a16z Podcast - a16z Podcast: Getting Network Effects

Episode Date: July 15, 2016

One of the biggest misconceptions around network effects (which are one of the key dynamics behind many successful and highly defensible software companies) is confusing growth with engagement. So how does one tell the difference between viral growth and network effects? How does one create network effects in different businesses? (Hint: it's not by accident!) How do you know when to hang in there because you see signs of network effects or just drop it and move on to something else? And what are some examples of teasing all of the above apart? In this episode of the a16z Podcast -- based on an event we hosted and slide deck we released all about network effects -- a16z partners Anu Hariharan and Jeff Jordan (who cover all things marketplaces, consumer, and more) share (in conversation with Sonal Chokshi) their observations, insights, and experiences. Because, why reinvent the, er, flywheel?

The content here is for informational purposes only, should not be taken as legal business tax or investment advice or be used to evaluate any investment or security and is not directed at any investors or potential investors in any A16Z fund. For more details, please see A16Z.com slash disclosures. Hi, everyone. Welcome to the A6 and Z podcast. I'm Sonal. And today's episode is all about network effects. And it's based on an event we did that was itself based on a set of slides that we released that you can find online. at bit.ly slash network effects A16Z. Welcome to our network effects event. Before we get started, let me actually just quickly introduce our guests. You probably know because you came here knowing who they are, but just to quickly tell you about who they are. Anu is an electrical engineer by training.

She used to work at Qualcomm, where she built the video streaming solution there. And then later she worked at BCG, and she's now a partner in our deal and investing team. And she covers all things, marketplaces with Jeff. Jeff's a general partner here. He was previously the CEO at Open Table, which is why the network effects theme is particularly strong here. And president at PayPal and SPP at eBay before that. And his board seats include a lot of our companies that have network effects like Airbnb and many more. It's actually funny because this whole deck started off as the result of an internal debate. And we're pretty opinionated as a culture. And there's a lot of strong opinions floating around. But I think it's kind of funny because this is probably one of the most internal contentious debates we've had. around what is a network effect? And I think it's because it's one of those concepts that's really easy to learn but difficult to master. So on that note, let's just start off by defining

what's a network effect. Yeah, simply put, I think the simple definition is as more users join the platform, it's more valuable to existing users. So it's really as simple as that, right? And it actually started with hardware. So think of the telephone in the early 1900s, the more people that had telephone,

you could actually make a phone call. So there was a network effect, if more people's, it was valuable to you because if more people had telephone, you could reach more people. So it sounds very simple at the onset, but I think the biggest misconception people have, in fact, we internally have debated about is, you know, growth versus network effect. So, you know, growth is different because especially as a startup in your early days, you're growing really fast, that's growth. But network effect is really value. How valuable is your service to the users as more users join the platform?

to be clear, though, it's growth and value, but what's the difference between that value and engagement? Is it enough to just say, we've got a lot of engagement and therefore we have a network effect? How do you actually know there is a network effect? I mean, we brawl over this concept because the easy observation is businesses with network effects typically are very defensible and often are prone to monopoly. You know, and we're trying to get in early into company, so that's not completely apparent when we're investing. So, you know, you're broad, you know, we'll have people standing on the table, there's a network effect. And here's why. And no, there's not. And here's why. And so, I mean, it is one of the key drivers of the

investment decision on a whole bunch of our consumer side businesses and some of the business side businesses. I mean, we actually told it, put a slide up to our pleas last year. Network effects equals moats. And, you know, if you can achieve it, you're defensible. You're less susceptible to price pressure. You're less susceptible to pressure of consumer acquisition economics and things like that. Because if your service is more valuable than the other, you can accrete that value throughout the entire process. I think it's valuable for users, which is really key, but that is actually where I think we get tripped up on what a network effect really is internally, because I think we sometimes confuse, oh my God, the product has viral growth. Is that a network effect? How do you tease those apart? Yeah, and I think that it's hard to do it, especially when you're a hypergrowth startup, which is why we have our own internal debates, you know, whether the company has network effect or not. I think viral growth is speed of adoption. Think of speed as vital growth. And usually if you have viral growth, it sort of implies that you're not spending anything in customer acquisition costs, right? So Facebook is a great example for the longest time, even to this day, they hardly spend money on acquiring users, right? And especially in the

initial years, they spent zero dollars. It was growing widely. So how do you actually decipher whether Facebook has a network effect? Because their MAUs and DAUs were growing quite fast. And the MAUs, you mean the monthly actives and the daily actives. But the one metric in Mark Zuckerberg has done a great job of saying this in so many interviews, the one metric that he really focused on in the first two years was retention. And the way he measured retention was daily actives by monthly actives. And if you see their chart in the first 18 months, it kept going from 52% to 55% to 57%. That showed that people really stayed in the network. That was a measure of the network effect and not just the viral growth in their users. And by way, just think about

the level of that number. I was talking to Mark in 2006. He reached out and asked to get together and we're chatting about business. And he got a text at the time, dinged it, and he reads it and he smiles. I go, what made you smile? And he goes, oh, this is, we get our daily metrics via text. I go, what metric? And he goes, it was this metric. He goes, how many of our cumulative historic users do you think were active yesterday? And I'm like, I'm just came out of eBay and PayPal. I'm like, two, three percent maybe. I think I got bold and, you know, six percent. And he goes, 53. And, you know, 53 percent of everyone who had ever signed up for the service was active the day before. And that, by the way, has only grown.

As, you know, they're installed-based age, it gets bigger in ages, that number's grown. And you're just kind of like, okay, that's a network effect. I mean, I think the thing that's most interesting about this whole discussion is it's not by accident, but it feels like it's by accident because you hear these narratives, you know, like, oh, my God, Facebook reached this number and it was taking off at campuses everywhere. And in reality, it's something people actually work at. And I think people here want to hear about how, because, I mean, what are some of the things that people do? Like, how do they start? I mean, Facebook's a unique case. Maybe we can start with Facebook.

We can have other examples, but what are some of the things that people do to get to that network before there's an effect? Yeah. I think the first question you want to ask is if you have a product is how do you bootstrap? What are some of the early growth hacks

that you want to do to build the user base? And then the second step is you're sort of testing your product market fit as you grow the users. And then what you have to quickly hone on, what is the aha moment that gets the user back to continue to use my product, which actually drives the network effect. And so in Facebook's case, contrary to what most people think, they actually went in a very clustered approach.

So the first was they first launched in Harvard, and you could argue that they could easily expand to all the schools, right? But I think their goal was they needed to get at least 80% of Harvard signed up on Facebook and sort of have that high bar of 50% plus engagement. So they wanted more than 50% to come back daily, which is a very stretched goal. And in fact, if you go back to some of the early investors, one of the criticism of Facebook was, well, you didn't grow fast enough. I mean, it's a different story that they did penetrate almost 80% of the colleges in 18 months and had high retention. But they did it in a very strategic way.

Until they hit that high bar on Harvard, they did not roll it out to Stanford. And the discipline that they got as a result where they were tweaking the product to get the right fit. So simple things, like the relationship status, they realized actually drove engagement really high because people felt uncomfortable asking someone else, like, what is your relationship status while you could go to Facebook and figure that out. Second thing that they did was they realized that if someone was connected to 10 friends in 14 days, they were much likely to return to Facebook. When they realized that they started, you know, every new user that they joined,

they worked on making sure that they got to 10 friends within 14 days. Maybe it'd actually bludgeon the crap out of you to get you to 10 friends. It would be like, you'd show up and like, do you know these 12 people? So one of the best hacks Mark had was, he was well known, he got in trouble for it. He hacked the directories in each of the houses. And so he pre-populated everyone on the Harvard campus into Facebook and then let people claim their identities. But you could then friend, it was populated when the first person showed up. It might not have had the same utility, but it was populated versus being the first person to show up.

It's like, you know, I'm in a stadium by myself. And so, I mean, I think he did end up in front of the president of the university for that hack, but the hack populated and made, you know, brought a bunch of utility. And so you would use the word growth hack. Virtually every marketplace we can think of did some, had some hack or hacks that enabled them to solve the impossible chicken and egg problem. The chicken and egg problem is, okay, an open table. It's only useful to restaurants if consumers are there. and it's useful to consumers of restaurants were there, where do you start? And so, you know,

there's a whole bunch of different hacks at different entrepreneurs that have used to try to address that. How do you get the flywheel spinning? Yeah. Well, what are some of the other ways that you've seen people stumble on this? Because sometimes people discover the hack by accident or is it deliberate? I mean, how do they really, I mean, do you actually wake up one day and say, okay, I'm going to figure out the hack for this market, please. Oh, yes, you do. No, no, when we're talking entrepreneurs, you're looking for the theory. What's your theory on how you're going to bootstrap? How are you going to get the firewheel spinning? How are you going to solve the chicken, the egg problem? Which side are you going to start on? So OpenTable ended up starting

completely on the restaurant side because there was zero utility to the consumer until you had a selection of restaurants. And so how do you get a restaurant to adopt it when no consumers are using it? So in OpenTable's case, is what Chris Stixon, you know, come for the tools, stay for the network. They build a suite of tools that they charge $200 for and laboriously rolled out one restaurant at a time throughout, you know, the country that had enough utility that a restaurant was saying, okay, I'll adopt that in the absence of a network. And as they slowly built that base of restaurants, and I mean slowly, I mean, a good salesperson would do three or four new restaurants a month. And, you know, there's, you know, there's 2,000 restaurants in San Francisco. That is

slow. And then at there was a point where there was enough restaurants on the system. It was a subset of the restaurants in the system. But the utility of making an online order was so great that they'd say, I ignore the fact that you don't have any percent of restaurants because the 20 percent you do have has a huge amount of utility. As soon as the consumer started booking any reservations through Open Table, the system exploded because all of a sudden the restaurateur is saying, oh, they're filling the seats. So that case was coming for the tools, but then stay for the network. And so the sales rep that initially in San Francisco and New York sold three restaurants a month, the same reps in most cases, seven, eight years later, we're selling 20.

I mean, they're ripping off one a day because all of a sudden, I have got all these tools. They're kind of interesting. I'm going to fill your restaurant. And each person I'm going to put in the restaurant is going to bring 50 bucks. And your margins on that 50 bucks is 30 percent. So you're earning 15 bucks and you're paying us a dollar all day long. And so that then, you know, the flywheel goes. By the way, that was city by city.

So the fact that you had sufficient restaurants in San Francisco meant nothing in Miami. You started again in Miami and in Tokyo and in, you know, in Munich, and it's a slow roll-up. Do you actually advise people then to, like, think about their strategy differently, you know, to go, like, because I would think people would stall if you can only get a local but not get global. Like, do you, how do you sort of balance? I mean, I feel like there's a pressure these days to, like, go global fast. How do you make those tradeoffs and decisions? I mean, different marketplace have different characters that the local ones are really hard and really slow. But if you build them, they can be really valuable and really tough to overcome.

So, you know, it is virtually impossible to displace open table at this point. It's the strongest. I've worked for some great network effects businesses, you know, eBay, PayPal, Airbnb, you know, of Pinterest. It's virtually impossible to displace open table from a restaurant right now. But they had to go market by market to do that. I mean, eBay was instantly national, almost global. Because if you were a take, there's 10 antique collectors on the service. It didn't matter.

one's in Michigan, one's in Miami, one's in Berlin. If they're collecting the same antique, there's utility there. And so that one was able to spin up much more seamlessly. The notion of the local marketplace, which I know some of you in here are doing, it has that extra complexity. I also think the question that you have to ask depends on what type of business you're operating. I think initially, the hunch is always, let's go global, get as many users. But I think the question you have to ask is, can you truly measure your product market fict if you are not close to those consumers or you don't control the zones you're in? So, for example, WhatsApp, right? It went global really fast, but it didn't start that way.

In fact, Jan used to hang out a lot with the Russian community in West San Jose. And the first feature of WhatsApp was literally just to show status, saying, you know, he just said, why can't there be just a status message on the phone that shows that I'm at the gym so that someone won't call me? And so he rolled out that at the West, and he told, I mean, his focus group was literally the 50 people that he used to hang out with over dinner on Fridays in that community. And they would give him feedback on how they're using the product. Quickly, he realized that they were using it to send messages to their family in Russia. And he was like, well, that was not how I designed this app for. But he sort of,

of built features around that to make it more seamless. And that's really how WhatsApp started. So, again, counterintuitive, because you see their growth and you're like, wow, they went global really fast, which is true, unlike, you know, like say Airbnb, which took time to build critical mass. But even then, they actually, in the initial days, were so focused in their small control groups to make sure that they got the right product market fit. Both of those examples that you guys shared with Facebook, OpenTable, those are familiar behaviors for people. What about a case like Airbnb and the early days where there is no such thing as the so-called sharing economy.

Like, it's actually a completely foreign concept to say to someone, yeah, you can stay in my room. Like, that's just, that's crazy at the time. How do you think about, like, what did they do there? And how do you think about other businesses that have that kind of a situation or there's anything really new and they're trying to build a network effect? Yeah, I think with Airbnb, and it's a classic case of a marketplace with strong network effects where they did not have any vital growth, right?

to that because forget about renting someone's home. You would not let, I mean, something more personal than your phone is your home. They see your bedroom. They see your restroom. You're not going to let a stranger in. So imagine how a company like that could take off. So their early days was really, really hard. But their biggest value proposition, it was, you know, that they could offer homes at much cheaper rates, especially when there were big conferences or events in the city and all the other hotels are sold out. I mean, that's literally how the concept was born, right? They rented their own place, literally rented their own airbed,

and they said, well, why can't we spend this into a company? But, you know, they did that for their own home a few times, but it did not take off instantly. So they really had to use traditional marketing and good design to sort of build that growth hack. So simple things they started doing where was they started targeting all events in various places, and they would make specific advertising in those. those areas saying, oh, is the hotel sold out? Why don't you try air bread and breakfast?

In fact, even, you know, in the early days, the story that goes around is that they, you know, they found it very difficult to raise. So they actually remarketed their serial boxes as Obama and McCain serial boxes. And they actually use that money to sort of bootstrap the company. So it goes to show during the election cycle. During the election cycle. So it goes to show, you know, the different steps that they needed to take to bootstrap. It almost took 36 months for them to build critical mass and see network effect. So it was not easy. But once you build that critical mass, you have huge defensibility. So one of the founder stories, it's so entertaining and almost always so patently false is the overnight success. I had dinner with Brian last night. He's been at

Airbnb nine years. I mean, you know, it's kind of like, so, you know, in the last three or four are gone really well. You know, the first, Ben Silverman at Pinterest, nothing happened there for two, three years. Nothing. He just kept working. for product market fit, you know, with his dozens of users and then 100 user, whatever. And then he got it and, you know, figured out the hack to get critical mass of content that then attracted the users. But this overnight success is rare in our experience. So I think we should drill down then because if I'm in those founders' shoes, how do you know to hang in there? Like, what are the signs, the metrics? Like, okay, I'm going to just stick around

for another 10 years or I have some sign here. So how do we tell? Let's walk through some case studies. just to have the data because we use a lot of words. I mean, let's look at the numbers. So this is the first example we're going to share is actually Facebook. And the reason we used Facebook was to tease out the difference between viral growth and network effect. Right. So this is the chart usually startups show at the start like, hey, we're growing really fast. You can see that we reached 800 and 800 million plus MAUs in a very short time frame.

So the chart looks great. But this is what we, and Facebook did this with pretty much zero customer acquisition. cost, right? They didn't spend a diet. But this is what we would call a speed of adoption, which is growth. This does not tell us whether that is a network effect. What tells us whether that is a network effect is actually this, which is, even as they kept growing, their retention, so this chart shows daily actives, divided by monthly actives, and you can see that it kept growing from 45% to 57%. And Facebook was actually one of the first platforms to actually do that. Because usually you've never heard where, oh, you increase users, do you actually increase usage?

That was not the norm. And so that's the sign of network effect. And this metric is different depending on what type of company you are. For a marketplace, like for Airbnb, you would actually look at the number of guests that booked rooms and how that trended over time, right? That would be the sign of the network effect. And we wanted to show the other example. Angry Buzz, they have a similar curve. They did, you know, lots of downloads.

and if you looked at the user graph, it was high. So they had viral growth because people were playing the game and they really liked it and they talked about it. So they may have acquired all those users with zero customer acquisition costs. But did they really have a network effect? You know, not really because if I'm playing the game and you're playing the game,

that doesn't necessarily mean I'm getting more value just because you're playing the game as well. And we just sort of find the gaming world you do see where, you know, new games come. Like the next angry words was, you know, candy crush and so on, right? So I think the key is really to tease out, is this vital growth or is this network effect? The last example I'd share is with Medium.

It's one of our recent portfolio companies, and so we wanted to include one of the case studies, which was slightly early, because it's not easy to, you know, sort of see signs of network effects. So Medium's user growth chart is quite good. You know, they're adding a lot of unique visitors and they're acquiring them with pretty much zero customer acquisition costs. But how do you actually say whether there is a sign of network effect? So what we wanted to find out is, well, how are people finding medium articles? Because if I discovered Medium through Twitter or Facebook, then if I really liked it that much,

am I coming back to Medium directly, or am I still finding those articles through other sites? So one of the things they explicitly measure is how much, you know, people spend a lot of time reading articles on Medium, but how much of that came direct? directly from Medium versus through social, which is Facebook or Twitter, or even other channels. And you can see especially for the non-viral posts, which is what they call the tail, because the top viral posts will always be viral. People will refer it everywhere and it goes. But for the non-viral post, Medium is actually the biggest source for the traffic,

which was, you know, early indicators that they are able to match. It's like a marketplace. They are able to match the readers to the right writers. And so those posts are also getting attention. and it's happening directly on Medium. So the question you have to ask for network effect is, depending on your business or company that you're building, what is that aha moment that gets the user back to your site or your app without much effort?

And the other thing is, I mean, these are all macro reads. You know, we're trying to make that call. You're trying to make that call early. We're trying to make that call early on your companies. And so you're trying to tease out what might suggest that. that these macro things will kick in over time. And typically that's micro things. And so in OpenTable, the entirety of my diligence on the company was show me San Francisco. You've been in San Francisco the longest. Show me how the key metrics are going in San Francisco. And so it turned out

sales post productivity exploded over time. The number of reservations per restaurant exploded over time. The percentage of those reservations that came from the Open Table website versus the restaurant website changed dramatically over time. The, you know, the, I mean, Just the attrition of restaurants went down over time because, you know, the company's providing more value. So it was pretty obvious early on at OpenTable when you took the micro thing, you know, that, okay, San Francisco's working. And that gave, before you even got there, the management team and the board, the confidence, okay, we're going to step on the gas and start opening tons of new markets, even though if they took years to turn profitable, they said, okay, this case, if you build it, they will actually come. Now, different, you know, so we go into markets, you know, Zuckerberg was pretty obvious on Harvard it was working. It wasn't necessarily obvious the business was working. But if you went into your, you know, your petri dish or your test tube and said, okay, I made it work there. I made it work in the Russian community in San Jose. Their behavior is showing increasing returns to scale. And so the best advice I have is try to, you know, and typically both to launch and then to measure is run the small, if you can, run the smaller experiment,

validate that it's working for you. If it's working, tell us. But that, I mean, that really is how you do it. And eBay, the early thing was collectibles. Collectibles just exploded on the platform. Because the needs of that community trading in the analog world were so horribly not met. And it turns out the venture guy who made the investment at eBay

was Bob Kegel at Benchmark Capital. Bob, who happened to come from Michigan, and there's a craftsman in Michigan who does really beautiful, and I'll remember that it, decoy ducks for hunting. Oscar Peterson was the craftsman's name, because it's also the name of the pianist. That's the only reason I remember it. Bob went on this black and white website that also had the Ebola virus and Area 51 on it, three things on Pierre's homepage, and search for Oscar Peterson duck decoys. and found like five, and he goes, I spend my summer driving around Michigan looking for these. I went online and they're five here. And so, I mean, he was in, that literally put him in the deal.

But, you know, it's those little early, you know, you're seeking those early signs that, you know, the behavior is happening at a small scale. If it's having a small scale, you probably can replicate it on the larger scale. It's not having on the small scale. You have to, you know, you got to figure out how do you get product market fit? And that's where if you're thinking of going to work at Yahoo, you know, for a steady paycheck, you know, if you can't get any of the small experiments working, that's when you might want to start thinking about it. I want to push back on one thing, though, which is if you're in San Francisco, it feels like what happens in San Francisco is not going to happen everywhere else. It doesn't stay in San Francisco?

No, exactly. It doesn't find Vegas. Exactly. So if you're building a business in Silicon Valley where you do, we are in a little bubble. in a lot of ways. Like, people are doing very unique things. And I think that's why I'm glad you brought up the example of, like, the Michigan example, Pinterest. I'd love to hear your thoughts on what that means for figuring out whether there is a network effect. So to amend it a little bit, in local businesses, we typically look that they can replicate it

in a market outside of the market they live in, because that kind of suggests scalability. So if someone in San Francisco, you know, a great example, and I don't think you would upset if I use it, Cherry. Cherry worked great. I mean, the cherry car washers, it was on-demand car washing. They were up and down this strip outside of Sand Hill Road every single day. There'd be like three cherry guys washing V.C.'s cars, a little embarrassing. And then you're like, okay, can I see this working in Minneapolis in the winter?

And, you know, you're just kind of like, oh. So, I mean, that was an example. Okay, we want to see it, you know, replicate outside of an environment where it's, you know, either you know, uniquely, economically affluent, or highly convenience-focused and things along those lines. But yes, we typically look in local businesses. They can replicate it outside of the initial market because that then talks to scalability. Right. I mean, a city is the original network effect. So there is a certain density there. Well, speaking of the trend side of things, we talked about the sharing economy. So let's talk about examples like ride sharing and crowdfunding.

And those are new technologies. I mean, they haven't been around before. Or maybe they've happened and now smartphones are here to sort of give us new behaviors around them. Do those trends have network effects? Yeah, so I think I would combine, you know, I would call right sharing, even food delivery, all of those I would put them, like the on-demand category in general. I think it's a new phenomenon after mobile, which has made it really easy for consumers. So the way to think of it is, so if you look at right-sharing services, right, it's, you know, 10 times better than taking a taxi cab.

The product is way better. use it, but it's a city-by-city rollout to a large extent like Open Table. However, the difference, I would say, in a right-sharing service, it's a point-to-point service. In Open Table, you offered the tools to the restaurants, and then, you know, the restaurants used it for the tool, and then you brought in the network, which is the diners, and, you know, you match sort of the two sides. In right-sharing, it's really point-to-point. So I would say the general right-sharing, we think it's more supply-side economies of scale. And what that means is, The more drivers you have on the platform, you can make sure that you get a good quality driver within five minutes.

But that's where it's tense, right? So it's almost like the Amazon first party, if you wanted to think of it, which is like it's still a good business. We just call it as different, but just, you know, supply set economies of scale, which is scale economy versus a network effect. However, if you look at lift line and UberPool, I think those could have network effects because think of this, you're a writer, you want more writers using those. Because then you can share a right to San Francisco from Palo Alto and therefore have a cheaper right. And you want more riders every place. Maybe you travel to New York and you want people to be using left flying an Uber pool there as well. So I think there is a subtle nuance is what our view is. I mean, talk about hacks. I mean, Uber's early hack was they're going to subsidize enough cars to drive around a city to provide a minimum level service that the consumer found attractive. And then as consumers find it attractive and start booking more sites,

they have to subsidize left, and ultimately the subsidized disappears in the city. Instacart's hack was you can order from any grocery and they'll go get it to you, they'll mark the crap out of it. But early on, they'll be showing those grocers that, hey, you can get some pretty good demand for people who want to have groceries delivered. They went back to the same grocers and they're cutting deals. And now I believe, well, over half of their entire deliveries are not marked up. They're through deals with grocers like Whole Foods.

But the hack to jumpstart it was, okay, I don't have. have time. I don't have the time or credibility to do those deals with the grocers until they experience the flow of order. So the hack was, okay, I'm going to mark it up. And then I can afford to do it for a while. And then as quickly as I can, I'm going to try to leverage the volumes into deals and then let me provide a better consumer service. So every business we see has an attempted hack. If you come in and say, oh, I don't really know how this, you know, I'm going to fit the chicken and egg, you know, not the best answer. You have to try to figure out how can and get this going. Which side starts? What is this? You know, and, you know, the deck lays out a whole

bunch of different approaches that different companies have used to try to solve that one fundamental problem is in a two-sided network who shows up first. Besides this network effects deck, you guys participated in the post we wrote around 16 metrics and 16 more startup metrics. I think it's worth bringing that up because you're talking about discounts and subsidies as part of a model to create that hack. And then the other thing is in that same post, we talked about the difference between paid versus organic marketing and the value of those. So if I'm doing a network effects business and you see as an investor like, okay, well, you've done a lot of paid marketing versus organic marketing to get this network. Is that a bad thing? Is that a good

thing? I mean, how do you tease it apart? I think that it's really, I mean, there is no definite answer. It really depends. So I'll take the example of Airbnb. You know, it was very difficult to build the critical mass initially and it was a strange concept for them. But on the Demand side, they were able to hack with traditional marketing and, you know, they had really great hacks around how they bootstrap demand. So they were really not spending much on demand. You know, it's contrary to everything we know because travel, people travel only twice a year. How do I do it? But they were really good at doing that in the initial days. But supply was really hard. You know, how do you get a host comfortable to list their home? How do I target the host? So they did end up

spending dollars on it. So it's not that we are, you know, it's not that sometimes you may have, have to figure out, you know, there are lots of papers on this, on marketplaces, which is what is your money side and what's your subsidy side? I think for Airbnb supply was really hard. So they spent dollars right from the early days to acquire the host. But in those cases, you have to ask yourself, what is the return on investment if I am paying to acquire the user, right? So what is my auto value per transaction and what's my lifetime value? If you do have good autoI economics, like the CAC versus LTV, then... I'm sorry, just say CAC versus LTV, so the customer acquisition cost versus the lifetime value that you're getting from the customer, then you can spend because there are certain sectors where you have to.

Travel is one where you have to because the velocity of use case is really low. But you just have to be sent, you know, the thing that you don't want happening is where you're paying lots of marketing dollars to acquire users. And then when you sort of dialed down your marketing dollars, the growth also slows down. Because that means you've been spending a lot of money to acquire. and drive the growth, but the company itself or the product itself is not sustaining that. So that's the thing we really try to tease out is if you're doing paid marketing is, you know, why are they doing paid marketing? What's the ROI, a return on investment from that marketing, and is it truly showing retention? One observation, a lot of the best networks that are

platformed here aren't spending anything on acquisition. I mean, it's just, it's kind of an interesting thing in marketplace. Facebook, no, WhatsApp, no, medium, no, you know, just open table, know, just they're not spending on acquisition. The model is delivering it. If they are spending on acquisition, eBay, I was the biggest advertiser in the world, on digital advertising in the world for a number of years. They were spending. But what happened is the more they spent, the more valuable the service was to the user. So it was a component of the P&L. But those users demonstrated network behavior. And the contrary example is, okay, an e-commerce business, you can pour money on it. You can grow like crazy. I've been an investor in a few. I've been an investor in a

of them that actually are no longer around that grew like crazy. But, you know, the only reason you're growing is the marketing spend and when, you know, you know, and who said when you pull back on it, they don't come anymore. There is no network effect there. It was just a growth effect. And so, you know, the key is how do you tease out that? So it's not that you don't spend. It's surprising how a lot of the most valuable network businesses didn't spend virtually anything. They've, they figured out the hack. Well, I'll do one last question. Then we'll open it up to everyone for questions. It's a question that probably, you know, your entrepreneurs hate getting from their investors, which is, or maybe they want to know, which is when to monetize. And how do you

figure that out? I should say that as investors, you know, we're not going to push anybody to monetize because I know it is sort of a debate everybody has. It's more you should think about when to monetize. And what types of monetizations do you have in the plan? But we've seen companies at all stages, right? So Facebook didn't monetize for the longest time. They were building a social network, focused on users, and now, you know, they monetize quite later, but they're doing it beautifully now. Versus marketplaces, on the other hand, you do tend to see them monetizing a little earlier. Why is that the case?

Because there is a transaction that is flowing through the marketplace already. So let's take offer-up our portfolio, for example, right? There's a lot of transactions going on. People, the sellers are listing their items. Buyers want to buy the items. And so there is a transaction happening. Now, if you actually turn on payments in marketplaces, it actually improves the overall process, right? You don't want to, you know, meet the seller and then message them and just say, let's meet outside.

Versus you actually want to be able to send the credit payment directly so that the transaction is done. So it improves the experience. And so you can actually turn on monetization sooner and you tend to see that in marketplaces. So I think it really depends on the stage of the company where you are and what sort of product market fit you have. have and when you can turn monetization. I mean, I think the business models are really different. I actually think there is a meme with a false dichotomy that it's either growth or monetization. And, you know, it's just a little bit like, okay, you've got 100 people in the company and you can't have, you know, any on, you know, you have to pick one or the other and no one else is allowed

to cross over. You're just a little, you know, it's there. We've had a couple of companies that shifted in the current capital environments and the uncertainty there, shifted from growth at all costs to, you know, kind of smart growth mode or what do you want to call it. The amazing thing, the profitability improved dramatically, the growth in a lot of the cases didn't slow down. And you're just sitting there kind of like, huh, you know, just like, oh, wait, you went from spending all that money and not spending much money and it's still growing the same. So for me, I worry sometimes it's a false dichotomy. I think in a marketplace business, if there's money flowing through it, you should be able to demonstrate you can capture some of it early on. I mean, if not, I'm trying to

figure out if you're actually adding value to both sides of the marketplace and why wouldn't they be willing to pay? So a marketplace without a monetization model is a little bit like, oh, really? You know, something like Facebook, you have a little more of a call on or Pinterest, you know, they're a multi-billion dollar value pre-revenue business. You know, the question is, yeah, there is a legitimate when question. But yeah, I mean, typically, we think there's more reasons to do it early than not to do it early on average. So no false dichotomy. Let's open up to questions. So when you think about the open table business model,

and originally you were selling software into restaurants, I'm guessing you had competition that was also trying to sell similar software. At what point did, and that one must have been like a feature war and classic competition, at what point did the amount of reservations that you were driving become enough of a hook for why restaurants would always choose you guys for the software, that it didn't matter that a competitor had a certain feature or, you know, a couple small things. It's a fantastic question, and it's got an unexpected ending. There were three companies trying to do exactly the same thing at exactly the same time in

the late 90s, early 2000s. One raised a round right before financial Armageddon came in, and a nuclear winter set in for literally. It was a half decade in terms of funding new companies in the first half of the 20,000. Open table got raised enough money and spent it so slowly. They were able to survive. the two direct competitors didn't. OpenTable ran unopposed for five years, which is the only reason they were able to get to their network effect. eBay, when I was there during that time, enjoyed the same thing. I mean, there was no company formation for five years.

So, you know, we were nascent in a lot of markets and a lot of product lines. And we just had, we were unopposed for five years. So when people get scared of the current market, it's like, yes, be scared, but also there's advantages in it. If, you know, if capital dries up to your competitive set and you've got. You know, you've got access to capital or whatever it is. It actually can be an attribute in there. But, I mean, if OpenTable had, actually, Grubhub just went through this. They didn't have nuclear winner, and they ended up having to roll up three or four companies to win the market. They seamless, campus foods, you know, they camp, campus foods set out campus, seamless one,

a couple key markets like New York, you know. And so Grubhub's a roll-up, a home away was a roll-up. You know, there wasn't a source of differentiation. that was meaningful other than local geographic, and so they pieced it together. They went out of business, yeah. Yeah, and it's an open table, ran unopposed because they couldn't make payroll. And then that was incredibly informative to that company because the network was so slow to build. There was the slowest network I've ever seen to build.

It's amazing. It worked. It wouldn't have worked without the capital environment and anything like its current form. And we have some local investments. They have to go faster than OpenTable went. I mean, InstaCard's going fast or offer-ups going fast. or just get they're more virtual. They didn't have, they didn't have to physically convert each restaurant one at a time. And the average owner of restaurants on Open Table owns one unit. So it literally

door to door trying to do it. So it's a business that wouldn't work today. Hi. I have a question for you that I think is maybe about branding, but maybe not. When you were at Open Table, were you ever tempted to offer a private label version of Open Table for a segment of the population that really wanted to be special. That's kind of what we're facing right now, where we have an opportunity to private label to a certain population that thinks that they should pay 10x for the same accounting because they're just really special in some way.

Yeah, yeah, we have special. So one of the things that made restaurants and attractive marketplace, when I first met with my predecessor Thomas Layton, the first thing I did, this is an ingenious model, where else can I apply it and start my own company? That didn't work. What made Open Table interesting was two things.

One was the incredibly high degree of fragmentation, which was a bug when you're trying to aggregate it. But once you've aggregated, it's a feature. And the second was lead gen was a really important part of the restaurant experience. Most people don't want to eat at the same restaurant every night. And so Open Table would put new butts in your seeds, butts who'd never been there, that you're trying to get. So when we started with mom and pop, started winning the chains. The good news is the chains would only have 100. restaurants, you know, McCormick and Schmidt, Maggiano's. There are a couple other M's in there.

I mean, Mortons, they only had like 100 restaurants, but they would say, like, we're Mortons, and they demanded. We held firm on two things. One is we branded everything. And so if they didn't want it, they didn't get it. And it was such a unique service. They couldn't afford to build it on their own at 100 units. The other was we forced to redirect from Morton's website to the open table, Morton's page on Open Table to create the brand awareness. And that forced off, that forced redirect, the fact that we converted people searching on Google for Morton, Chicago into open table users is the open table growth hack. I mean, that was the open table growth hack by far. Because the supply base was so fragmented, we were able to pull that off. And, you know,

enabled to do that. I'd contrast that with Fandango. Fandango basically has probably be six or seven customers. There's a large movie theater chains. It's AMC. It's whoever. And those guys could build their own thing. And every once in a while, they pull out of Fandango. AMC two years ago, I think, pulled out a Fandango. And they could because they had the scale. I'd go to AMC, Redwood City. If they're not on Fandango, I'll go to AMC. And it's like, oh, I can buy it through them because they have that level of critical mass. So part of it is the market you're at? Do you have suppliers have power? Or does the marketplace more? yeah then that's that sounds pretty fragmented to me so and then branded marketplaces we would

value much more highly than white label because white label becomes you're selling a utility you're selling software and the marketplace standpoint you're selling customers hi do you believe in the idea of network effects um with brands so for instance Nike the more people that think Nike's cool the cooler it gets the more people think Nike's cool Yeah, I think we've had, again, you will hear differing views in our show. I think we all have different opinions on this one. We've talked about this, whether brand has a network effect. And we actually, and it's funny in the deck, we actually, that's why one of the charts that made it there was, I don't know if you've seen it, but there are three types of laws.

We call it, which is Sarnov's law, Metcalf's law, and Ritzlaw. So for brands, we actually think it's more like Sarnov's law. You know, some people argue, yes, it has some network effect, but not directly incredibly. value to the user, which is why the value is not as steep as Metcalfslaw, for example. But the other side of the argument you usually hear as well, that's like, you know, it's almost like Angry Birds, you can argue. They didn't have a, I mean, they died on and they go. But when you're playing the game and the game is hot and still having conversations about it and, you know, but you have to start, the question is, is that really value and sustainable value?

And is that something we would call a network effect? It's a really intriguing question. I think it's a, I think it's a different. source of value and a different sustainability. You also run into dis-economies. Once everyone's wearing Nikes, people stop wearing Nikes. So you're kind of, I worked at Disney when it was a long time ago, it was very, very hot on the consumer product side. And, you know, we couldn't sell enough. And then five years later, you can't give it away to the same audience. And, you know, just kind of, you know, Disney became, you know, I was there in the, you know, Lion King and all those. And then Disney animation hit a, hit a pause button, Pixar clean their clock. And you could, you

You literally couldn't give away Disney stuff. So it's, you know, one of the most valuable brands in the world has done. And Nike's done a little bit of it, too. They're hot and not back and forth. So, you know, the brand and the fashion overlay, I think, I mean, I think they're unbelievably valuable commodities. I'm not sure I'd call it a network effect. Going back to your statement about comfort the tool and stay for the network or the

community, it seems like some of the strongest network effects actually happen when people not only use a tool from the top down. So Nike branded shoes or an app from Uber, but when customers or users give back data or create something, which then adds to the network that we benefit in a stronger and more coherent way when we have a creative tool or something that someone, even if it's a post on a Facebook wall. And I guess I'm curious to know if you think that there's room for more creativity there, that these tools are actually better at creating network effects or fast. or it looks like Pinterest took a long time, but I'd be interested in that perspective. When Pinterest started working, it went very fast. But, you know, so it took a long time to get the product market fit to make it work. There's some fascinating ways entrepreneurs come in with the, you know, what is the value that

more masses come? We're seeing more and more data network-based effects. I mean, just, you know, companies are saying, like, I'm doing this to get the data. And with the data, I can provide a vastly superior service. and here's Y, A, B, C, D, you know, the hottest job right down in the world is the data scientists. Both my kids are in college. I say, you sure don't want to get a PhD in machine learning, data science. But, you know, we're getting more and more things coming at us with, you know, either a corpus of content or a, basically, a data, we've called a database network effect.

And there can be some really powerful ones where, you know, the aggregated data in the system, Pinterest should be able to put better content in front of you than anyone else because they have so. many more signals and so much more content. Yeah, it could either in the form of personalization. It can also come in the benefit of B2B, I think, a lot more. For example, a lot of these SMB players use shipping, right? And they can't negotiate great discounts like how Amazon does because they consolidate. But there is a shipping provider that is sort of servicing all of them, a tech company, and it's collecting all the data. They can actually give better information back to the players in terms of, hey, if you're in Dallas or in this microzone, and use this or like, you know, they say it's priority, you know, two-day priority, but it

actually goes overnight, 80% of the time so you can do money-saving. So we're seeing interesting startups that are sort of tackling those problems using data that customers are contributing because the transaction happened on the platform and using that to sort of give recommendations, better recommendations on who to use. I was really curious if you had any advice. If there are adverse socioeconomic or political or geographic effects or even seasonal effects that could adversely affect the network effect. I realize I'm saying effect quite a bit. Is there any advice on how to guard against or at least plan or buffer for these adverse effects as they're coming in that could actually disrupt this trajectory?

Are these like seasonality or can you get any? I think seasonality or probably the best example is 9-11 where everything kind of took. a crash. And if you started a company that day, odds are you would have very hard time getting customer acquisition. Same thing for, you know, at the onset of the Occupy Wall Street movement. If you were trying to sell something to affluent individuals probably wouldn't have as much success. I mean, I was managing a network of ex-business during 9-11, which was eBay. You know, our volume literally just dropped 40, 50 percent overnight. New York. user stopped completely. It was a big divot in the business for two or three or four weeks,

and then everything popped back. We never regained the lost users we had during that month. They never came back, but it recalibrated. I was managing open table during the financial crisis meltdown. We did our org meeting for the IPO in mid-August, 2008. We told the banks on Thursday, before the Monday Ord meeting, who got the business? Lehman Brothers didn't get the business. They went bankrupt Saturday on that news. Merrill Lynch did get the business, so they traded the Bank of America on Sunday. Over the weekend, the business, the entire restaurant consumption market dropped 15 percent because of the panic, and we worked Monday morning. That one lasted for like 15 months, but very few of the startups are long term. You need capital to pursue it, but just

You know, if it's a month, you typically, you know, it's not going to completely bury it. You know, if the shock's more than a month, it's a hell of a shock. And so it's, I mean, I think the externalities, other than, you know, I'm thinly capitalized and don't have a lot of time to run the experiment. Other than that, typically you can understand it outside of it. Thank you guys for your great questions. Yeah, that was good. Thank you. And let's hang out and talk some more.

Thanks, everyone. Thank you. Thank you.