EM Group Chat #142: A Deep Dive Into Network Effects With Sameer Singh From Speedinvest

So welcome back to Everything Marketplaces, where we talk with founders and leaders from some of today's top marketplaces. So this is episode 142, which is a really great group chat that we just had with Sameer Singh, who's a network effects advisor and investor. So Sameer's the creator of breadcrumb.vc, where he shares content and resources on network effects, and is also a venture partner at Speed Invest, where he invests in early-stage marketplaces. So this is really a great chat with Sameer, where we got to learn more about what led him to network effects and marketplaces. Did a deep dive into network effects, or sort of Network Effects 101, discussed more on how Samir evaluates marketplaces when investing in them, and also had a great group Q&A. So I really enjoyed this conversation, and you're going to find it a great watch to the end. So Samir, welcome to the group chat. Now, you know, it's of course great to have you join us here today. So I want to start off by saying, you know, huge thanks for taking the time to do so in advance. So I'm looking forward to diving into network effects and more on the investing side with you here. But before we do, I thought it might be great if you can start off by sharing a little bit more on your background though, for those that might not know you.

So quick bit about me. Originally, I was a part of a family office in India that was trying to build India's rocket internet. And while I was doing that, this was around the time the iPhone and Android came out. So I started writing a blog on the operating system wars and mobile application businesses. That led me to a startup called App Annie. So I ended up moving to London with them, worked there for 5 years. At one point, pretty much every consumer tech company, at least that was building a mobile app, was an App Annie customer. And so I ended up learning a lot about network effects there. When I left, I kind of wanted to combine those 2 things that I really enjoyed, network effects, which I kind of dabbled in when I was writing the blog about operating systems as well. and investing, which is kind of what I was doing before that. I didn't really know how, so I spent some time just writing what I knew about network effects. The pandemic gave me a year to do that. That got me into the Atomical Angel program where I basically started implementing everything I had written that became my investment thesis. And everything I wrote also turned into a course on Maven. And finally, that investing journey also led me to Speed Invest where I'm now a venture partner. So I spend 2 days a week with Speed Invest. spend the rest of my time angel investing and predictably only invest in companies with network effects.

Yeah, so that's a really great background. Thanks for sharing with us. So there's definitely a lot we're going to get into, but I guess going back to the very beginning, what initially led you to network effects and marketplaces and why did you ultimately decide to focus on them?

It was a very accidental journey, right? So I always had a bit of a knack for unpacking products and businesses and trying to think about them a level deeper, which is what I was trying to do in the original blog as well as right from VC. And I coming across this concept called network effects that seemed to be behind most of the companies I found interesting. And that there was a lot going on there outside of just quote unquote EBITDA margins, which is what a lot of analysts like to focus on. And so that basically led me down the rabbit hole. And so accidentally because of my journey at Apani, ended up coming across a lot of these companies and just kept going deeper and deeper. In terms of The way I think of network effects is, I mean, they are a mathematical concept, right? They've been around as long as the universe has been around and for us, as long as humans have been around. Languages have network effects, civilizations have network effects. And so if you've got something that enduring, we've seen that it has a pretty massive impact on the companies that actually have them. The study by NFX that shows that 70% of all value created in tech is from companies with network effects. And you really see the power of them when you look at them up close and center. Of course, not all network effects are that powerful. Some can be weak. And so there's a lot of nuance in what's a strong, defensible, scalable network effect, what isn't. But fundamentally, that concept I found very, very intriguing just because of the potential impact it's had.

Certainly. Yeah. So I guess before we get into some of the more specific topics, I thought it might be great if you could maybe kind of give us an overview or kind of a Maybe like network effects 101.

The definition I think everyone knows, right? The addition of a user makes the product more valuable for all users. So essentially the utility of your product is tied into adoption, which is both good and bad. Good because when you get that going, that can be very, very powerful. Bad because on day zero, you don't really have anything. You need to find a way to kickstart your product. There's lots of different ways to categorize network effects, like NFX has their own. I have my own. The way I divide them is into 4 very broad buckets. Data networks where you're collecting data from users to increase the value of the product for the same user. So, Waze or TripAdvisor would be in that bucket. An interaction network where you're connecting users to enable ongoing information exchange between them. There you could have social networks, collaboration, P2P payments, stuff like that. There's marketplaces, which we all know you're connecting buyers and sellers to enable transactions. Airbnb, Uber, stuff like that. And there's what I define platform, which is not sort of the bastardized term we use in the tech world. It's when you're allowing application developers to build applications on top of your product and then connect them to the users of your product. That's an actual platform. And then you can have some sort of SaaS overlay between those 2 middle 2 categories. And you kind of think of this as a spectrum going from data network all the way to platform, where adjacent categories tend to have Some overlaps. And then I have subcategories within those depending on certain characteristics that make those types of network effects more defensible, more scalable or less. That's how I end up sort of categorizing them. So they tend to be, the framework tends to be a bit more flexible and a bit more applicable compared to a lot of other frameworks that you see out there, which sort of look good in theory, but can be very, very hard to Apply in sort of complex, messy reality.

Oh, certainly. That's a, that's a really helpful breakdown for us. So I guess, you know, maybe using some examples of just marketplaces that we all know of, right? So if we use like an Airbnb and Uber, could you maybe describe kind of like the network effects in play at both of those marketplaces and, you know, use that as a way to kind of point out how they might be different?

The basic construct of a marketplace and a network effect, both of them meet, right? So adding more drivers makes Uber more valuable for riders, vice versa. Adding more hosts makes Airbnb more valuable for guests and vice versa. People also conflate the two sometimes. But if you actually look at the dynamics within each marketplace, Uber supply is very, very commoditized. So when I'm trying to book a ride, I don't care who the driver is. I just want him to get to me in 4 to 5 minutes. And below 4 to 5 minutes, rider behavior tends to not change. And so as a competitor, all you need to do is hit 4 to 5 minutes, a wait time of 4 to 5 minutes relative to the size and location of your customer. Right. That's— it's challenging, but it's not that hard to do if you're willing to burn some money, which we saw happen in the case of Uber. Airbnb, on the other hand, the supply is highly differentiated. Every property is different by location, by price range, by quality, by the individual attributes it has. You know, some might have a fireplace, some might have parking, some might have high-speed broadband, the number of beds. differs. And as a consumer, you care about each, if not each, many of those things. And so as a marketplace, Airbnb needs to have critical mass in each of those, you might say, category combinations, apartments within a certain price range in Paris that have 3 or 4 or 5 of those constraints depending on what their users are searching for. That's a huge amount of variability, really hard to solve the cold start problem, but also massively defensible once you get to scale. It is very, very hard to create a comparable experience, which is why even now Airbnb doesn't really have any direct competitors. Like it's never had any direct competitors outside of like Wimdu that tried to subsidize their way into getting a couple of apartments, could never match up to the experience and then flamed out. So that's one axis. The second axis is scalability. How easy is it for you to leverage the existing scale of the marketplace to expand into new locations? With Uber, insanely hard because it's a very hyperlocal marketplace. You need both demand and supply in the same location. So you bootstrap in one location, then you go into a new city, a new location, and start off from scratch all over again. So your CAC starts out really high because you're subsidizing everyone. And as you start or successful in getting more people on, word of mouth spreads, CAC starts to decline. You move to a new location, goes right back up, right? So it's It's a challenging market piece from a unit economic standpoint. Airbnb, once you have a host in New York who's attracting guests from everywhere, the next time those guys might want to go to Paris or Prague or London. And so you can leverage that demand to attract hosts in those regions as well. So a whole lot easier to scale. And you can see this in the unit economics of both Uber and Airbnb. Airbnb had a much easier time getting to profitability. In their case, it was actually something they could control. In Uber's case, it took a while. It took a CEO change. That was a challenging journey.

So those are 2 really great examples of marketplaces, of course, at scale. But we have a lot of early-stage founders here in the community. So how should founders be thinking about network effects at the earliest stages? And maybe how can they start to see them kind of take shape or form and maybe even start to measure them?

Well, the first thing you got to do as an early-stage founder is recognize that products and companies don't have network effects, interactions have network effects. So figure out what interaction it is that you're actually trying to create, which are the 2 users you're trying to connect, how far apart do they need to be, what's the nature of that interaction. And that really will define the structure of your network effects. First, what category it falls into and also if it's more scalable, more defensible or not. I have a quiz on breadcrumb.vc. It's about 10 questions. If you go through it, you'll get a quick overview of kind of what that looks like. And so that's one bit of it. In terms of when you know you have network effects, like once you have liquidity, in theory, you have network effects. They might not be massively defensible at that scale just yet because you're a seed startup, but they do exist and they are measurable at that stage. If you define a network effect as Adding a user makes the product more valuable for all users. So what metrics there are for that, again, there's a quiz for this as well, but you can broadly categorize marketplaces into sort of 2 types. There's high-frequency marketplaces, low-frequency marketplaces. If it's a low-frequency marketplace, there's 2 things that really tell you if you have enough liquidity or not, if your network effects started to kick in. There's a metric called the search to fill, which is of all users that end up on a particular landing page or marketplace, what percentage end up making a booking or transaction? And people ask me, great, what's the benchmark for that? The answer is there is none. You can't have a benchmark for that because what a search to fill is for an Uber or an Airbnb will differ widely. It's very, very specific to the kind of marketplace you have. Instead, you need to look at the trajectory. this thing needs to follow an S-curve. Once the S-curve starts, that curve starts pointing up, you know you have the beginnings of liquidity. Once it starts to tap out, you know that kind of adding more supply or demand is not really going to meaningfully move the needle, at least in that category of supply or in that particular region, depending on the kind of marketplace you are. So that's one metric. The second metric is the utilization rate. In particular, if you have what I call high-touch supply or limited amount of supply, right? So Airbnb and Uber are good examples of that. You would basically measure what percentage of available inventory or SKUs have been booked in a given time period. And again, see if that follows an S-curve. Once it starts going up, you know you've got the beginnings of liquidity. And when it starts curving up, you can say you have network effects at that point. Because adding more supply and demand is genuinely changing user behavior. If you have what I call low-touch supply where the amount of supply is technically unlimited, like Udemy for example, right? Once a course is on, the number of students that can take it is infinite. There you don't really have a good measure for the denominator there, right? Or how much inventory you have available. You could either measure what percentage of your courses have received a booking. That's one thing you could do, or you could measure share of wallet. What percentage of that instructor's earnings or revenue is coming from you? That can be tricky to measure because you have to figure out third-party data in addition to your own. So you can do that via surveys, but that's not something that's going to give you sort of a reliable S-curve. So that's one of those things just to keep in mind. And then you need to look at these 2 in combination with your supply density, which is basically supply to demand ratio or the number of listings for each category, whichever one sort of makes sense in your case. So as that changes, that supply to demand ratio, you'll see variations in the utilization rate in the search to fill. And so you'll know what's the optimal amount, because if you have sort of too much demand, then something's going to break. Or to one supplier, something's going to break. You can also measure percentage of buyers who become suppliers and vice versa if you have that kind of marketplace where users play both roles. Now that's for low frequency. High frequency, all of that still applies except for one more thing, which is purchase frequency or what I call core interaction frequency, which again should follow some sort of an S-curve. How many transactions do people make? The last thing you need to keep in mind is that what time period do you actually need to look at this data for? Is it weekly, monthly, quarterly? That's very specific to the marketplace you have. So for example, if Airbnb looked at purchase frequency per month, that would give you nothing. Or if you look at retention per month, that would give you nothing. They look at annual retention cohorts. That's your purchase frequency on Airbnb. Whereas for Deliveroo and Uber, monthly frequency is good enough because that's the frequency we even might even be able to look at it weekly, but that's probably overkill for most marketplaces. So you need to align sort of the time granularity with the frequency of use for your product.

Those are definitely super helpful metrics here. And I'm glad you mentioned the point as far as spectrum of frequency, right? So I know there's a lot of B2B marketplace founders here in the community. And one of the questions that was actually added to the doc is, With the kind of, I would say, lower interaction and kind of lower frequency of purchases, does it typically take longer for network effects to start to form with B2B marketplaces or?

To answer the question, yes, but not because of the frequency. So you see low frequency marketplaces in the consumer space as well, right? So you can still, though the path to liquidity is sort of different, depends on how differentiated the supply is, how hyperlocal it is. If you have both of those, it's very, very hard to build liquidity. The bigger issue with B2B marketplaces is not frequency, it's the distribution channel. So by and large, for most consumer marketplaces, consumers you can acquire via a scalable channel. Now you can debate whether paid advertising is profitable or not, and the unit economics there might stop, but it is scalable. B2B marketplaces, especially the ones that are challenged, tend to have this issue of trying to do direct sales on both sides. That's obviously a very long, slow process. So, you know, you get one buyer on board, he wants to buy something, you go, all right, I think that seller has something, you get that seller on board, you put them together, it's not really a fit, then you find another seller. And meanwhile, you have the existing seller and you need to find a buyer for him. And so it's really hard to transition from this brokerage to an actual scalable marketplace. So if you really want to scale that B2B marketplace, you need one side to have scalable acquisition channels. Like there's occasional situations where you might be able to do SEO or paid advertising for that side, like Alibaba for some, as an example. But in the vast majority of cases, that tends to not be the most effective way. The most effective way is buyer-to-seller virality where either you focus direct sales efforts on the buy side or sell side and they onboard the other side because you give them a tool or some sort of financial services like invoice financing that incentivizes them to bring onboard everybody they transact with. So that's a very, very good hack like Fair, probably one of the most successful B2B marketplaces, like they have that at the core of their marketplace. The second is situations where every customer plays both the buyer and seller role. And in that case, you're just doing direct sales to customers and then populating your network or your marketplace on the back of that. You don't have to worry about which one's a buyer, which one's a seller, they will end up doing both. So those 2 I tend to focus on quite a bit. Those tend to be very structural. So if you don't have that and you're forced to do direct sales for both sides, then you're probably not in the right space. it might be better to think of another area or maybe potentially experiment with other GTM channels to figure it out.

That's definitely super helpful. So I'll have to hold myself or refrain from going too deep into it with you here. So I know we'll have some questions from founders about that when we get to the Q&A. But you mentioned that of course you're investing in marketplaces and network effects businesses through Atomico and Speed Invest. So could you maybe share a little bit more about, you know, maybe, you know, as far as like the check size kind of stage that you're investing in, and then maybe even a few examples of recent marketplaces?

Sure. So in terms of check size, it's roughly $30K for my angel check and between $500K to $1.5 million for the speed invest check. Both are at pre-seed and seed. Speed invest invests right from like idea stage to Proper seed rounds, which are typically up to a round size of $3 million. Personally, I don't go that early. I like to see an early MVP and early behavioral metrics to sort of validate the marketplace. So some of the metrics we've talked about, search field utilization for very high-frequency consumer marketplaces, it could be user retention as well. So those are the kind of things I'm usually trying to validate. My investment approach is there's a few things I'm looking for. One is, is the founder solving a problem that they have some sort of personal relationship with? That really matters to me at a personal level. Second is obviously every network effect marketplace or not is built on an interaction. Are they creating a unique interaction, something that these users could not do before and that solves a real pain point? Third being the metrics. that we actually talked about in the fourth actually fits the criteria. I'm actually not a big believer in the market sizing religion because I have never seen an investor who has accurately estimated a market at seed stage. It has never happened. And in fact, most successful investors will tell you this. I was recently in a talk and a partner at Accel Basically said she went back and read their Facebook memo and they were off by orders of magnitude. And that's happened in every deal that I've ever seen. So it's essentially astrology for VC. So that one I do not care about at all. In terms of examples, there's actually 2 companies from the community that are part of my angel portfolio, not the Speed Invest portfolio. One is a company called Nium, which is, you might say, a A unique spin on fashion resale. The idea there was that most people go and try and resell clothes on a Vinted or a Depop or a Poshmark. But generally speaking, the things that get sold are ones from mid-range to high-end brands, whereas the average thing you might buy from a Zara or H&M just doesn't get sold because people don't want to pay up for it. And those things end up in a thrift shop or in a landfill, and that tends to be the vast majority of clothes. The idea on New is you can upload one of those things, New approves it and gives you a virtual coin and you can use that to claim somebody else's clothes and you pay a £1 or $1 fee every time you claim something else. So essentially it's recirculating clothes and the level of site switching here is incredibly high. Last time I checked, I think it was 4 to 5 times a month, which was extremely high for a marketplace. The level of liquidity they had relative to the number of users they had was incredibly high as well. The second is a company called Homefans, who I actually found on the community. Homefans is a marketplace for sports experiences. The best way to describe it is it is a guided tour of football match to start with. So they started out in Latin America where often there's a You might say that there's a bit of trepidation to go watch a football match there because it's a bit rowdy. And so the local host, who's typically a season ticket holder, gives you the authentic local experience, takes you to a sports bar where local sports fans go, gives them a tour of the stadium, watches the game with them, tells them about the team, about individual players. And so it's a guided experience through So the dynamics of that marketplace are very Airbnb-like. So supply is very, very differentiated. This interaction obviously was very unique. It's not something that existed out there. And they also have the Airbnb advantage of having a positive cash conversion cycle because people often buy things ahead of time. And so Home Funds is already cash flow positive and they haven't raised a seed round yet.

Those are 2 really good examples. Of course, I'm biased here myself with them since in the community. So also too, I cannot overlook the line as far as astrology for VCs, right? Yes. So I actually had a question for you and that's, you know, as far as, you know, what are the primary kind of differences when you're evaluating early stage marketplaces when it comes to kind of, you know, from an angel kind of point of view versus say Speed Invest? You know, from a fund?

Now, the angel point of view is simpler, right? I'm the investment committee. So my process generally is first call with the founder, digging into the metrics, second call to go over questions about the metrics, and then yes or no after that. There's like a bit of documentation work and IAM I need to create for Atomical, but by and large, that's pretty much it. With Speedinvest, beyond that phase, and usually that phase, you know, I have an associate from the team involved and there's sort of a A broader list of questions. And then there's a pitch to the entire team, what we call a workshop. It's about a 1.5 to 2-hour session with the founders, which is kind of a longer pitch plus a rapid-fire Q&A from the entire team. And we come to a decision on the term sheet after that. So it's not that much longer, but because it's essentially institutional investing, there needs to be a stamp of approval from the team IC.

Yeah, no, certainly. I thought it'd be a helpful context for, you know, founders in the community to just kind of see, you know, see the differences from an angel versus say a partner at a fund. So, and then are you guys, you know, investing primarily in UK-based startups or?

Across Europe, across Europe, including the UK. Personally, as an angel, I can only invest in Europe. Speedinvest can do a percentage of their fund in emerging markets as well. So typically, maybe anything anywhere between 5 and 10% of the fund might end up going to Southeast Asia, Middle East, Africa, maybe even Latin America, but the US is off limits. Got it, got it.

So yeah, we don't get too many European or UK investors here as far as on the group chat, so wanted to ask about that. But yeah, so you and I could probably go on for quite some time here with things, but you know, you of course have some awesome experience, you know, not only insights within network effects, but also on the investing side, and I'm sure you see a lot of deals. So are there any common misconceptions maybe with early-stage founders that you're seeing specifically around network effects and maybe some of the related kind of topics, you know, as far as like defensibility when it comes to pitching?

I mean, there's a few, there's 2 very common ones. One is confusing virality with network effects, which is, you know, the idea of virality, in particular product virality, is that You spread— users spread the word about the product in process of using it. So Zoom, for example, has virality, right? You want to use Zoom, you've got to share a link with the other person who wants to join the call. They don't need to have a Zoom account. And because they don't need to have a Zoom account, adding more Zoom account holders does not increase the value of the product. The value of the product is the same. And so Zoom does not have any defensibility from network effects. And so as of today, there's quite a few Zoom alternatives. And Hopin was kind of one of those companies that had a similar experience, right? It was touted as a network effect product, it really was not, it was purely viral events product and we know kind of what happened with that. The second one is that anything that looks like commerce is somewhat equated with marketplaces, right? There's a rule, the idea of a marketplace that enables an interaction between a buyer and a seller. If the company is the seller, then you are not a marketplace, you are an e-commerce company. There's not a dramatic difference between the business model of a Lime or a Bird and a Herds. They're like asset rental companies. The same thing for all these 10-minute grocery delivery companies. There's not a huge business model difference between a Gorillas and a Tesco outside of the fact that Gorillas just decided they wanted to burn more money. So fundamentally these things look like marketplaces but they're not. And the thing about defensibility of network effects even when you do, there's a spectrum there. So that I can— you can sort of excuse lots of founders and investors at the level of nuance that most have not been exposed to. That's okay but these 2 are the ones that really, really matter because it's pretty binary if you have a network effect or not in those cases. And if you invest in a company or raise funds in a company that you think has network effect, but actually doesn't, that's a pretty bad thing. That's not going to end well.

Certainly, that's going to be really helpful for our founders here. Noah, do you want to come on?

So I know we've talked about this a bit, but making your product into a workflow so that your actual client or the user uses it in their workflow.

So my question is regarding to that.

So what are the kind of key elements you have to keep in mind when building the product and in like, so that the companies actually use it as into their, like put it into their workflow? Because like we wanna make it 10 times faster for companies to work with influencers, but like how do you keep that in mind? Do you have any key, like what is your mindset going into that? What would you say to that? It's a tough question to answer because it's deeply specific to each workflow. The whole idea of embedding yourself in the workflow is that you want to create switching costs, right? So if the marketplace doesn't have a strong basis of defensibility, if the SaaS adds a meaningful switching cost, that gives you some stickiness, overcomes the weaker defensibility of the marketplace. So the way you want to create it is as you use it, there's more and more data, information, whatever that gets embedded into the system, which makes it stickier and stickier. Whereas, so even if you hate the product, you can't leave later. A basic example being Salesforce, right? Salesforce probably has one of the worst UXs I have ever seen in most software products. I think most people who use Salesforce will agree, it is horrendous, it's horrible. And the same goes for most ERP systems like SAP. But when you start using it, you're basically Any information you need to drive your business gets stored in there. And trying to rip all that out and moving to something else is going to cause a major disruption to your business. So it's not just the financial bit of it, it's the non-financial bit of it that really makes it hard for you to switch. That's really the gold standard there.

Hey, Sameer, so actually, so one that was added to the doc was around as far as thinking about like, you know, expanding from your first kind of few markets. So say you have, you know, initial kind of liquidity, in this first kind of, you know, 1 to 2 markets. Are there any kind of metrics that you look at that, you know, shows that, hey, you know, you do have liquidity and it is kind of, you know, if and when, you know, to expand or?

It depends on the kind of marketplace you are, right? So if you're a hyperlocal marketplace, then yeah, you basically want to make sure you hit the top of that S-curve on search-to-fill utilization or whichever metrics are relevant to you. So you know that, okay, this market is now stable. And then you can move on to another marketplace because— sorry, location, because that's going to require your full attention to rescale that location. If it's something like an Airbnb, truly cross-border marketplace, then it doesn't matter as much. You might still be building local supply in some cases, like HomeFans and Airbnb, you had to build supply in each destination. So you're going location by location. But the moment you kind of have critical mass and kind of the S-curve is starting to point up, you can look to acquire locations in new regions because that's going to help the S-curve move up further. And whereas if you're a purely cross-border marketplace like Udemy where you don't have that issue at all, then frankly it does not matter where your supply is based. In that case, it's better to optimize for a particular category you're building. courses for marketplaces, for example. You wanna make sure you have a strong S-curve within that category before you move.

Yeah, those are 2 great examples. Appreciate it. So, hey, Dimitri, do you wanna jump on? Question?

I have a question about, I guess, flywheels is probably the broadest way to describe it. So in addition to the cross-side network effects that you've talked about, there are also other things like returns to scale, to scope. There might be network effects that are based on the data that a marketplace has available. Are those things that you sort of consider in picking out which marketplaces are going to be prominent or are those sort of second order effects that are kind of nice to haves? To me, very much the second. Data network effects in particular are a category of network effects I'm deeply skeptical about. The vast majority of data network effects tend to be very, very weak. It's only the rare cases where one that data is central to the product, like a Waze, for example, and the data tends to decay very quickly. That's when it tends to be defensible, it tends to be— scalability is sort of a bit of a mixed bag there. Even something like a TripAdvisor where you kind of start out with reviews, then you pivot into being a marketplace, right? Whereas if you're building reviews for a marketplace, like that's not a data network effect, that's curating that particular marketplace. Economies of scale for a managed marketplace could be interesting, right? So an Amazon, for example, you're building out warehouses, logistics, fulfillment centers because that improves the experience of being in the marketplace. So yeah, that can be a factor. But if it's purely inventory carrying and your argument is economies of scale because we can buy inventory more cheaply like that, I'm more skeptical about that. To me, it's not even a tech company, forget a marketplace. Like, I mean, that's retail. So that stuff I don't look at very seriously. So if it's a managed marketplace, if you are building economies of scale, then you also have to ensure that the kind of unit economics you have are commensurate with the investment required in a managed marketplace, right? So for example, the Vestiaire Collective, they kind of have this verification system where you verify every single item before it goes on to a buyer because you're selling luxury products like a Burberry handbag or whatever. But the AOVs there are large, the take rate there is large, so it makes sense. So I've had recent pitches where a founder is trying to create that but for regularly priced fashion e-commerce, like the economics don't make sense there. If the AOV is $25, like best of luck verifying that, that's not going to work. So yeah, that's kind of how I look at it.

Awesome. Hey, Jason, do you want to come on?

Yeah, thank you for that. Sameer, thank you so much for joining the chat with us. The question I had was, if you have a marketplace business and maybe you're looking to move to a different region, so going from the US to Europe or maybe Europe to APAC, what are things that marketplaces during expansion often get wrong when they're tackling new regions? So again, depends on the kind of marketplace you are. So if you're the kind of marketplace where The supply and demand both have to be localized to the same sort of continent, right? So something like, let's say I have a portfolio company called Home Cooks that delivers frozen food, right? There's no way in hell that food's going from the UK to the US. Like they can do it within the UK, they can do it within France, maybe even like France to Belgium, but it's not, it's sort of, there's a border to it. So in that case, you've got to be very careful about when, because you need to ensure that the playbook is perfected. Before you move on to the— there's an awful lot of testing you need to do to figure out what works, what doesn't work in your GTM, in your product, in your messaging. Once you perfect all of that, then you can expand. If you haven't perfected all of that, the problem is when you expand to a new region, there's going to be more local constraints, so more local nuances to keep in mind, and you're trying to optimize for both of them. That's an awful lot of load on an early-stage marketplace like that, especially if it's a new language, new cultural dynamics. I mean, forget about it. It's a nightmare. If it's a marketplace like an Airbnb where you're connecting supply and demand across those regions, then you kind of need to figure that out. It's again one of those marketplaces where your barrier to liquidity can be a little bit higher because I'm guessing in that type of marketplace, supply is going to be a bit differentiated. But you kind of need to take that leap because it depends on where you have the most valuable supply. For Airbnb, the most valuable supply was Europe because that's where people wanted to travel, right? So they had to figure that out. You'd still have sort of some sort of nuance on ensuring you have sort of a base level of liquidity in your existing locations before you expand to a new one. But expanding to new locations is sort of a part and parcel of building liquidity here, whereas in the first case, why it wasn't. So it really depends on, are these is this an interconnected network or are these sort of discrete networks? That's like the biggest thing to keep in mind. The second is localization is a whole lot harder than it might seem. It's not just language, it's generally different behavioral traits. I had a portfolio company that was called Smitten, which expanded from Iceland to Denmark and Sweden, and they were sort of flying everywhere except Denmark, and they couldn't figure out why for the longest And it's only later they figured out that smitten meant infected, which, and it was a dating app. So obviously that was not ideal for that company. So there's an awful lot of nuances, stuff like that that you're honestly not going to know until you expand.

Awesome. Well, this is a great last question. So yeah, we're almost out of time here. So once again, I appreciate taking the time to join us here today, Sameer. So this is a really great chat. I know we covered a lot of the ground and we had a lot of questions here. So we appreciate taking the time to walk us through it. through network effects, your awesome investing experience, and help answer some questions from the group. I know when it comes to marketplaces, it can always be— it depends as far as the marketplace when it comes to answering. But right before we wrap things up though, do you have any kind of like last tips that you might want to share with us here as early stage marketplace founders and teams?

I guess the most important thing at the early stage is just figuring out your first viable network. Andrew Chen calls it the atomic network, I call it the minimum viable network. That's pretty much the most important thing to figure out in your early days. So do you have enough utility within that sort of early base, forget market size, forget everything else, like that's the only thing that matters at the early stage. Sometimes founders lose track of that and start focusing on look at this big market, we want to tackle a big market first, like no, do not do that, target the smallest possible market first. Like, that's really, really important because that's going to help you improve all those search-to-fill utilization rate, uh, core action frequency numbers. Trying to go broad will not do that.

No, I'm glad you said it. So I feel like I'm always trying to reiterate that in the community. So that's a great way to wrap things up. And, uh, you know, last but not least, time for a quick plug. Where can we keep up with you at?

So, uh, you can follow me on LinkedIn or Twitter. Uh, on Twitter, I'm Sameer_Singh17 because Indian names are unfortunately all too common. Or you can connect with me on LinkedIn. I also write at breadcrumb.vc. Anytime I have something new to write about, that's where that will end up. I also teach a course called Applied Network Effects about network effects predictably, which I teach about 3 times a year on Maven. So if you're interested in that, just Google for Applied Network Effects and you'll find it. And if you need to send me a deck, I'm at samir@breadcrumb.vc.

Get ready for some, for some, some decks coming your way from, from founders here. So we once again really appreciate, you know, taking the time to join us. And I'll include links in the recap post here too that I share after. So yeah, thanks for joining us, Samir, and thanks everyone for joining in the questions today.