top of page

7 Questions to evaluate Network Effects Moat

Network Effects is the most hyped and confused topic amongst startups and investors.


"Network effects have been responsible for 70% of all the value created in Tech since 1994. Founders who deeply understand how they work will be better positioned to build category-defining companies."

- James Currier @NFX, SFO based VC.

Here are 7 questions to help you bust those false claims.

What is the difference between Uber's network effects vs Airbnb such that uber is swimming amongst so many competitors like DiDi, Lyft, Grab, etc whereas Airbnb seems to dominate with no other clear competitors?

Why do every additional user on Airbnb or Cameo add value but not every supplier on Fiverr or Upwork?

What does it mean when a network effect is strong but score weak on defensibility? Or what if a startup scores well on the strength and defensibility but is not the first in the market? Is it still worth building a slightly better version? What are some network effects that will really be advantageous to have as a first-mover? In this Platform Series with Prof Julian Wright and A. Prof Andrei Hagiu, (>10HBR) world experts in Platform Strategy and Network effects + Angel Investor in >200 startups, we talked through the above questions and gave examples of companies evaluated on those questions.

Another way to use this video, is to flip it and ask what can I do to build the network effects and increase the defensibility for my business. We even have Brian Cox from Succession (TV Series) mentioned in an example. (Beware of swearing n video!)


(00:00) [Music] Welcome to our platform series with Andrei and Julian. In the last video we talked about three methods of how you can move from a product to a platform business in this video we will be talking about one of the most hyped and confused topic network effects: not all network effects are created equal.

(00:20) With all the hype and claims from startup to investors about their network effects we will explore the seven questions that you can ask to evaluate how defensible or strong are these network effects. Let's set the context with the first question: what are never effects and why is it so desirable?

(00:37) So network effects arise for products or services where the value for an individual user or customer increases where as more customers use the same product, so think social networks like facebook, you know the value to individual users increasing the number of users or things, think of any marketplace, so you have buyers and sellers, the more buyers participate in the marketplace the more valuable it is to sellers and vice versa. Why is it so desirable? So as I think we had talked about in the last video the reason like every investor and every entrepreneur wants to build or to invest in the next big network effect business is because the number one network effects create huge potential like huge scalability so businesses can grow very fast at very low cost and because they can be very defensive so the more people or the more users join a particular network or a particular platform the more valuable it becomes.

(01:34) Obviously it's harder for them to switch to a new entrant or to a new competitor and at the same time they do this at a very low cost in the sense typically when you have platforms with network effects the platform doesn't really have to control or do everything basically the value is created by the participation of the users so if all goes well you can achieve a lot of value at very low cost; you grow very very fast. What is the first question they can ask first and most primary question would be just to ask like to what extent the users really care about how many other users are buying the same product or service right.

(02:07) If they obviously they don't care very much there's really not much of a network effect in the first place so often these network effects are not actually that material and we want to sort of look at like how important really are they; so you know just to give you an example I mean obviously there's you know andre mentioned facebook and um businesses like that which have very strong network effects right, and very important for their business but there are other businesses which you know people talk about as you know there's network effects and this platform businesses and so on.

(02:40) For instance p2p lending platforms right; p2p lending platforms where you know investors invest in loans that are directly with businesses that taking out loans and these were you know the examples where it would be like lending club right, so lending club was one of the first in the US that did this; individual investors put in money and they were lending out to individual businesses it could be also other individual you know. What are the network effects in this kind of business right? From the investors point of view well they want to have multiple different businesses they can lend to so obviously, I care like there are more than one business they're lending to but they don't need like thousands and thousands they just need enough to diversify their portfolio a little bit. On the borrower's side the company side they don't actually need they don't really care how many investors are putting their money in they just need to access money they just want to borrow money they don't care where it comes from. So the network effects here are actually not very strong right and interestingly if you look at these p2p lending platforms over time they moved away from having individual investors towards institutional investors.

(03:48) So now most of the money is coming from a few large institutional investors if you look at lending club, which is one of the biggest in the US, about 95 percent of their money is coming from institutional investors; so the idea of like having lots of individual investors lending to lots of individual borrowers has actually not really worked out that well right. So I just want to emphasise here so this is a very in some sense it's almost like it may sound like a silly question but it's actually just making sure it's very important to ask; it's a very basic question it's very important to have very high standards right, so like okay yes by definition network effects involve users caring about the presence of other users but you really have to dig a little bit and say like do they really really care about other users on the same side or on the other side. 

(04:36) We see this all the time where like if you look hard enough you can start basically seeing network effects everywhere but the point is like in some cases they're just like not very strong, they're not very important, like lending club is a good example and another actually another common confusion that people make here, which we see all the time with startups, is they tend to confuse network effects with virality; so they would basically say, oh you know this product has network effects because people will tell their friends and they would adopt the same product. No, that is viral, that just means that that would be true for any product you can put up a shoe or two a brand of toothpaste and if you do a somewhat clever marketing campaign, people tell their friends. That's not a network effect; that's just basically people becoming aware of the product. I think it's very important to separate the two.

(05:20) Come back to the question do people really care when they're buying the product about the number of other people or other users that use the same product? Maybe the wework example also would be a good one to clarify because obviously you know there are some discussion that we will have you know network effects and there is an article that you have linked here about whether they really have network effects or not sure; I mean that's a good example of over claiming network effect; so I think the legendary founder of WeWork at some point was claiming to everyone who would listen that his business is not just like renting retail uh real estate space, it's an operating system for life and therefore it has network effects because you prove there's like lots of other services that local businesses would provide to the companies that rent space and WeWork. So to some degrees the fact that WeWork has lots of tenants attract some local businesses to offer their service, we have to again come back to the question, when businesses decide whether or not to rent space at WeWork, do they think about how many other businesses are providing services like whatever local stuff at WeWork, and no one really cares. I mean they're thinking what's the price, what's the space and so on; so it's just not very important. Let's talk about the second question. So the second question that we like to ask is…

(06:40) To think about how quickly does this extra value get as you add additional users diminish right so you might have a situation where there's a lot of value added as you add more users right to each individual user, but that only lasts for some time and then it peters out and of course you know most network effects will eventually diminish right as you add the thousandth facebook user you're adding quite a lot of value, as you add the million you're probably still adding quite a lot of value, but as you add the billion facebook user like how much value does that user add to everyone else you know.

(07:15) At some point it must diminish at least when you run out of the population on earth, so you know we're trying to understand like what does that curve look like as you think about adding more and more users, and thinking about the marginal network effect that's generated and seeing how much that diminishes so you know one example where I like to think about where this you can sort of concretely think about diminishing would be something like Uber right. With ride hailing so you know you asked how many drivers you need in a city to get to the point where customers are pretty happy they can get the core Uber and it comes within a few minutes right at that point as you add more and more drivers, the additional benefit in terms of this network effect is really diminishing right. I mean once you get down to a few minutes you double the number of drivers it's not going to add that much more value so that sort of concretely illustrates this idea.

(08:09) Even though you know Uber does have quite long lasting network effects because you need to get to that point where you get it down to rights within a few minutes but you can see it diminishing at that point. Just to sum it up and to compare with the first question; the first question was about the total level of network effect; the second question is more about like how quickly do you get to this total level. So in some cases the value of the network effect flattens out very quickly so you know after you get a few users basically you get all the value of network effects that was Julian's Uber example; in some other cases you actually need a lot of users to generate the total value so the network effects keep increasing for longer; so yes obviously if we have a choice we prefer to invest or you know to look at network effects that you know remain that keep growing for much longer.

(08:55) But at some level it's also a static thing right, because as you know product features and functionalities as they grow; I mean you can't really fully predict in the beginning as well because the startup obviously would add product functionality features that could increase value and increase the connection between the user that will add value, and also increase additional network effects wouldn't that be? I think that's a very good point Josephine, so I think that's so it is as we go through these questions I think it's important to keep in mind.

(09:29) There's a part of it which is kind of exogenously given by the nature of the market by the nature of the product, but certainly a lot of cases the company startup can do something about it right. I mean you can sort of do something with product design or adding new features in order to kind of push the envelope. Now I think there are limits so again there's exogenously given limits and we try to get at those, so I mean with our questions we try to get at the exogenous limits but obviously this should also suggest to the startups well maybe there are ways in which we can enhance the value of our network. 

(10:00) I would like to flip it to see how we can add value and you know make that happen, instead I mean if you think about social network, and we talked about facebook right, it has a pretty strong network effect that's long lasting I guess, and maybe it's helpful to think about when that might not be the case right, and that might not be the case if you have a social network where you don't actually want too many people right, it's somewhat exclusive, so if you think about a club like a country club or that kind of social network, well actually you know as you add too many people the value becomes less and less not only diminishes it may even become sort of the marginal value becomes negative, so you know you can see in those types of social networks there's a very clear diminishing network effect. So I think it's very important to be clear here this is not about so when we say the graph like how quickly or how slowly the network effects are increasing, this is not versus time this is as a function of number of views, it doesn't matter if it's today it's in the future it's like total like we're looking at okay suppose like everything goes well I don't care about over what time, it's how much do people care.

(11:10) So back to the first question, how much do people care over other users, this is not over time so it's like in general. How much do they care about other users? Second question is as a function of other users, how much does the value to an individual user increase? I think the key thing was really obviously you take time to accumulate the use of because of the same time but obviously is for a number of users, but the thing is as an investment really you're investing in the pretend future potential of the startup right from the perspective, you're investing in the potential but you know you're really thinking about market size right, and what is the potential of this and ideally we're investing in a business that has a very large market size where that size is realised through the network effect right; as it becomes more and more valuable more and more people join and we want that process to go on for a long time and reach a large scale so the business becomes very large through the network effect. So question three is asking for marketplaces in particular do buyers view the suppliers as differentiated or distinct right, or do they view them as interchangeable homogeneous providers of the same product or service?

(12:19) We like to contrast marketplaces where really what matters is having a wide range of suppliers even sometimes what we call a long tail of suppliers that are going to appeal to different types of consumers, like you know your Airbnb, your ebay right, which have thousands and thousands and thousands of suppliers and you know they all have they're all distinct, and they'll provide different value to different types of consumers as opposed to marketplaces where the supply side is very homogeneous. And so this might be the case of something like Fiverr or Upwork or Uber where you know at least in Fiverr and Upwork for certain basic services and Uber obviously ride-hailing it's a fairly homogeneous service, the consumers don't care that much about the particular supplier as long as they provide the basic service at a reasonable price, that's all they care about and so this distinction is important, because if you have a lot of differentiation on the supply side then the sort of the network effect is going to be much stronger, much more value can be created by adding more and more suppliers because you get that heterogeneity, which is what the consumers are looking for right, they can find their perfect match.

(13:36) I would add this also ties into so it's the strength of the network effects but obviously it also ties into defensibility as well, if the supply side is differentiated well you actually need this base of differentiated suppliers, but if say it's a commodity it basically everyone draws from the same pool of undifferentiated suppliers, well it's much easier for someone to come along later and imitate what you're doing and drawing from the same pool of supply so it's less defensive. 

(14:04) Would you talk more about the one that has the distinctive you know supplier like Airbnb and Cameo? Thanks for letting me discover Cameo because I had so much fun with it, you know having Brian Cox from Succession, you should tell us you could describe Cameo so people know. Josephine you should let us know your favourite actors and we'll offer you whatever like a personalised message from your favourite actor from coming for your birthday. I was surprised to find Sarah Palin there. I'm surprised I was supposed to find Kenny G there because I'm like is he still alive? Because someone and you see someone I'm sorry I'm like my god this man did not age at all, you know from the perspective and obviously everybody's watching Succession. In Australia might be very popular well it's doing really well yeah. Cameo is a great example of this because you know they just have such a wide range of you know different celebrities actually not just you know from movies or music right, like you get some pretty unusual characters on there, from ex-politicians, like Rudy Giuliani is there now, like anyone that would make fun of Rudy Giuliani you can get him yeah.

(15:30) And you have comedians, you have the reality tv stars, you have composer like David Foster is it? Or catherine miffy you know kind of thing and it's such a… I wonder and I will ask you for a question in terms of defensibility, is it a lockdown thing? Also there's exaggerated demand for it. Do you feel that after the lockdown it might not have the hype, sort of like the novelty of it might wear off you know? From that perspective because it's like, why do they do it? Like if I record this for three minutes I make 300 bucks you know as an actress, like why? Well I mean there's many use cases right like I mean definitely lockdown help because people are bored at home and looking for something novel to do, but I mean you can think about many different use cases I would just like to get, so for me I would get like some basketball player because Andrei is a big basketball fan and you know like surprise Andrei on his birthday or something and get this guy to talk to him you know like people do it as a lot of them are gifting I think you know, but for other people I think a lot of the use cases don't necessarily depend on the lockdown. I think it was already doing very well before that.

(16:37) Before the lockdown yeah, but it's a phenomenal way for a lot of these communities and artists to find work with their time as well given the fact that they can't perform, so that's like a really interesting one, so thank you by the way just to make it clear to your audience so this is basically what they offer is access to c-list celebrities and below. You're not gonna find Brad Pitt or anyone that actually like anyone that's really famous just to be clear, like my favourite basketball players are not going to be there if they're still serious like you have to go look pretty carefully so, but yes it's true like these are people who probably again, I don't know, they probably don't have a lot of work and this is a very easy way for them to make money, it's super easy it's super simple and the fact that they're c-list right that just sort of illustrates this idea, like this long tail of supplies right, like these are not the most popular suppliers that or content creators that everyone wants but still there is some demand for them from some people, but that's the distinctiveness actually add on to the variety, which is what you're saying, which is like Airbnb any additional you know artists you know whatever add to the flavor of the marketplace itself. I mean when I use Airbnb I'm looking for very specific things right, which you know not everyone else is looking for, and so having that richness and supply really helped in finding the right matches and that becomes very defensible.

(17:57) I'm always going to search on Airbnb first because I'm more likely to find that particular place that I'm looking for so what would be the fourth question. So the fourth question, we can make this short and sweet, and I think is pretty obvious, is the network effect global or local? And you know you can ask the question well which one's better? It should be obvious that it's better like all other things equal we want to invest or we want to see global network effect. So a good example good contrast would be say compare Uber to Airbnb in case of Uber the network effects are very local in the sense that I'm in Boston, so if I open my Uber app I care about my wait time so therefore the number of drivers they're in Boston, I couldn't care less about how Uber how many drivers they have in San Francisco or in Singapore or anywhere else for that matter, and vice versa the drivers only care about the number of users in their area or in their city not somewhere else. Now contrast this with something like Airbnb which is fundamentally like a global business like network effects are global, again I'm in Boston but I want to travel to Singapore to Australia, so obviously it matters to me that Airbnb can offer houses or places to stay all around the world, and the same thing for the hosts right, they can get people from all around the world so all other things equal, I'd much rather have Airbnb because you know the network effects span the globe as opposed to being completely low, and same as we talked about before it also makes them more defensible. In the case of Uber they basically have to fight city by city so you know they get Boston, they have to go to San Francisco and start from zero. Airbnb once they had a bunch of cities now it's very easy for them to move into a new country, for example, because they already have lots of other countries lots of other users so that gives them a huge advantage. I would just add on that while we would much rather invest in you know Airbnb than Uber, of course it's also much harder for Airbnb to get off the ground right like building a business which has global network effect really relies on adoption across a wide range of places you know.

(19:52) They might start somewhere but the ability to grow that and get to that outcome is much harder, so you know we also look at obviously whether the likelihood that's actually going to get off the ground and that just illustrates how much more defensible it is right once you've built one of these global.

(20:08) But you know a business with global network effects it's very hard for anyone else to come in they have to somehow enter in multiple markets at the same time and grow the business to compete. I guess you're assuming that these are what I call network effects, or marketplace that need a physical delivery or physical kind of a goods or whatever right, I mean like Airbnb obviously you need to fly you know Uber is location specific if it's a digital delivery type of a marketplace it's immediately always global right is that even like yours wouldn't that be location doesn't matter yeah that's a good point I think in general.

(20:43) That's true I'm trying to think I mean there are languages differences. I think in general history exactly but there might be for actually for example we were just discussing recently a marketplace for tutors for languages right, so it is by definition global but it's not as global as Airbnb because basically you need the language speakers from like very specific countries, for example if it's english so I guess like you can find like there are people that want to learn english in many countries in the world and you can find many native english speakers while in english countries, but and yeah so it is global, it's not as global as Airbnb I think by and large just again I think you're right so with digital products and I mean the other part is like you're right so I think it's as long as there's a physical component then there may be some for like international travel or or shipping friction, but with digital things in principle there should be like the geographic barriers vanish, that's why I think because then there's no local global definition in any digital delivery, whatever buy sale connection or whatever exchange that it is with the digital deliveries regardless, just like your NFT marketplace it doesn't really matter that's right yeah. I think the future will be there will be more of these like global ones, but again I think it's very important to realise what Julian pointed out. I mean there's lots of reasons I mean there's different countries have different regulations, different policies so geographic barriers still matter to a certain extent, even for the most digital of marketplaces, I mean you'd be surprised that something like Cameo, which is clearly international, there are a bunch of competitors in other countries right, like if I'm gonna do the k-pop market or the you know particular countries, Bollywood, I mean I need first of all I need to have I need to get those people on there right, and I if I don't have the connections even though I have a lot of people in the US and Australia and UK coming to my site, it may not help that much right and so you could get a local player who has those connections, build up a cameo competitor in their local market because they have you know the right they can get the right see this celebrities on there from their local market right okay, so I guess you know that's in their own language right like their own language.

(23:03) I don't think it covers all languages yeah okay, so I think your global local like physical distance culture barrier or whatever anything that's specific to the delivery of the connection because, even like your example with sprinkly right I mean like that your favourite simply it's like the connection is has to be local because obviously I got to go to the swimming pool around me so it's like the connection just has to be within the cluster of the local swim simply would be just like Uber right, they have to go mark it by market and build up, they have local network effects they have to go to each new market but the one thing that helps them which is more about virality is they get word of mouth through press because it's an interesting business right, and so that helps spread the word across different locations. I would add one thing to swiftly though I think it's an interesting example I would put it somewhere in between Uber and Airbnb certainly maybe closer to Uber.

(23:57) What you're saying was that because swim please you say it's like in between like Uber and Airbnb and I just wanna explain that is because like simply in some ways is because he has a distinctiveness of the different pools I get to see which is I feel like Airbnb right, but it's different, it's similar to Uber because of the locationality kind of thing that is required, so it I guess whether the network effects are local or global depends on how people are using. If it's mostly let's say families that want to go in their own neighbourhoods that's fine then it's like Uber, but if it's people that travel internationally and you know say I want to go to Australia and you know when I'm there I just want to go to someone's house and find a swimming pool, that makes a little bit like there's an element of global, but I think Andrei is also saying like Airbnb you might plan your trip right, and as part of planning your trip you might want to have some time swimming when you come to Australia, nice weather right, so you might also find some Swimply locations, I mean you could say the same thing for Uber right, there are people who travel who want to have open their Uber apps and be able to use it everywhere, yeah I guess the question is, do you think that this is so it comes back to our first point like how strong is that effect? So I think there's an interesting point here so maybe it comes down to what I would say is like yeah because you can so in principle yes you can but I think it's when do you make the decision? It's like if you make if I have to, so in the case of Uber there's no question I don't make like I'm not going to plan my Uber before I leave my house and I'm not going to book my Uber in Singapore before I leave from Boston.

(25:35) In the case of Swimply I don't know but I guess you're right like it's more likely that I get to my destination then I start looking for this kind of stuff. For Airbnb obviously you do this like once you're at it when you're at home you book the thing, but you might if you're in Singapore like I am you might download Uber on your phone because when you travel you're going to have that option. In fact I do that so I have Uber on my phone Uber is not in Singapore I can't use it in Singapore but I still have Uber on my phone I have to I have my you know account there so that when I travel I can use it so there's a little bit of a international global network effect, but the point is it's minor if there's some better service that when I travel I can use, it turns out like you know using another provider is better, then I'll just switch right, but by the way so I think there's two separate points in there and that's a nice transition by the way to question five. So what you just mentioned with Uber I would say it's an issue of like switching costs or like familiarity. What I mentioned I think there's a separate thing, it's like, do I make the decision to draw like to basically like to transact with someone on this marketplace, do I make it before travelling or do I make it like one time and once I'm in a specific spot? I think both matter but this last one that you mentioned obviously provides very good segway into question number five, well that's before we go to the fifth question, because you have a key distinction that you say that the last four question that we have talked about it's really about the strength of the network effects and the next three that we're going to talk about are really about defensibility. Can you just quickly talk about like what is the difference?

(27:08) You've been thinking that if it's strong now effect it just means you know good defensibility, so what is the subtle difference between the two? The defensibility we already mentioned several times you can see there's a pretty blurry line between these two, but I mean broadly the idea was in those first questions.

(27:24) In some sense we're talking about what is the potential right? What is the potential you can reach? And we're trying to understand that you maybe have to reach very high potential in terms of growing business through network effect like zoom you know zoom has massive network effects and it's growing very big, but then the question is, is that defensible? And Uber is another example right. Uber has strong network effects because you want to join where there are more drivers and drivers want to join where there are more you know people looking for rides, so strong network effects but how defensible is it? 

(27:57) So that's where the next questions come in. We're sort of asking okay even though there may be very strong network effects doesn't necessarily mean it's very defensible business. I agree I would add so the one thing I would so I think the right sort of, I've done this with students over the past few days and it just occurred to me like a good metaphor to have in mind is the following, so of course they're related, there's no question that if the network effect is stronger maybe defensibility is higher but they're certainly not equivalent I mean I think those Julian's point is very important, it's very possible that you have extremely valuable, extremely strong network effects but let's say what if it's very easy like there's no switching costs like everyone can multi-home on multiple platforms at no cost whatsoever well there's not a lot of defensibility so it's possible to have very strong network effects but not very different, so the way I think about defensibility is like you know I show students like I showed them a castle on top of a mountain and it's basically defensibility; obviously it comes from like the metaphor comes from war so this idea that it's not about like how good my features are relative to my whatever my opponents, but it's basically what you're looking for in defensibility is like, is there something inherent in the position that I'm occupying that makes it harder for the second mover to come beat me; so obviously if I put my army on top of a hill I don't have to be better than my opponent; just by virtue of being on top of the hill, I just have an advantage relative to the whatever the next next army is alone; so maybe another way to say if I understand this correctly is that if it has strong network effects the potentiality that this business will be a good and strong, the market itself is a good potential market and business that will occupy a valuable and you know speak enough market a business in a big and a valuable market itself right, but it if it is not highly defensible in terms of never effect is that probably it will become a dual pulley or… 

(29:50) it will become a multiple you know players that's in this valuable market right, because for example Airbnb highly defensible and strong but possibly highly defensible and that's the reason why you don't have a another similar player around you know for the matter, but Uber highly variable and highly strong you know network effect therefore it's a good company, good value market, high growth but then you have a lot of different players you know they're in the same market because it's highly not defensible, the defensibility is not as strong is that how I think that's a really good way of thinking about it.

(30:25) The network effects of the market may be very strong but it doesn't mean an individual player can therefore monopolise it right, it's not necessarily going to tip to one player if it's not a very defensible kind of network effect right, and you know you see clubhouse would be another example where you know it looks like there's pretty good network effects there's discoverability and so on, but isn't is actually not very defensible because it's easy to copy; so the fifth question is about multi-homing so how difficult is it for buyers suppliers or users to multi-home and let me define multi-homing first.

(31:02) So multi-homing is just the idea that the users are going to join multiple competing platforms right in order to interact with users on the other side, so developers may develop for ios and android because they want to reach all the ios and android users right, merchants may accept mastercard and visa because they want to reach all the consumers who have those different are using those different cuts and there's lots and lots of examples like that so people are joining multiple platforms we say they're multi-homing right, they're adopting competing platforms and the question here is how difficult is it for users to do that? 

(31:42) So if you're thinking about you know users multi-homing on android and ios from the consumer side that's kind of difficult, you're going to have to go and purchase both a android phone and an ios phone right, but maybe for mastercard and visa that's easy right as a consumer it's easy for me to get both types of cards right and use them, so multi-homing can vary and our point here is if multi-homing is very easy intrinsically, quite easy there's no barriers to it then you're going to have less defensibility because it becomes much easier for a new platform to come in and the existing users of the incumbent platform don't have to give up that platform in order to use your service, they can multi-home they can use both at the same time and that makes it much easier for platforms to enter. So if you look at those platforms where you see a lot of multi-homing like payments mastercard visa american express and so on you see you don't see tipping right.

(32:38) Competing platforms Uber and lyft right same thing multi-homing is easy on both sides lots of people drive for both Uber and lyft, lots of people use both Uber and lyft on the consumer side. Food delivery another example booking platforms like booking expedia all these examples you don't see tipping, unlike Airbnb, you don't see tipping to one player because multi-homing is quite prevalent. What would be some examples of that would stop somebody to multi-home or create it make it more difficult for them to multi-home? So this is where I mean not to beat up on Uber but I think the comparison between Uber and the Airbnb is pretty interesting here too right. So as Julian mentioned I think multi-home is very easy on Uber both for the drivers and the riders, now think about Airbnb contrast is with Airbnb I would say so in the case of an Airbnb they do have competitors but they're very small so the competitors would be like vrbo and home and away. From the perspective of users I think multi-homing is relatively easy I mean it's pretty easy to have accounts on both and you can I guess you can search for homes on both. The interesting part is if you think about it from the perspective host so yeah it is easy to create accounts on both but it's actually pretty difficult like because you only have say one house it's hard to manage the calendar like or the scheduling for one house on two different platforms, like you probably don't want to do that because it can lead to obviously the conflict so in that I mean that's one factor that can make you know scheduling is a big deal like if it's you know physical property under the limited capacity; obviously that makes multi-homing harder and of course by the way you can think about it so you can always take the perspective of the company and think about well if multiple costs are low what can I do to make multi-homing harder.

(34:18) And therefore I guess force users if you believe that users will choose you you want to force them to choose and stick with you and not your competitor you know, so there are things you can do to make I guess to make them make some investments in using you that basically makes them less likely to use other platforms I think.

(34:33) That was a good one the calendaring because I didn't think about that the reputation and ratings people are selling something that you are you know on amazon and ebay you know that you really build up you know your credentials monetary or non-monetary, cost of joining is one, so my other question is what about utility itself of the platform of usage of putting a product onto it or which actually then create data stickiness onto the platform because if you have more data already on one platform that you're invested in it you would not shift it you know for what example I'm having in mind is obviously a client I have is that it has a little delay it's like a delivery for a certain product but it could be used for the sme or the small players itself to actually invoice their client or to create their client, collect the client data at the same time so there's a utility on the platform that they provide as usage.

(35:26) And as is more invoice and customer queries or quotation or requirements come to the platform they are actually creating and owning holding those customer data on that platform so that it's more difficult for this person to want to switch to another platform, so that's actually a very important one especially when we look at b2b businesses right we often trying to look for those that are building up something like a system of record for their clients right, and that actually tends to not just lock in the client's terms of switching costs but it means multi-homing is probably not going to work right because they won't have all their data in one place. So yeah I think that one's actually very important. Another point we mentioned briefly which is you know the platforms can do things to enhance uh the stickiness and reduce multi-homing so it may be useful to think through like Uber's case right.

(36:19) Because you know Uber has some very obvious options it could use to reduce multi-homing the most obvious is to say to the drivers you have to be exclusively on Uber and you cannot drive for a rival if you want to drive on uber you can only be on Uber. Now this is this kind of exclusive dealing is something they actually did in singapore and they put in those contracts they actually grab had that when it came in as well, but the issue with that is it's usually found to be anti-competitive by competition authorities, so Uber and lyft will not do that in the US because it would be anti-competitive.

(36:56) So they can't but there would be an obvious way to create you know a cost to make it almost impossible to be on both platforms right that you're going to be kicked off if you get caught multi-homing. So another thing they could do is they could try and you know put give drivers incentives to do all their driving on one platform right so they have they design a reward system a loyalty system where if they do more than 40 hours a week driving for Uber then they get like additional incentives and rewards right in terms of the payments so this is sort of thing they do which is kind of like an exclusive contract but it's done in an indirect way to try and avoid the competition concerns. Another thing they might do is just say okay we're going to employ you as a worker right as a full-time Uber employee you know then they have to give them all the benefits associated with employees so they don't want to do that.

(37:50) But you know these are all the kinds of issues that they can think about on the driver's side similarly on the rider side they can try and reduce multi-homing by creating a membership plan so you pay a certain amount per month then you get you know rides on Uber up to that point and that you know tends to reduce multi-homing once someone's in one of those membership plans that's good. Let's go to the sixth question right so the sixth question asks how easy is it for users to coordinate their adoption decisions? So this requires a bit of explanation; it's important to understand that with network effects, a fundamental reason that network effects lead to defensibility is the fact that it's very difficult once users say have coordinated on one platform it's very difficult for them to coordinate our new platform, so let's say you know there's a coordinator on the platform everyone likes and there's a new platform that's better that comes along.

(38:44) Well even if it's better if the users can't talk to one another let's say they can't have a whatsapp group and then say hey we should all jump on the new platform it's better then the coordination is going to prevent the new better platform from emerging, so in most cases that's going to be true.

(39:01) So think about for example Julian and I always give the example of craigslist. So craigslist is like online classified that's been around since like 1996 and it's a really crappy experience I mean it's literally the experience of that they had in 1996 with very very few improvements. Now again the reason they can get away with that is because a lot of buyers and a lot of sellers users and buyers and sellers cannot coordinate to say like what the hell are we doing here we should just go on somewhere else there's lots of other better options so in that regard because you know buyers and sellers want to discover one another there's just like there's no way for them to coordinate right. I mean this is true in most markets but there are examples so this is why it's important to realise there are some cases um in which actually it's a lot easier for users to coordinate on new platforms so take the other extreme something like you know.

(39:45) Video conversa video chat or video conversation apps like zoom for instance, well on zoom you typically don't want to talk to strangers it's basically you talk to people that you already know, so when when we use zoom well typically someone sends a calendar invitation to everyone else by email, well they could just as easily send a calendar invitation for google meet or microsoft teams or something like that, so coordination here is actually very easy, so if zoom was really that much worse than google me or microsoft teams it would be very easy for people like I have no reason to use zoom I can for the next call we can just jump on on one of the other ones. Now of course there are other things like that we talked about I mean it's basically like you already learned how to use zoom, there's some switching costs from a coordination standpoint there's absolutely like it's very easy for people to coordinate on different platforms so just focusing on that aspect that actually is a concern of defense for defensibility in the case because there's always like if someone comes up with like truly a much better platform like most people can actually choose to use. Let's give the example that zoom decides to charge a dollar for every time you use zoom for every user right or even 10 cents right. What would we do we would all when we want to communicate we'd say would send an email to each other or whatsapp would say let's not use zoom anymore because you know we don't want to pay let's all use microsoft teams so that's the coordination of all of us to switch over and use a different platform whenever we want to communicate. Now that coordination happens amongst the people initiating the call right like if I want to have a zoom tomorrow with a with a group I'm going to send instead of sending a zoom an email to say that speed on zoom will say let's meet on microsoft teams because we save that 10 cents or a dollar each you know so the point is even if they start to charge a small price you know a lot of people would switch a lot of like you know consumers would switch of course they still can charge for business users which is a different market.

(41:41) But their regular sort of just household users of zoom would just switch to another platform and they can do that because they can coordinate okay so in some ways is to coordinate in some way to go to another platform or to transact in another platform kind of thing is to orchestrate that in that way.

(41:57) I mean imagine trying to do that with we want to switch away from using ebay or craigslist or amazon right when all the sellers and all the buyers want to switch away we have no way of communicating amongst ourselves that we should do that and transact on a different platform right because it's just the nature of that those marketplaces is they're offering discoverability so we don't actually know each other before we transact and therefore there's no way for us to communicate that oh we should actually go on this other marketplace because it's actually better right there if you think about craigslist there's other alternatives that are much better in the US you know have a much easier to use can take a photo and upload of the product and you know it's just a much better user experience like offer up, but still craigslist persists for a long time just because it's hard to get all the buyers and all the sellers who don't know each other to coordinate and move over to that better alternative like, from this regard that's any time there's discoverability that you don't know so the way to think about this there's a certain this is not broadly it's not like they're not that many marketplaces where this like coordination is easy in most marketplaces that will come to your mind.

(43:07) Coordination is very hard but there are I think it's very important it's very important to realise that there are many in which actually coordination is quite easy like video chat apps like it's very easy like we want to use house party you want to use whatsapp we can just say okay we'll use whichever we prefer for whatever reason that actually makes it less defensible; precisely because I know everyone I want to talk to and I can we can easily communicate with one another and say we'll just use a different platform. The way we like to think about is I want to come back to Julian's point is discoverability, there are other types of marketplaces where if you're only going there to transact with whatever people that you already know, coordination on a different marketplace could be easy the moment you actually go there to discover new transaction partners there's a coordination problem and therefore it's defensible. So an example what Andrei is just saying would be something like fitness instructors right so a marketplace for fitness you could maybe online fitness right, so if you have instructors that already have a large client base right and they want to switch to a different platform they can just switch and then they can inform all their clients that I'm now on this other platform and all the clients come across, so it's not difficult in that case to for suppliers to and the users to switch to a different marketplace if it's much better or cheaper because we're assuming that the main purpose of that platform is not discoverability but the moment that you know that platform is primarily one that you go to to discover a new fitness instructor then it becomes much harder, because you know they're going to attract users coming in there who are just looking for new instructors and instructors want to be on there because they want to get new users.

(44:46) I was just going to add just to make it clear the example of the fitness instructors and users the way they can coordinate is like a lot of these instructors have instagram followers right, so if most of their users are following them on instagram they can say listen I don't like this platform anymore let's just switch to the other one; that's how coordination works in that case so yeah that's a problem in that case you would have coordination becomes easy and it's certainly a defensibility problem. What is the seventh question? Yeah so seventh question is whether the matching of buyers to suppliers happens synchronously or asynchronously? So very very easy example to think about in the case well since we beat up on Uber and all this time like we finally get to say something positive about Uber; so in the case of Uber the matching happens synchronously means like if I want to ride it has to be right now.

(45:32) Most of the time yes you can schedule rides but most of the rides are basically like spontaneous, I need to ride as soon as possible so in that regard if you if most of the matching has to be done synchronously, that actually is defensive it creates more defensibility because it means you have to have liquidity, you have to have a lot of drivers at a given moment in time so you need in liquidity and therefore actually becomes harder for a competitor to come in because they have to have liquidity at any point in time. Now contrast this with marketplaces in which the matching is asynchronous so maybe you can get to do the booking ahead or like you know it's a long time in advance, all other things equal that actually makes a little bit easier for competitors to come in so a good contrast to Uber I was trying to figure something in the same kind of space, there's a company in france called blablacar which basically is also like I guess riders and drivers both private, but in this case it's basically for scheduling a long distance ride. So let's say I want to go from paris to north or some somewhere else in france and I don't have a car but there's someone someone is driving there next weekend and they have a car so they can list.

(46:37) Okay we have two seats in my car and I'm looking for people and they have to pay me 50 and you know I can go and book it so this is like a little bit like longer term there the matching is asynchronous and I would say there would be well in that case if someone wants to come in it's a little bit easier to basically build up both sides, because you don't need liquidity at every single moment in time, so it's a little bit easier to get started one of the entrants in singapore into the right-hand market that's exactly how they entered they only focused on next day right and cash daring so pooling right where we have more than one passenger riding together so schedule rides that was their focus, because that's much easier to enter right, they don't need to have drivers at any you know at a moment's notice everywhere across singapore, they can just direct their drivers to where those book rides are for the following day.

(47:29) So it illustrates the point. Another example would be in online gaming so there's a bunch of companies that enable like expert gamers to come in and help you play your game and win your game and you know get you through to the next level or beat your friend that you're playing against so you can basically just hire them, but you need them on the spot right like I'm gonna die I need someone so these are like really providing instantaneous help within game, and so this is an example where you really need the marketplaces to enable that they really need to have enough liquidity so wherever you are, whatever game you're playing, whatever time you can just get that help and you know that helps provide a little bit of defensibility, because they actually need the real-time you know component like it's not so easy for another one to come in and compete against one that already has can provide that expert instantaneously.

(48:23) They have to build up a sufficient base of expert players on that particular game to compete so it does sort of raise the sort of the barriers to entry, so basically your last three questions and on the on that increase the evaluated defensibility I guess that goes to the point where for the startup that are actually that has never effects the answer this that score well on these three questions, it's good if they have a first mover advantage right because everybody start from chicken and egg curricular mass, once you have critical mass all these three defensibility then makes sense you know what I mean because everybody anyway have to start from that chicken and eggs in order to have the liquidity in order to have that hard to coordinate to leave you know all that stuff right, actually you know just related to that point if you're evaluating investing in startups and you looked at these seven criteria and you say on this market you've got all seven things are holding, this looks really good if you're the first mover it's like that's great you know you definitely want to invest if you're the second mover, yeah there's an existing incumbent and you have all these seven features it's actually a bad thing right like yeah it's actually going to be impossible to break through, exactly which is what I was thinking to myself you know until it depends on the third aspect, well it's possible but then you know which means that actually it's about like the first you know the last three is like if they are the first mover they'll be great but the second one then is like well then you know how do you overcome it it's not possible, that's right, so let's go to the eighth question. So what are some ways that you know startup could actually increase defensibility you know of their network effects? I mean you can go back actually on all the previous seven ones and you can say well if we like if the answers to one of those questions was not satisfactory, so let's say the switching costs were the multi-homing costs were low you can ask well how can you make the multi-homing cost higher so that's one way to increase defensibility.

(50:18) Again you're the first mover now I want to raise the multi-homing cost because well everyone else makes it difficult for everyone else to come in. I can also increase just so both switching and multi-homing costs so I can get maybe the more so how do I increase those? Well generally speaking the more I can get users to spend time and to invest resources and customising their experience on my platform the less likely they are that they're going to use a second platform, you know the more they learn the more they you know they put stuff that's personalised they're not going to use a second pop. 

(50:52) Back to the coordination issue anything I can do to make coordination harder so if people don't like think about coordination when we think about is like let's make discovery very important so the more people come to my platform to discover new transaction interaction partners the better that is for defensibility. Again I'm the first mover I want people to come here to basically discover you know buyers to discover new sellers and sellers to discover new buyers the more that happens there's very little chance that the second mover comes in and it's going to be able to get them because they know well I can get all the buyers that I need there and I can get all the sellers I need on the other platform why would I move to a new entrance. Well so on the second one thinking about increasing discoverability I think that's a very real option for a lot of startups like they are building say our fitness and you know fitness instructor platform.

(51:43) A lot of them just focused on getting fitness instructors on board and using their clientele to build their network which is fine like that's a way to grow quickly but if they really want to be defensible they have to sort of turn on that discoverability part they have to enable the users to discover new you know maybe in a different area of fitness new instructors right, make it a destination where people come to discover the new instructors and make it a place where instructors want to come to where they can get new users and then it becomes much more defensible, so it's really deciding whether they want to do that and opening it up to discoverability.

(52:20) We see this on interesting we see this on online courses as well right like online courses they can just be providing tools for instructors to provide these classes teachable is a platform that does that right, and we talked about I think this and products and platforms right they can open that up to discoverability they can open up a way for their students to discover other courses right, and then it becomes a place that people come to to find out and discover new courses they want to take, and then it becomes very popular place for instructors to come to get new students right and that is a very real option that these firms face these startups face when they're building right, do I want to enable that discoverability and that does you know the plus side of that is the defensibility that it creates. What are some of the common mistakes that investors make about uh about network effects when they look at investing in businesses with network effects? There are many I would say a high level what we see that we found really annoying is basically over claiming network effects. Again at this point everyone and their grandmother is aware that network effects can lead to like very high value very high defensibility and he has very high valuations from investors, as a result there's a very you know there's this temptation over claim network effects even when they're just not there, and we see this all the time so the common mistakes would be like well claiming network effects when in fact it's virality. So some of the examples that we talked about we see lots of versions of that or simply just like claiming okay we have network effects therefore this is very defensible. The whole point of this whole like one hour and a half in the conversation that we had is precisely to discuss that just because you have network effects is absolutely no guarantee of defensibility.

(54:10) Network effects come in all shapes and sizes, some of them are great and make for great investments, some of them are actually no better than investing in a restaurant business next door. I mean just to emphasise Andrei's fault we just see it all the time like when we're looking at deal memos right like we've got network effects and then we're like where are the network effects? You know either they're very minuscule or they're just completely confused, they're not network effects at all right, and it just comes up again and again and again. The other sort of similar point is they'll say we are you know we are Uber for x so you know we're Airbnb for y, and when you look at what they're doing it doesn't make any sense at all right like not a good understanding of the business models of those of those businesses. I guess the only other thing I would add is you know on the sort of on the positive side sometimes there's they're actually overlooking the potential to add network effects into businesses or turn the businesses into platforms, so they're actually underselling the potential of the business because they haven't realised there's this product or platform transformation that can happen that is actually creates a much more valuable product or service than the original one that's you know being put forward.


bottom of page