?"

Meet your guest, Andrei Hagiu

Associate Professor, Harvard Business School

Spotlight S2 Ep39 Andrei Hagiu

Andrei is a leading expert on platform strategy and business models. Prior to becoming an Associate Professor at Questrom School of Business, he taught platform strategy courses at Harvard Business School and the MIT Sloan School of Management. He uses the insights drawn from his research to advise and angel invest in start-ups building platform businesses, as well as to consult for large companies seeking to turn their products into platforms (e.g. ADP, Equifax, Intuit).

Listen to the next episode

Ep. 42 Home S2 Ep42 Jay Kaplan

Decoding Cybersecurity: Jay Kaplan on How to Protect Your Business from Cyberattacks


29 min

Jay Kaplan is a renowned security expert and entrepreneur who has served in many high-profile cybersecurity roles—including at the Department of Defense and the National Security Agency. He was also selected as Forbes 30 Under 30 in Enterprise Technology. After seeing a gap in the cybersecurity space, in 2013, Jay co-founded Synack, a company that’s developed a premier security testing platform. Today the Synack platform protects federal agencies, DoD classified assets, and a growing list of Global 2000 customers. In today’s episode, Jay talks about the rise of ransomware and crime syndicates, and he offers simple yet effective security strategies organizations, and individuals can apply right now. What are some of his biggest tips? Focus on your people, sensitive data, and endpoints.

Episode transcript

[00:00:00] Andrei Hagiu: You want to make it…

[00:00:00] Andrei Hagiu: You want to make it very easy for individual users to adopt it, right? I mean, that's kind of the key because it matters to you to have a lot of users and then it's easier from there to actually become a platform later on.

[00:00:20] Dan Saks: That was Professor Andre Hayu. Andre currently teaches at Boston University. And previously lectured at MIT SLO and Harvard Business School. Andre's research and teachings are entirely focused on platform businesses, think Airbnb, Amazon, and Facebook. His extensive knowledge on what makes these businesses successful in addition to their unique strategic challenges.

[00:00:45] I first met Andre as he wrote a Harvard Business Review case study on AppDirect as a multi-sided platform. Andre leverages the insight from his research to advise an angel invests in startups and consults with large companies seeking to turn their products into platforms. In today's episode, Andre talks about how to create a successful, multi-sided platform business, what the difference between virality and network effect is and the future of Web3's decentralized platforms.

[00:01:16] This is Daniel Saks, president and co-founder of AppDirect, and it's time to decode building platforms and marketplaces from the ground up.

[00:01:28] Welcome to Decoding Digital, a podcast for innovators looking to thrive in the digital economy. I'm your host, Daniel Saks. And I'll sit down with other founders, CEOs and changemakers to decode the trends that are transforming the way we work. Let's decode.

[00:01:52] This is Daniel Saks. Welcome to Decoding Digital. I'm super excited about this setup. So Andre was a professor at HBS when we first. And we were just starting AppDirect and we were doing research on platforms. And our business is a multi-sided platform, which means you have to connect buyers and sellers.

[00:02:13] And from an academic perspective, there wasn't a lot of research out there. But Andre really pioneered research around multi-sided platforms and now runs the Substack on Platform Chronicles and continues to publish work and teach. He's an innovator and angel investor, and a friend. So really thrilled to welcome Andrei Hagiu.

[00:02:34] Thank

[00:02:35] Andrei Hagiu: you so much, Danielle. It's been a while. It's great to be on this podcast to catch up.

[00:02:39] Dan Saks: Yeah, it's fun how the industry's come full circle. So before we go too deep on some of the research that we did together back when App Director was founded, I wanted to start just by getting your definition of what a platform business is, and I think that will give great context for our viewers.

[00:02:56] Andrei Hagiu: Sure, I know there's lots of definitions out there. So the one that I use, which I hope is quite clear, I think of a platform business as one which connects affiliated customers. So there's two things that have to happen, is the relevant customers have to be on the platform, obviously and there has to be a direct interaction.

[00:03:17] The platform enables direct interactions between these affiliated customers. So this could be, I mean, obviously could be something like, say social networks or messaging apps like WhatsApp or something like that, where the direct interaction is people connecting one another on the social network or exchanging messages.

[00:03:36] Or of course it could be two-sided platforms or three-sided platforms, where they're buyers and sellers. And the direct interaction in that case is there's a sale, there's a transaction between the buyers and the sellers. And the key things I would emphasize here is like what makes a platform business a platform?

[00:03:53] It's the notion of direct interaction, which means the platform doesn't fully control the transaction, right? So if you go to eBay or Airbnb or something like that, I mean, of course Airbnb and eBay have rules around what kind of certain things that can't be done, but by and large, the terms of transactions are controlled by the various parties.

[00:04:11] Right. Same thing on social networks. Yeah, of course. Like Facebook and others, they have their own like terms of service, blah, blah, blah. But again, by and large, like it's very different from traditional newspapers which have full control, editorial control, right? I mean, this is still like people are kind of posting and discussing as they see fit.

[00:04:27] So that's how I think about it.

[00:04:29] Dan Saks: So I remember first finding your article on multi-sided platforms and Web 2.0 was starting. Some of the examples you gave were pretty old school like Nintendo, where there's developers and games. Yeah. And people program the games, but there's been like this cambrian explosion of now platform businesses.

[00:04:47] And I feel like now people understand the business model, but at the time you really defined the concept of a multi-sided platform. And what I thought was really fascinating about it, that we kind of really reinforced in the early days of AppDirect beyond the network effects and the moats. For the notion where a good ecosystem or a good platform can reduce the search and transaction cost for that market, therefore creating more value for everyone and therefore justifying a bigger take rate or therefore compensation than what would've previously existed..

[00:05:16] And back in the day, like, you know, you take Nintendo, it's still required physical distribution, physical cartridge development, but platforms in the digital world, particularly the multi-sided platforms, have truly transformed and really enabled consumers to have a much more cost effective way of interacting with everything from banking to gaming, to content videos, you name it.

[00:05:44] But I wanted to kind of go back to that early phase, like what made you see that platforms would be such a huge transformation back in the day when there were very few examples at the time?

[00:05:55] Andrei Hagiu: I appreciate that you're giving me all this credit for seeing it. I mean, I think quite a few other people have already understood the potential of platforms at the time.

[00:06:04] You're right. I mean, I started doing research on this a little bit before the obvious Cambrian explosion, but I would say the seeds of the large platform businesses that we see today, so like Facebook, Airbnb and things like that, were kind of sewing around that time. Right. So I mean, again, I think business typically is like a part of academia in understanding these phenomena.

[00:06:22] I think we can be pretty good. Like sometimes, you know, academia is decent at sort of like formalizing making things a little bit more rigorous and maybe asking interesting questions. There's a lot of interesting stuff in there. So to answer like directly your question,

[00:06:38] I think what I found particularly fascinating was this notion, unlike in say, the traditional economy, right? So we basically have these like linear value chain models where basically, you know, raw materials producer like does something to the inputs that passes them on to the next node in the chain. They keep adding value and then the end, like there's a finite good producer that takes it and adds the last thing and then sells it to consumers and maybe then you get to retail.

[00:07:00] What I thought platforms were interesting is basically this idea that, if it all goes well, if I create a really valuable platform, like I said, all the value is basically enabling other people. It could be companies, could be users to interact in a very efficient way, and the platform doesn't have to do much, right?

[00:07:18] I mean, it's basically, I don't need to touch the product. If, again, if it all goes well, of course the reality is more complicated. Some platforms do a little bit of touching, but in the purest form, in their purest forms, platforms are just basically like, I make it easy for the two or more sides to find each other.

[00:07:33] So that's the search cost part. And the second piece, potentially, I could also make it easier for them to transact one another. And you know, you can do this by, you know, creating trust. You know, especially if it's people like who live in one like users in very different countries, very different locations that they know each other.

[00:07:48] So the magic quote unquote, of platforms is they create incentives and so it's like creating economic mechanisms for, there's the right incentives and then value gets exchanged and created without the platform having to like, it doesn't have to incur all the marginal costs that a traditional retailer or something like that.

[00:08:05] I think that's sort of the beauty of it. And again, I think there are a few people that have understood this early on. I thought, okay, that's super interesting. And I think it was pretty easy to extrapolate to see the first platforms like PayPal and eBay say, Hey, this kind of business model will probably go far.

[00:08:20] Dan Saks: Yeah, and when we started in 2009, one of the big challenges that we thought through is like when you take any market, what are those search costs and what are the transaction costs? So I just wanna give the live example of how you helped us critically think about our business. And we were saying we think that SaaS is gonna be a massive market, and now there's probably maybe tens of thousands.

[00:08:43] So the market grew faster than we thought. But what we also realized is that those SaaS companies were spending more than 50%, sometimes 100%, and they were burning on sales cost. So that high cost of sale is a huge search cost for them to find the customer. On the flip side, Customers trying to find the right software was really complicated as well.

[00:09:05] And by looking at your framework of a multi-sided platform, we said, okay, theoretically here's the market size and the more value we can extract is based on or predicated on the more search costs we save. Both the developers and the buyers, as well as the more value added through technology we can enable through the features or capabilities that we use to make people transact easier.

[00:09:31] But I want to kind of play it forward 15 years, because it's been a while since you wrote that initial case study and you invited me to the Harvard Business School class. Your students grilled me. And I remember that we actually chose the strategy that your students didn't choose ironically. So the case study that you wrote, and people can, I think, still find it and buy it on Harvard Business Review, but it was called Game Time Decision for AppDirect.

[00:09:56] And the quote that you chose was “I skate to where the puck is going to be. Not to where it has been.” And that's a Wayne Gretzky quote. So as Canadian founders, we love that you chose that as the headline, but you really kind of framed our starting of the business and saying, is AppDirect gonna be the marketplace where businesses go to find, buy and use software, or are we gonna be the platform that enables any technology seller out there to be able to sell off our marketplace?

[00:10:23] And that was like the biggest strategic discussion that we could decide in the early days. I mean, I feel like there's so many founders out there these days saying, how do we build the marketplace for X or for Y? And the chicken and the egg problem is real.

[00:10:41] Yeah. So what have you found are other strategies to address that initial chicken and egg problem?

[00:10:48] Andrei Hagiu: Yeah, I have lots of things to say about the chicken neck problem, but I think particularly with SaaS, there's something interesting which is illustrated by your case, right? Because people might say, okay, well yeah, of course you should go and partner because then there's guaranteed business, right?

[00:11:00] I mean, the whole point to these partners, they have reach and you immediately have a customer, they'll pay you. Like, you know, you don't have to go through this period of like, I'm not sure if this works right. But I think there's a downside, and maybe you can speak to this a little bit, and I've seen this with other companies where they basically get stuck in this mode of being a white-label provider to larger distribution businesses.

[00:11:18] And there's a very real concern. In some cases, they might actually not let you get to the point where you want to get, which is basically they might not let you become a marketplace. because they might view that as a fundamental conflict. Like they might say, okay, well what are you doing?

[00:11:33] You're trying to get control of my customers, or something like that, right? They don't want to give you too much power. It's not so trivial. It's like, oh, let's just go and find partners and that's it. Like you have to be very clear like, I'm gonna use these partners and eventually I'll get to the point where I want to get, but you know, a lot of business I think, get stuck in this wide label mode.

[00:11:49] It's very real.

[00:11:51] Dan Saks: Yeah, and one thing that I've found is that the technology choices you make in the early days really define how you scale over time. And a lot of times as investors, especially with like the public SaaS multiples, you can kind of take charts and say, okay, you're growing at 80%, so therefore you can burn X and your sales and marketing costs are Y, and your cost of acquisition is this.

[00:12:10] But one of the best ways that I found recently to describe effective, scalable businesses that have good gross margin value is like product-led growth. So we just had a few podcast episodes on product-led growth, but the notion of, instead of making your customers having to go through a lengthy year long enterprise sales process, talking to a lot of people, having a lot of barriers in mind.

[00:12:33] You can go to a website, try some of the value, easily consume it, and then scale based on consumption. And I almost think like for the platform businesses, there's probably an equivalent, which is frictionless-led or platform-led growth or something like that. You said it earlier where? As a platform, you don't want to get in your way.

[00:12:52] And if I were to be critical of ourselves, I'd say we could've grown probably in half the time if we would've got out of our own way by enabling the different parties in the ecosystem to be able to transact faster on their own without needing us to go outbound and do the partnership and then manually pair people and all that.

[00:13:14] But do you have a term for that or is this something you've seen?

[00:13:17] Andrei Hagiu: No, I think it's similar. The one thing I would say is it's interesting because in this case it's basically like your, initially, this is relates to the chicken egg problem. I think the strategy here, this is like a very valuable strategy for SaaS companies that want to become eventually platform businesses.

[00:13:31] And again, just be clear, platform business means something like you're connecting various customer types, right? You're something like a marketplace. And one of the strategies, and this is illustrated by your experience with AppDirect, is I'm gonna initially, instead of being a platform from day one, I'm gonna start with all this single modes.

[00:13:47] So I provide a product to one side. Once I get enough customers by providing that product, I'm gonna turn around and then get customers on the other side. Right? And I think that's kind of what you did, right? And I think you're right. If the vision is. I'm not gonna be just, again, a B2B SaaS problem, which is, by the way, it's very valuable, but I want to be more than that.

[00:14:08] I want to be a true platform business in a sense. Like I go B2B initially, and then I open it up to the other side consumers or something else. Then yes, I think it makes sense to go a little bit faster, right? You wanna make the product much easier to experiment, and sometimes you definitely want to use enterprise sales, like something like, you know, the Telcos or something like that.

[00:14:26] But you wanna make it very easy for individual users to adopt it, right? I mean, that's kind of the key because it matters to you to have a lot of users, and then it's easier from there to actually become a platform later on, a true, full-fledged platform marketplace later.

[00:14:40] Dan Saks: Yeah, and you spoke to this in one of your Substack articles, but what's key to a platform business is the virality or the network effects.

[00:14:48] So can you give examples of some of the best network effects that you've seen?

[00:14:53] Andrei Hagiu: This is one of my favorite top screens. I spend a lot of time, so you know, just to backtrack, right? I mean, yes. The biggest problem that any network effect company has is like chicken and egg, right? So we just said like one of the strategies we're getting past chicken and egg is like you start with the product, right?

[00:15:05] There are other strategies. I can talk through those, but you know, very briefly. So the first one is a yes. You offer a product to one or the other side, and then you move from there. The other one you could say, I'm going to, especially if it's a marketplace, I've seen this happen where you target some suppliers or sellers who already have demand and you say, Hey, bring your already existing demand onto my platform.

[00:15:27] We're just gonna make it easier for you to transact on my platform. Right? There's a bunch of interesting questions that arise there, like what's their incentive to bring their customers onto your platform? If they're transacting outside, they could say, listen, like my platform is just a more efficient way to transact with them.

[00:15:41] And don't worry, even if your customers discover other suppliers, we'll make it work in terms of the pricing. We'll give you whatever, like some portion of customer lifetime value or something like that. So anyway, there's other ways to get past the chicken and egg. Now, in terms of, this is on the tactical level to make this work, what I'm particularly interested is an angel investor, right?

[00:16:00] How defensible, at least in principle, are these network effect business. And you know, you can look around and of course, most investors and entrepreneurs want to invest in or build the next, I don't know, Airbnb or the next Facebook or the next Amazon or something like that. Unfortunately, of course not all network effects are equal.

[00:16:21] Like there's some companies that do have network effects, but they're just not as valuable or not as defensible. And this is where I have a sort of like a battery of questions that, you know, I think it's conceptually interesting. I think it's also very useful for an angel investor that you should be asking in order to figure out whether these are really promising network effects that's gonna truly give you the next YouTube or the next Facebook, or just something that's gonna be like not particularly effects.

[00:16:44] so we can get into some of that. I mean, I think part of the motivation for me writing about these is like, I've actually seen it, like I found it super frustrating. You see a lot of entrepreneurs that have in their deck somewhere, they say like, oh, we have network effects, and it's very superficial statements.

[00:16:58] Okay, we have bars and sellers. Therefore network effects and even more frustrating than that when they confuse virality with network effects. Others have written about this, but it's like, I think all of us can relate to this problem. Virality is like any product can get virality. Virality just means like, okay, well if you do a marketing campaign and your customers tell other customers about your product, it's of course it's a very useful tool.

[00:17:19] But that doesn't mean that your product all of a sudden has magically network effect. Network effect is basically like my willingness to pay us higher for this particular platform or product if there are other customers on this platform. So it's a much, much higher standard. If you hire someone who's good at marketing, they should be able to create virality Around the most boring product

[00:17:37] Like on toothpaste or anything, right? It's just like, do a very clever marketing campaign, use social media. It will go. Network partner. So if you think about like, you know, just generally, like what makes network effect stronger? I mean, there's a few things, right? So the first thing I ask is, well, first of all, is the network effect real?

[00:17:55] So it's not virality and is it meaningful? Like, is it truly the case that say buyers really value the presence of sellers? And you know, how, like how much, how many sellers do buyers need? How many sellers do they want to transact with? What's the average transaction value? Things like that, right? So you want to get a sense of like what's the size of the network effect, like what is the magnitude?

[00:18:16] Then there's all kinds of questions about, is the network effect global or is it local? So think about like Airbnb, it's a global network effect, right? After Airbnbs in say 10 countries moving into the 11th country becomes easier because they already have posts and travelers from the first 10 countries, which obviously helps when you try to get a new country.

[00:18:36] Well compare that with something like Uber. Uber, there's some positives, but I think it's fundamentally not a very defensible business because the network effect is very local. Like the fact that Uber has, say dominant market share in San Francisco does not help them in any way whatsoever when they go to a new city, say Boston, because fundamentally drivers care about riders and their geography and their locality and vice versa.

[00:19:02] So again, most business will be somewhere in between. Of course, you want a marketplace or platform with differentiated supply because that makes the network effects more valuable. If the supply set is commoditized well, it's much easier for competitors to come in and basically draw off in the same pool of non-differentiated suppliers.

[00:19:20] Dan Saks: So just to give Uber maybe a nod there, they were able, with the lack of obvious global network effects, they still were able to scale pretty effectively globally. Maybe not in China, but would you argue that's with bad unit economics and kind of just like bullying tactics?

[00:19:35] Andrei Hagiu: That's exactly the issue. So this is like another problem that I see with a lot of network effect companies.

[00:19:40] There's this myth that you really need to be first. Like, you know, because there's network effects no matter how strong they're to go as quickly as possible and dominate the world. And again, in some cases it's a terrible strategy precisely because a lot of them go faster before actually figuring out the unit economics before figuring out what exactly is the product market fit.

[00:19:59] This was very real. I mean, it was like five, six years ago, there's still investors that found like Uber's gonna be the dominant ride hailing platform in the. I always tell like why There's absolutely no reason. It's again, the network effects in the US do not help you in China or in Southeast Asia. And you look at it, they basically, they lost in China to Didi. They lost in Southeast Asia to Grab and Go-Jek.

[00:20:20] Yeah, I mean they kind of had to retrench. I guess they're pretty big in Europe and the US but everywhere they face competition, if you think about like all the big publicly traded network effect companies. Like I would not invest in Uber. I just don't think that that fundamentally is a very defensible network effect that has a lot of upside.

[00:20:39] I would consider investing in Airbnb. There's a lot more upside if you think about like there should not have a lot of competitors for Airbnb. Same thing for YouTube and it's global network. It's quite clearly like it's very global versus local plays a big role. The other part is that I think it's like differentiation on the supply side.

[00:20:54] Again, Uber is a pretty much commodity on the supply side. Not sure in the case of YouTube, not sure in the case of.

[00:21:01] Dan Saks: So you're definitely touching on some of the concepts of one of my investors, Peter Thiel, in his book Zero to One, he speaks to this, but one is last mover advantage. Like you don't need to be the first.

[00:21:09] Yes. The other is how do you become a monopoly? And I think he hated on Uber as well, potentially for similar reasons, but what you're kind of making me realize is a lot of people would think, oh, I wish I had found Uber, and it's a good company and it's a good product and maybe bad culture or bad publicity in the end.

[00:21:25] But you're making a really good point that there's a high distinction between. Being focused is a really good unit, economic business or platform, and seeing how to get the flywheel turning versus let's say like bulldogging your way and spending your way with bad unit economics into a market.

[00:21:40] Andrei Hagiu: Exactly.

[00:21:41] So I think this is why, like the questions that I'm asking, I have a whole battery and we can go through most of them, but it should be pretty intuitive. But I, I find it very useful to go through those. This is, I would say it's, even before I look at the company metrics, it's just to get a. in this market, does it actually make sense to expect if the company does everything well, is it reasonable to expect that there's gonna be a monopoly marketplace?

[00:22:05] Because that's what you want, right? I mean, monopoly or close to Monopoly, that's where we want to invest in. I guess that's Peter's point. So again, most times the network effects are not some kind of magic potion. You drink a little bit of network effect, and all of a sudden the market's monopolized. There's lots of marketplaces in lots of spaces with network effects that don't lead to monopoly.

[00:22:25] Like ride sharing is a good example and there are others, right? So that's why it's important to go through all of these, right? It's more like the potential, trying to figure out what's the potential for like very strong, defensible network effects in here. So after that, then you go into like, how is this particular company doing in terms of executing against that potential, which is, you know, a small set of other metrics and stuff like that.

[00:22:47] So let me give you a couple other things that I think are very important. So talked about the network effects. Real should be network effects and not virality. That's again, critical distinction. Supply chain that should be differentiated, global versus local. And the other thing is important. Again, I always talk about like how quickly does the network effect taper off.

[00:23:07] Network effect basically means like the more users we have, the value of the network increases. Right? Well, but sometimes it doesn't increase forever. Like even for the most valuable network effect, it's after like a million, 2 million, 10 million users, it starts to taper off and sometimes maybe even becomes negative.

[00:23:22] So you want to get a sense of, okay, well what kind of situation are we in? Do we really need a lot of variety here? Does the value of the network increase even if I keep adding past a thousand, a hundred thousand a million users, or does it taper off pretty quickly? I guess part of that has to do with whether suppliers are differentiated, for example, but you know, there are other factors that go into this.

[00:23:43] And the other thing that matters here, so this is sort of like the value of the network effect. The other thing that I feel like it's very important, is the issue of multi switching costs slash multi-homing. So switching costs, I think everyone understands because it's not just platforms. It's like any product.

[00:23:55] Well, of course I want the switching costs to be high because, well then my customers don't switch to competing products. with platforms, so switching costs matters, but it's not just switching costs. In most cases, what you worry about is like, what if my customers are present on multiple networks and not to beat up on Uber, this is another problem.

[00:24:13] This is another serious problem with their business model, which is, should have been obvious from the beginning. Well take the US right. What do you think drivers are doing? Well, they have both Uber and Lyft and they're always comparing the two, like where's the biggest potential to earn and what do we do as users?

[00:24:29] As drivers? Well, same thing. Basically we have both apps and you open both apps and you can see like which one gives you a low price and it's very easy to multi-home. It's almost like one of the worst cases in terms of Multi-home because it's so easy and I could just compare them in real time. In other cases, obviously it's more difficult, and I think this is one where like you also want to talk to the entrepreneurs.

[00:24:49] Like there are things you can do to raise the switching costs and the multihoming cost. Give people an incentive to mostly transact on your platform and not to sort of divide the transactions between you and the competing platform. But I mean, you can see how that's a very significant issue. If you think about, again, you can have network effects, but if everybody multi homes, that network effect is, I'm sorry, was not gonna give you a very defensible position, right?

[00:25:12] So you have to think pretty carefully. Like is there a way to give people an incentive to be exclusive to your platform? And yeah, there are ways to do that and that's something for instance, in talking to entrepreneurs, like I expect them to have a good answers for this. There's layers to this, like the first thing that everybody says, oh, you just give them financial incentives, right?

[00:25:28] They just reduce your take rate. Take Uber, they go from 20%. They say if you do more transactions on Uber, if you do more rights on Uber, it'll go to 10% than to 5%. That's a terrible answer because I think about Uber, they're already not making money. They're losing money. They'll be like, their the solution made even less money?

[00:25:46] Like that doesn't work. Right. So in general, what I think is important is to think about providing incentives to one or both sides to be exclusive, but it doesn't require paying them to do so. Because of course you can just pay them, right? I mean, I can say like, I'll give you. I don't know.

[00:26:03] I give you $500 if you're exclusive. That's not a real solution, right? So you wanna do something that at least, ideally it doesn't cost you that much. So something better than just like loyalty programs and discounts would be nice. So, for example, you can reward people algorithmically. So in the case of Uber, there's an example I do with my students in class.

[00:26:23] You can come up with a solution for the Uber case, which is very difficult. But one idea is basically to say like a driver that completes, more rides on alert would be shown at the top of the queue, right? It would be given preferential access to say, more valuable rides or something like that. And same thing for the riders.

[00:26:40] Usually I spend quite a bit of time when I talk to startups like this obviously is an important point and yeah, that determines whether or not it's the defensible one or whether it will end up like in a very competitive situation.

[00:26:52] Dan Saks: Yeah, it's a really helpful framework in order to assess what are some of the problems down the line.

[00:26:57] As an entrepreneur, I would say we faced each of those problems and every time you were giving an example, I was like, okay, here's where we went wrong. Here's some critique. It triggers a couple things to me. So one is, as a multi-sided platform, you always have this issue of like, who's your primary customer?

[00:27:11] And I'm sure a lot of the decks you see would have these slides similar to what we'd show, where you have a flywheel effect. And we have different partners and it's hard to pick one. But sometimes you just have to say, this is my primary customer and that's gonna make my business scale. Do you think, a, do you agree with that premise of having a primary customer and B, do you think it matters what side that primary customer is, or is it just the fact that you focus on one to ignite and then you kind of go in a sequence?

[00:27:38] Andrei Hagiu: No, I think it's a very fundamental question, both for platforms, specifically multi-sided platforms, when you have more than one side, right? So like marketplace, you have two sides, buyers and sellers. Social network. I used to think that there's users advertised, there's potentially third party app developers, and I mean, I think in principle we say, well, they're all our customers.

[00:27:57] We care about all of them. But the difficult thing is, I totally agree that you do need to have a very clear view. There should be one that's the priority because at some point, push will come to shove. Let's say, you know, maybe like there's conflicting incentives of course, if there are features that you want to add that help all the various customers to benefit from, like there's no problem, of course.

[00:28:19] Like there's no issue. The issue becomes, there'll be a feature or one type of customer benefits the other one doesn't particularly like, right? So then you have to think about like, how do I resolve conflicts of interest between the multiple sides? And of course, You should have an answer, like, you should always resolve conflict of interest in favor of the most useful side.

[00:28:37] So how do you choose that? An easy one. Say the, any kind of advertising supporting platform, like social media, right? It's like there's advertising users. Well, should you care more about advertisers? Should you care more about users? I'd argue you should care about users because without users there's no advertising business.

[00:28:55] Right. Versus like users don't particularly like advertising. So there's a bit of an asymmetry. Now, of course, even there, I would say it's complicated. As we know from the Facebook experience, it's complicated because users create the network effect clearly. So if you focus on them, they don't generate revenue, right?

[00:29:13] So it's advertising that creates revenue. I think realistically, yes, it's important to have one side that matters more.

[00:29:21] Dan Saks: I remember with the Zappos example with Tony Shay, they said that they didn't really take off until he actually would buy the inventory. So he'd bulk buy, this was old school, but bulk buy the shoes from Nike to be able to get the right inventory to then say, Hey, we have enough shoes that someone's gonna take us seriously.

[00:29:37] Are there other strategies in the digital world that drive that kind of demand of one side?

[00:29:43] Andrei Hagiu: This is good that we start with the definition at the beginning. So Zappos, it's not a platform. What's interesting about Zappos is initially very early stage, they wanted to be the eBay for shoes. And then very quickly Tony realized like, I can't be a platform.

[00:29:58] Like I can't be a marketplace because I can't guarantee quality of service. Within a year they moved to a retailer. So if you look at Zappos today, it does not meet the definition of that of a platform business because they buy and resell shoes. Right? And it's a retailer. There's no direct interaction between Nike and the customers like Zappos buy this stuff.

[00:30:18] When you go to Zappos, you buy from Zappos. Like they ensure everything. They price the shoe, they do delivery and things like that. So I think those are a little bit different, right? I mean, they try to do marketplace initially, but in this case it's quite clear like their customer is basically the end consumer.

[00:30:32] And by the way, that's why they have one of the best customer experiences online. There's no ambivalence. There's no like, well, what We have buyers and sellers. Our customer is the user. Of course, like we have to buy stuff from suppliers, but there's only one customer group, which is the end user. In other cases, it is more complicated.

[00:30:51] Like if you were to say it's something like eBay, then over time maybe your allegiances have to shift a little bit.

[00:30:58] Dan Saks: I think I'd clearly get an F in your class. I've been mixing all the different terms.

[00:31:02] Andrei Hagiu: No it’s fine. I mean, so I thought a lot about this. Obviously I tried to make them very internally consistent. I think there's value in that because this is a good illustration, right?

[00:31:10] It's very tempting to start talking about platforms and then you get into like Zappos and Netflix and things like that, which are just fundamentally not platform businesses.

[00:31:18] Dan Saks: So in that vein, with all the research you've seen on platform companies, is there a right or wrong or is it just purely execution?

[00:31:26] You could have an idea, it's the right idea, but you just have to navigate this wild rollercoaster. And the ones that make it are like the grittiest entrepreneurs that can figure it out. Or do you think that from the research you've done, there actually are academic answers that can propel a business.

[00:31:41] Andrei Hagiu: So the answer is yes.

[00:31:42] I do think that there's some right or wrong answers. So what we're just talking just now, right? I think it's a good illustration of this point. So the questions that I talked about, let's say, how valuable are the network effects? Are these strong? Are they defensible? Those questions don't apply, in such that they're independent of execution.

[00:32:01] This business saying, okay, I see what you're trying to build, and I'm just gonna assume that you're basically executing perfectly. And the first thing I'm gonna try to figure out is like assuming perfect execution, is this an idea or is this a concept in this particular market that can lead to a very defensible, dominant marketplace?

[00:32:20] So that's part number one. And I think, yeah, it's very important to do that because you can have someone executing amazingly well in a terrible industry. And you know, I think there's a quote that's like great CEO, great entrepreneur meets terrible business. One of the two reputation stays in tech, and it's not that of the entrepreneur.

[00:32:38] You can be amazing. But if you're in terrible business, fundamentally, because again, network effects are not at large or not sustainable, it's just not gonna work out. But then once we've done that, so we've convinced ourselves that at least in principle, this is the situation in which network effects can't actually lead to defensible businesses, then I agree.

[00:32:53] Then it's all about execution. And there I would say there's definitely like good and bad strategies, right? I mean, there's good and bad approaches to solve the chicken and egg problem. There's good and bad approaches to scaling. Like we talked, you know, like a lot of network companies move to scale way too quickly.

[00:33:09] They first need to figure out things like that. There's the stuff we didn't discuss, you know, a big one is leakage. The idea is like my marketplace helps the buyer and seller meet, but then they take the transactions off the marketplace. I'm sure you've faced some version of this, and it reminded me as you're talking initially, right, with AppDirect, so you're reducing search costs.

[00:33:27] You make it easy for people to search and find, but you don't stop there, right? I mean, your point is like, I also make it easier for you to transact with them, which in itself is an insurance against leakage, right? If all you're doing is making it easier for buyer and seller to find each other, why would they come back to you?

[00:33:43] Right. Once they found each. There's a lot of things. So once it comes to execution, there's a lot of ways for it to go wrong. Basically. There's a lot of things that you can do wrong. But yeah, I would say there it's basically, it's execution and it's kind of stuff, especially if you're looking at early stage startups in a lot of ways, you really have to form some judgments about the capability of the entrepreneur to like experiment and figure things out.

[00:34:05] Like quickly learn what works and what doesn’t.

[00:34:09] Dan Saks: So thinking of the next frontier of marketplaces and well specifically platform businesses, we've talked a lot about centralized marketplaces or marketplaces with multi-sided platforms, but there's a whole movement around Web 3 and decentralized platforms.

[00:34:23] What's your take?

[00:34:25] Andrei Hagiu: How many hours do we have? How many hours do we have? Very briefly, obviously, it's super interesting, which also super exciting. I would say. This is where like there are a lot of other people that have great insights into this. From a16z, it has a lot of interesting podcasts and writings.

[00:34:39] To me, what's super interesting is obviously the idea that Web 3, at least the principles, allows people to build decentralized or semi decentralized marketplaces. Now, as with anything, we have the maximalists. We immediately rush to the conclusion that it's the end of centralized marketplaces and you know, basically Airbnb and Uber are obsolete and Facebook are completely obsolete.

[00:35:02] You know, tomorrow there's gonna be decentralized equivalence. We're going to give the power to the people, and all the centralized ones will go bust. The reality, of course, is more complicated. I don't think it's gonna happen, but certainly I think the Web 3 marketplaces, fundamentally that's part of their selling point and their power, is that in some sense they will return some of the value to the users.

[00:35:24] And there's almost like a credible commitment through decentralization, essentially, not screw over users once the platform becomes too powerful. Right. So one of the problems with Web 2 platforms that become very dominant is that once they're dominant, they increase fees or they find ways to kind of change the terms on their users in order to essentially like extract more value.

[00:35:48] And that's a problem with Web 3 that provides ways for marketplace to commit not to do that, which in principle should be more appealing to users, to buyers and sellers. So you can probably build more, say like equitable, more sustainable marketplace in terms of sharing the value of users. Now, I think what's interesting is that the devil's gonna be in the details.

[00:36:09] I certainly do not believe that truly decentralized marketplaces have any chance. I mean, it's almost like saying, do you think you can run a company in a decentralized way? Something where like all the users can discuss and vote in a discord group on all the decisions. I mean, that's nonsense, right?

[00:36:25] There's a good reason to have like priorities and authority, like, you know, centralized authority for some critical decisions, especially early on. But it is definitely possible to have interesting models where say you centralize some decisions, you function very much like the traditional marketplaces for some things, but for other things that matter a lot to users, you can allow them to basically vote upon, they can sort of become a little bit more decentralized, get their input and so on.

[00:36:49] Dan Saks: For sure. So I’ll link to your Substack in the show notes. Are there any other resources you'd recommend for the viewers?

[00:36:57] Andrei Hagiu: I think my Substack. All my contact info is in there. I put a lot of my research and all my thinking around marketplaces in there. So yeah, that's a very good resource.

[00:37:07] Dan Saks: Nice. I could keep going for hours, but Andrei, thank you so much.

[00:37:10] So good to catch up.

[00:37:11] Andrei Hagiu: Likewise. Thank you so much. Yeah.

[00:37:18] Dan Saks: Thanks for listening to Decoding Digital. Make sure you never miss an episode by subscribing to the show in your favorite podcast player. To learn more, visit decoding digital.com. Until next time.