Starting and Scaling Marketplaces with Eventbrite GM Brian Rothenberg and Casey Winters | Greymatter

Hello everyone. I'm Casey Winters, growth advisor in residence at Greylock. At Greylock, we're always trying to learn from practitioners about how to solve real problems for startups. Now we're bringing these practitioners in to Grey Matter, our podcast to jam on how we solve real problems so others can learn as well. This week I'm excited to welcome Brian Rothenberg. Brian, why don't you introduce yourself?

Hey Casey. Thanks for having me. So I've been working on marketplace businesses for about 13 years, ranging from various positions from founding my own marketplace startup to working on product to working on growth and marketing. Most recently I've been at Eventbrite for about the last four and a half years as the first VP of growth and more recently transitioning to a GM role, responsible for Eventbrite's self service business and increasingly thinking about how we evolve to a marketplace. I've also spent time at TaskRabbit as co head of marketing there. And before that I started my own VC backed company called skillslate, which was a local services marketplace. And then prior to that I was at yah, who is a PM for their local marketplace businesses.

Cool. So this week we're gonna talk about Brian and my favorite topic which is starting and scaling marketplaces. Marketplaces can be the hardest businesses to start and to scale, but they're very lucrative once you are able to scale them. Just look at the success of Airbnb. Usually the first question that Brian and I receive when we're talking to marketplace founders is is how to solve the chicken and egg problem. So Brian, why don't you define the chicken and egg problem for people that haven't spent as much time thinking about marketplaces and how you think about solving it?

Yeah, it's the classic marketplace question of how do you drive buyers to a store with no products on the shelves, essentially. So you have to typically either have supply before you drive demand, or you can start with demand first, but you have to have a reasonable way to get supply to create that virtuous cycle that is a marketplace. I always joke that marketplace businesses are a love hate thing for me. They're so difficult to start, but once they get going, they're so powerful. It just sort of keeps me motivated to work on them. But it's really tough to get the flywheel going.

So when you think about supply and demand, like the classic response, people say, oh, you seed one side, that's all you got to do.

And it sounds so simple.

Yeah. What does that actually mean? And then when you think about which side you need to seed. How do you make that definitive answer to say, okay, I'm going to do demand and that's going to drive supply. I'm going to drive supply first and that's going to help me drive demand. How do you think about that?

So I think more of my experience has been on seeding the supply side of the marketplace, but again, it can be done both ways. At my own startup Skill Slate that I co founded, we tried to hack the supply side. So we were trying to roll up the long tail of local service providers. More of your single mom and pop providers that weren't yet on Yelp didn't have their own websites. And we created what we thought was a novel hack. We crawled the web, classified listings, people posting online, advertising their services, and then we layered on top Amazon's Mechanical Turk and some of our own technology to help extract the data, build structured profiles for these service providers. And we effectively invited them to claim their listings that they didn't even create. We also pushed them to the web so that they could be discoverable via SEO. And both means led to a lot of those profiles being claimed and helping to seed that initial marketplace. So that was one hack that we did. Another approach is something I call the Trojan horse approach, and it's not quite as malicious as that sounds, but it's really about creating a solution for the supply side, like a standalone product or service that meets their needs before you even have any demand. And so an example there using Eventbrite. So Eventbrite started as a SaaS ticketing platform for event organizers. So. So it has utility even without any ticket buyers that the marketplace is driving for them. It's a tool to help them manage their business, help them manage ticket sales online. But as you roll up the market, as you get more and more of these event organizers onto that platform, you create the supply, the inventory of events for people to come discover things to do in their local area, and then you can start to layer on that demand side over time.

Yeah, I think OpenTable is another classic example of the Trojan horse, right? Yes. Allowing reservation software for your business and then being able to sell that reservation software, that reservation inventory to people looking for where they can book tonight. Yeah, absolutely.

And I know on the OpenTable example you have a lot of experience with the local marketplace approach. And so that was another example of super hyperlocal approach. They focus on a few cities and really work to build supply and demand density within those cities.

Yeah, that's a very good point. I think when you look at most marketplaces and how they scale, it's not just like, oh, we seed one side and then quickly fill on the other side. It's you do it at a very, very hyperlocal level. So grubhub started in Lincoln park in Chicago, and that was the only neighborhood that had actually had a product for a while. And it's interesting to hear you talk about supply seeding strategies, because grubhub actually was a demand seeding strategy in the very beginning. We basically picked up menus for all the restaurants we could find, put them online, and then when people were searching for delivery menus, we showed up and then we were able to then go to restaurants and say, hey, we have all these people that are looking for food delivery. You should advertise with us so you can get in front of them. And that was the original launch strategy and it did work. But we found, similar to you over time, that a supply led strategy actually was more effective. So now with grubhub, when it launches a new city, it will send a couple salespeople to that city called Paratroopers, and they need to come back with a certain number of contracts for restaurants. We'll put those contracts up onto grubhub, create those listings, make sure they accept online orders, advertise them on AdWords, create SEO pages for Linkin, Park, Thai or whatever restaurants you were able to sign up, and then start bringing in demand from having the menu data directly from the restaurant. And that allowed GrubHub to scale infinitely. But that first city was done with a demand first approach, which. So it is interesting to see how grubhub kind of worked from both sides. I think it worked a little bit better with a supply side approach, but both can work.

What led to that pivot of shifting from demand to supply? Was there an insight?

So grabbing menus for every city across America was very difficult. And I think you quickly learn that certain restaurants are the ones that help really build an interesting product for the demand side for people that want to order food. So if you can learn from existing services like Yelp, which restaurants are actually important to get their menus up and to take online orders, you can cut out a lot of the work. That used to be like a super manual, you know, TaskRabbit didn't exist back then, so you had to find people that were like, willing to go grab menus for you or you had to do it yourself. So from a scalability perspective, once the sales process was refined in terms of eliminating every barrier to someone signing up. You only pay when you get orders. You can cancel at any time. It's a set fee, no hidden fees. As soon as we were able to get through all the barriers that a restaurant owner would have to say no, then it became much easier to do supply. But that learning process of how to do sales at the local level very effectively took a couple years to hone you, you know, in the home market before we're confident scaling it elsewhere.

It's very related. Jason Finger, the founder of Seamless Web, was actually an investor in my startup, Skill Slate. And so we learned a lot through him. And they had an interesting approach where they started with the demand side as well. They'd go to big investment banks, big law firms in New York City and say, we're creating this restaurant food delivery service. Are you interested? You know, for your bankers that are working long hours for lunchtime, et cetera, they'd signed big contracts with those firms. Then they would go to the local restaurants in the area in each key category that mattered, like Chinese restaurants, Thai restaurants, Italian, and they would say to those restaurants, we're only going to onboard two Chinese restaurants in the area and we're going to your competitors next. So sort of they forced this competition dynamic and were very successful in onboarding both sides, which got the flywheel going.

Yeah, it's an amazing story. You know, Seamless and grubhub later became one company. But Seamless, starting with the demand of not regular people ordering food, but specifically banks and law firms, is an even more hyperlocal approach to building the demand side first. It's really interesting. Let's talk about horizontal versus vertical as well. For starting a marketplace. Obviously there are examples of both being successful. Grubhub doing something very specific. Food delivery and only food delivery. But then there's also examples of going super broad. TaskRabbit, I believe when you started went super broad. Thumbtack's been successful going super broad from the beginning. How do you think about whether you should start with something very specific or start with something a little bit more open ended?

I feel very conflicted on this question. I initially took the philosophy that going very vertical in one local market makes the most sense to see if you can drive product market fit over time. I've seen the opposite work really well too, like Thumbtack being the key example. Thumbtack and my company, Skill Slate started at the exact same time. We focused on one metro area. We were cross categories, but it was very hyper local focused. They've launched In, I think, 1,000 categories across the whole United States and they've been dramatically more successful than my company was. So I don't think that there's a right or wrong approach, but I tend to favor single category versus cross category. Part of the reason being, I think horizontal platforms, you can't be everything to everyone, and so you open yourself up to effectively being disrupted by vertical players over time. So horizontal platforms that reach some level of scale, unless you're eventually verticalizing. And I think that's a trend that I'm seeing. Like, if you look at Thumbtack, if you start to request a dog walker on their site, they literally have a very tailored flow for picking the perfect dog walker. And so I think that's a very smart approach. Start horizontal, pick off and improve the user experiences for each vertical. So that's one interesting approach that I've seen.

Yeah, I remember when TaskRabbit launched in Chicago and actually went with a coworker to their launch party, and as he was reading about the service, he was like, I can think of a better service for every one of these specific examples. Right. And that's the danger of going too broad at the start, is that there, you know, there's a better specifically focused marketplace on this thing that you can go do today. So why would you do this generic version that probably isn't as focused on it? But I think Thumbtack had a more interesting way to scale supply very easily. Like, they didn't have to do direct sales like grubhub, and that allowed them to say, like, okay, is there a first mover advantage? Just scaling this out as far as possible and then, hey, if you can get all of the supply nationally, you should probably just go do it. But if you can't, you're going to spray yourself very thin trying to sign up. In Grubhub's case, restaurants in D.C. and LA and Chicago with a very small team. So I think that's another consideration that helped them.

Yeah. At TaskRabbit, I completely agree. The paradox of choice. Consumers would come and they'd say, oh, this is a cool service. It's really interesting. And then you'd ask them, well, what are you going to do? And they literally couldn't name a use case. There were too many choices. And then if you look at, like, I'm sure many of you listening have seen the famous Craigslist image with all the competitors that picked off every link on Craigslist. If you did the same thing for TaskRabbit today, you'd almost see the Same thing. Grocery delivery. Instacart is a $4 billion plus business. Deliveries. Postmates is a multi hundred million dollar business. Like they've just been picked off by so many others. They've really retrenched and focused on fewer use cases and doing them much better. And I've heard that's worked well for them. But it's just interesting to see these players take different paths.

So we talked about kind of supply versus demand and starting a marketplace, but once you're past that path, there's like the constant calibration of should I be focusing on demand right now or supply or a healthy example of both. One of the things that I've seen startups do successfully is really try to figure out what's constrained at this very moment, what's constraining the growth of this business. Is it because I don't have enough people providing this service? Or is it because the people providing the service don't feel like they're getting enough customers? And that requires like having a deep understanding in every combination of buyers and sellers where you're at. So in something super broad like thumbtack, I'm sure they have like a giant spreadsheet that's like, okay, for dog walkers in this neighborhood we need more dog walkers, but for dog walkers in this neighborhood we need more dogs. Right. And building that up takes some time. It takes some data acumen that can be hard to build. I'm sure you guys saw some things like that at TaskRabbit and as you've expanded the categories in Eventbrite, you had a similar approach to trying to match that. Have you been able to automate that better than a spreadsheet?

Yeah, it's a really hard challenge. And I think one thing is there's the flaw of averages. So looking at a marketplace across all metros, across all categories can hide a lot of potentially fatal flaws. So I think to your point, the geography focus, seeing how is your supply and demand matching within a geo within specific categories? I think figuring out which categories matter the most is really important. And making sure you have liquidity within those categories. It might be, you know, for Eventbrite, music, food and drink and food tasting events, I don't know as examples. But making sure you have depth of inventory, there can be qualitative measures like you might have, similar to Google search quality raters. You might have people who come and score like you give them a task, find an event in food tasting that matches your needs. Can you find this or not? And so if you have qualitative measures. You can look at quantitative measures as well to see is there supply and demand matching at a hyperlocal level, at a category level, and is it meeting user expectations? And then how do you scale that out more broadly?

Yeah, that makes sense. Related to that, as you're scaling cities, and I think both of us are a fan of scaling locally at first, at least. There's also the concept of how you staff for that. Right. So when you're thinking about scaling, do you believe that you need to have like local people in every city that you're in? Do you feel like HQ can manage city growth without like full time employees on the ground? And how have you thought about staffing that in the past?

I've tried many different ways and I can't say that I found the magic silver bullet at TaskRabbit. We found it to be highly effective to send a small team, a couple people, to each launch market. And they were tasked with understanding the market dynamics, how they're different from other markets, seeing the similarities between other successful local markets and other challenges that may be, how geographically dense or spread out are the areas, what types of services are most in demand. And then they were tasked with building out that initial supply and training the initial supply on the service side. And then furthermore, the next step was typically to do local pr, which was in coordination with headquarters, layering on the demand side around paid marketing, et cetera, and putting the two together. And the team was there for several months to get the supply and demand going and to get user feedback, to create a sense of community among the supply side, to help train them and onboard them. So I've seen that work reasonably well. At eventbrite, we have tried distributed models where there are local city managers. We've since retrenched from that, so there's no perfect approach. That was a huge piece of Yelp's expansion strategy. So I've seen it work at other businesses. Do you have a perspective on that? Did grubhub?

So I think you brought up a really good point in figuring out like centralized versus decentralized is how much training needs to happen and does that need to happen every day or is it just a period of time? So for grubhub, I'd say we had a very similar model to which you described, which is we sent those paratroopers to the market to build the initial base with restaurants to learn about how the market worked and if it was similar or different to other things we'd seen. And then originally those people are also tasked with hiring a local person that would stay there and, you know, they'd spend like their first week, you know, with the experienced salesperson. Seeing how it worked over time, we found that we actually didn't need that local person and that we could do inside sales from HQ and be just as effective. And then those salespeople wouldn't over time be like a victim of their own success. We'd be like, I've signed up 500 restaurants, which means I have a very small amount of leads to go after to continue to hit my goal. Then they could just easily move to another city, which is harder for someone that lives in Los Angeles to then go do the same approach in Phoenix. But compare that to Uber, which has to train drivers on how to accept rides, manage quality, and do that basically every day because they're always adding new drivers. I think that's a very different dynamic that probably requires at least some local people on the ground all the time. And I think that is how Uber has scaled mostly with local teams. What you also mentioned that I thought was interesting is the tension that will bring up if you do keep local people. So originally those local people were in charge of supply and demand at the local level, which meant they sent the emails, they figured out the paid marketing budgets. So they were these Swiss army knives to do whatever they needed to get the liquidity. But as you scale, there are specialists that are going to be better at that. Right. The person in Chicago that's doing emails and paid marketing and, and events and vetting new drivers is probably not going to be the best adwords person you can find or the best email marketer you can find. So I know that they struggled once they got to scale trying to go back and centralize some of those functions so that they could get a higher ROI on those functions because it's an OPS led company and those people have all the power. So they probably could have thought more about how that thing would scale over time or set expectations with the local teams over. Here's the things you're going to own now. Here are the things you're going to focus on in the future. Those were issues we didn't have at grubhub because we never gave them those powers early on. Yeah, so it was basically talking about some of these tactics that both grubhub and Uber used in the early days to drive supply and demand. Let's go a little bit more in depth on what tactics usually are the most successful at driving supply and demand. In your experience, what are the things that have really driven demand at the marketplaces you've worked at. What are the things that have really driven supply? Did that change over time and were there any tricks there that you can share?

Yeah, I think it's almost a little viewed as stale or doesn't work anymore. But search is incredibly powerful, both paid and organic. And I think one aspect of marketplaces, or some marketplaces at least, is typically there's some aspect of user generated content seeding the supply side, which is very attractive for, for SEO. And then if there's an aspect of the supply, promoting their own listings or their own events like that is a sort of page authority or sorry, domain authority hack, page rank hack to help create massive link authority that can help the overall site rank. And so an example there is eventbrite. Event organizers create their listing page on Eventbrite. They typically have their own website. They link to that event page, creates a huge volume of inbound links. And then that content on the Eventbrite event event page is typically largely unique. And so it's very discoverable. You can create structured directory pages at a local level which tend to rank really well in search. So that's one approach that I've seen work really well. Are you seeing that working today?

Absolutely, I think, yeah. Right. Search is seen kind of as old hat, but usually marketplaces are trying to organize existing demand. And where do you find existing demand is what people are searching for on Google. Right.

Tapping into other networks that huge.

Yeah. So I think in general growth strategy is tapping into huge networks. The two hugest ones are Facebook and Google. Right. And Facebook is not as good at giving you free traffic as Google is. But you brought up perhaps an even more interesting point which is getting supply on your side to help grow you as a marketplace. That's something it took us a long time to figure out at grubhub. But once we did, we were able to tee up a bunch of strategies around that. Create websites for restaurants that they could promote because only 4, 40% of restaurants had their own website. Then help them get it in Google Local so they rank better for those terms. Give them flyers that they can hand out that send people to those websites because the order is going to be fulfilled by grubhub. We're able to go down the line once we have that insight and do so many more offline things as well as online things. And I think Yelp beat us to the punch with kind of people love us on Yelp. And that's fantastic. That being this physical evidence that brought People from offline to online and later grubhub was able to tap into that, and restaurants loved it because we're, like you said, providing more infrastructure to help them scale their businesses, kind of in the Trojan horse way. And I still see a lot of marketplace miss that opportunity. And obviously something that you keyed on very early, But I think GrubHub is better late than never. But it's something that, because it's offline, because you already are like, this person is like someone I'm trying to keep happy. Why would I ask them to do work for me? It's like they're willing to do the work if they're going to get more business, which is kind of an intuitive finding when you think about it. But most founders don't start there.

Yeah. I mean, we had TaskRabbits early on, wear the bright green TaskRabbit shirts when they were out doing services. And that was a huge awareness driver. Offline has been huge for TaskRabbit. Another example and anecdote that I've heard is Instacart. They have started and stopped at various times having the bags that they give out. And I heard that you can't track easily what the impact is, but they saw their numbers decline when they stopped, including the bags and the deliveries. So there's all this stuff that we can't measure. But offline can be so powerful.

Yeah. At grubhub, we had kind of a similar thought process of there's this journey to an order from a person starting on Google or opening the Grubhub app directly to placing an order to 30 to 45 minutes later, receiving an order. And as we got more in depth in understanding that journey, it was about how can we thoughtfully insert GrubHub into more of that experience than currently exists. So creating bags, creating a car that's grubhub branded, whatever the case may be. And you have to see which of those make sense and can you actually measure them in an effective way, which it sounds like. Instacart got some macro data, but it was hard to attach a CPA to is really important because usually marketplaces are a new way of doing something. And in the case of grubhub, it's not a new way of doing something that people are going to virally talk about. People at grubhub were not like, yeah, you know what I did on Friday night? I stayed home and I ordered pizza. It was awesome. And it just makes you sound kind of like a loser. So ways that we could get people to talk about the brand or experience the other people experiencing the brand got more important over time. And then trying to figure out if you can measure that to see if the exact ROI was a challenge. But sometimes we could do it.

Yeah. So I think that's right. Sometimes there just isn't that viral mechanic built in, but sometimes it is. Like for events. With Eventbrite, for example, events are inherently viral. So for a long time, Facebook was the largest traffic driver for all of Eventbrite. And so over time, the company leaned into that, built a relationship with Facebook, has been an early API adopter to. To adopt features more quickly, et cetera. So, again, seeing that spark what's natural in the context of the business and leaning in can be a great approach too.

And I think you've also seen an event organizer bringing people to come to their event. A lot of those people end up creating their own events later. Right. How did you track that? Were there ways to optimize that further? Cause I think that's a lot of marketplaces see, supply can become demand, and demand can become supply over time.

Yeah. I think some of the most valuable marketplaces have that overlap of supply and demand, and it can be a huge lever. Other companies like GoFundMe have that dynamic. Other examples are. I'm blanking on right now, but for.

Eventbrite, like SurveyMonkey is one.

Absolutely. That's another great example. But we saw it in the data. It was just happening. And so we surveyed customers to understand how they came to realize that if they first bought a ticket, that they could also use eventbrite to organize events. And over time, we worked to understand, well, what are. Are the gaps for the people who don't make that connection and started chipping away within the product. So it was small things to start, like changing our global header from just an eventbrite logo and contact the event organizer to the eventbrite logo. Find events, create an event. Right. It's like setting up the marketplace and that there are both sides and many other things in the product. Trigger notifications, looking at the data. And we were able to more than triple that conversion rate over a couple years of effort, which is really meaningful for the business.

Yeah, that makes a huge dent. Yeah. Grubhub didn't have this opportunity. We had restaurants that, like, four years after they signed up, and at that point they'd received probably 100,000 orders from Grubhub. Like, man, I went on the website for the first time. This is really interesting. You have all these Restaurants like, yeah, why do you think you've gotten 100,000 orders from this business? But they, you know, restaurant owners at that time, you know, pre mobile, they never went online. They just kind of skipped the whole online revolution. So they're like, I have this fax machine. It prints orders. I don't know how it works, but I'm going to take it. And then later on, one or two of them might have actually ordered food online. But Airbnb is a company in the Greylock portfolio that's seen something similar to Eventbrite, where they have this natural upsell. Once you book a room on Airbnb, they're like, why don't you make your room available on Airbnb? You're not going to be here for this time period. And that's worked really well for them as well.

I've heard paid acquisition can also be a great strategy. So olx, the international Craigslist competitor, their initial strategy was there was so little competition in international markets, specifically Latin America. They would buy long tail keywords for, like, all the permutations of local cities and the categories on Craigslist. Effectively, they would start driving traffic to their site. And supply just started creating local listings because they felt like, hey, here's exposure. I should be there as well. And so it was a paid acquisition strategy where they were buying clicks for pennies, and it eventually seeded the market and then created the demand side over time. So that's an interesting approach too.

Yeah. And they've obviously been very successful and replicated that in lots of other areas. I think for grubhub, I think you're always trying to say, are there opportunities in paid where you value the inventory a lot more than the place that's selling the inventory values it? And in our case, transit advertising was a really good method for this because we started buying it in 2008 during the recession, all the national players were pulling their advertising budgets. And you could get nickel CPMs on Chicago transit. And for Grubhub, which had all of its orders between 6 to 8pm local time. Where was everyone right before 6pm local time time? They're either in a car or on the train on their way home. Right. So we just would send. We would buy like maybe 500 units of advertising, but we would send them like 5,000 and say, hey, if no one else buys stuff, just put our stuff up.

Yeah.

And no one was buying anything because all the budgets were being pulled back during the recession. So we got this, like, all this amazing inventory that was just so perfectly suited for the audience we were trying to reach. And they actually had higher lifetime values than a normal user because every time they went home, they would see grubhub again to get reminded about it, which.

Is it's like acquisition and retargeting.

Yeah, yeah, I called it, like, built in, offline retargeting to your acquisition. And I know TaskRabbit played around with paid, because I remember seeing transit ads for TaskRabbit as well. And sometimes those inequalities exist in the market and you can really leverage them. For a while, we felt like it was years before the secret got out that transit was a really cheap way to reach people. And then the prices started going up. Of course, we did AdWords and Facebook ads as well, and those meaningfully contributed. But a lot of that depends on what your lifetime values look like and what type of CPAs you can really drive. Another big problem we had, I wonder if you encountered this problem is so you have two sides of the market and you have an understanding of the lifetime value, in grubhub's case, of a restaurant and of a diner. But if you spend up to your lifetime value to acquire them, you spend double what you'll make because you're only making money once from the collaboration of two sides. How do you think about solving that problem in terms of benchmarking the CPAs you try to hit?

To be totally honest, we haven't solved it yet. We spend more from a paid perspective on the supply side, and we're just starting to scale up paid on the demand side. We're trying to think about it in terms of can we measure incrementality on the demand side and then only pay up to what we can deem to be incremental and have that be how we think about our unit economics? Have you found any ways to solve for that?

Well, as we encountered the problem at grubhub, I did some research on how other people looked at it, and I found a few approaches, and I can't really say which one is best, but it's just how companies solved it. So some companies like Yelp, they only pay to acquire one side. So they're like, we have a strategy that's going to get one side of the market completely for free. In this case, Yelp, SEO with demand. Right. And in other examples, people will pay for both sides of the market, but they'll have a long lifetime and a very short payback period to make sure that there's a lot of slack in the system. So, and this is kind of what ended up happening with Grubhub, where we'd have a commission for direct sales on a restaurant that we'd make our money back on in a few months. And then we had a four month payback period on demand and the average restaurant and diner would stay with us for like four plus years. So we're like, okay, we have enough slack to feel like there's profits there and now it's proven that there are a lot of profits there. Now it's a public company. The third way I've seen it handled is if you feel like there is a batch of users and this is usually a supply side strategy. If you feel like there's just a group that you have to get and it makes the demand side work, you almost think of it as a sunk cost. So you're like, we'll pay whatever we need to do to get the airlines. And then if we get the airlines, we're golden because then we can drive all the demand and make profits there. And usually that's a venture backed startup that's like, I'll burn the capital to do the sunk cost to get the supply side and then there's plenty of profits later on. And obviously that's probably the least attractive of the three, but not necessarily a horrible strategy.

Yeah. Didn't Simon Rothman talk about why Uber won and the approach around like using capital as a way to brute force your way into network effects at a local level and the end state is so valuable that even if you burned a ton of cash on the way there, it's worth it? I think that's maybe what you're saying.

Yeah, I can't think of anyone that's burned more capital than me. And you know, evaluations are a construct of how valuable they are. They are the most valuable marketplace startup of all times. Right.

So maybe evidence that.

So maybe evidence that that's a good strategy. Right. But if there is a certain game theoretical aspect to that, Right, which is if they had realized that a little bit later, like for example in China, by the time that they were doing it, Didi was already doing that same strategy. So then it creates like a more of a zero sum game where no one wins. But if you can realize that insight earlier, which I think Uber did in the Americas, in Europe, then you can get so far ahead with excess capital that it's hard for someone like a Lyft to maintain a really serious competition with them. Very good point.

Have you seen many aspects where you think you're advertising to, say, the demand side of the marketplace, but you're actually pulling in the supply side as well. It's almost like halo on both sides.

So at grubhub Transit, advertising certainly was that. Because a lot of times the issue that a restaurant would have is that they had never heard of us. So if you started buying outdoor ads on the bus or on the trains, restaurant owners would see them, and then that would make them trust you more, which would be more likely that they would sign up with you. It also helped with restaurant retention because it's like, oh, they're doing stuff to help drive demand. I. I might not be seeing that in the orders yet, but now I know that the salesperson wasn't lying to me, that they say they do a lot of marketing to help drive us orders. Right. So in that part, it helped. We never got really good at tracking that effect. And I always, if I had a salesperson come to me and say, like, hey, I can't make enough sales because no one's heard of us, I always said, oh, well, that's just a bad salesperson. Cause I see this salesperson over here that, like, people also don't know about us there. And he came back with 30 restaurants. So we never said that we had to have that to drive sales, but we certainly saw qualitatively that it helped restaurants stick with us when we weren't necessarily driving the band that we should. And we felt like it helped unlock the conversion rate of a normal sales call a little bit higher too.

Yeah, I've heard some startups do say out of home TV in a local market, thinking that they're advertising to the. The demand side, and then both sides sort of lift simultaneously. We've seen anecdotes at eventbrighter Evidence as well for our pages that rank highly in Google for San Francisco events. Those pages actually acquire the supply side too. The supply side sees where the demand side is coming from, and they want to be there as well.

Yeah. I think a smart supplier will be like, well, what would someone that's looking for my services do? And then see who shows up. Right. So it makes sense. Cool. When you think about marketplaces, and we just talked a little bit about Uber, you typically think of, oh, they're winner take all or winner take most, or that's what the venture capitalists say. What plays into who wins in marketplaces, usually, in your view. And do you agree that they're usually winner take all or winner take most, or do you think there's room for lots of players.

So I think they tend to be winner take most at the very least, if not winner take all. And I think the general driver of that is most marketplaces have cross side network effects whereby if all of the demand is there, all of the supply wants to be there. And if all the supply is there, all the demand wants to be there. So it creates this mechanism where both sides tend to grow which has benefits around. It can lower your customer acquisition costs and increase the amount of revenue you can make per user. It increases retention. To your point before and when those dynamics kick in, there's typically only room for one, maybe three, two players in a given market. It tends to be too hard to catch up. And so I think the first person to liquidity and liquidity ultimately drives those network effects is the winner, not necessarily the first one to the market.

Yeah, I agree. I think it's first to scale in a market, not first to market. And the important word there is also market. And we've defined markets as local. Right. Which is why even though Uber globally is a very strong number one in certain locales, they are getting beat very thoroughly. In China they had the exit. In other places like Brazil, they are definitely like number two to 99 taxis and whatnot. So there's a battleground in every locale, but then there's usually one major winner in every locale. I agree with that.

That's part of my love hate thing with marketplaces. They're so hard to get going and then if you to have scale them to each new market and it's the same amount of work, it's really tough without a playbook at least so it can be painful.

Speaking of a playbook, one of the things that investors like about marketplaces is that so much gets solved by the people on either side of the marketplace that is like, oh, these are very capex low. You don't have to do a whole lot of things. Once you get the liquidity, everything just kind of solves itself and it's very profitable. But there's this very strong playbook on what works in markets and usually that is can be capital intensive. What do you think marketplaces should be exceptional at, should own? What do you think they should push to either side of the market in terms of building that playbook that is defensible?

I think obviously product is core, right? Building the core product that both sides love and find value in. And getting to product, market fit is key. Customer service is important. So how do you arbitrate when there are disputes between the two Sides and to that point setting clear trust and safety guidelines for marketplace. I think ultimately people just want a well lit path and understand how to play within the market. Who's responsible for what? What are certain refund policies? What happens when things go south? Is it all on me as the demand side? All on the supplier or some combination of the two or responsible for the marketplace? I think marketing and sales are obviously key. How do you build the supply and demand? Demand on one or both sides and then quality of supply matters a ton. Like you don't want just all of the supply people don't want to browse through a thousand crappy listings to find the one good one you want to have curated sometimes or at least great algorithms to surface what people want. And I think that's on the responsibility of the marketplace.

Yeah, and I think that's a painful lesson for marketplaces to learn that actually a key thing you want to do over time is kick out supply that doesn't meet a quality bar. And it was a while before grubhub we realized that we had to fire restaurants that just, you know, weren't meeting a service level. And I agree with you said about like customer service. One of the things that we realized early on at grubhub was if something goes wrong, they don't blame the restaurant as the demand side, they blame us. Right. Which means you might as well take ownership of the experience because they're going to blame you when anything happens. And that meant building a much bigger customer service team than we anticipated, giving refunds ourselves, even if it was because the food was bad or the food came late and that was because of the restaurant and not us. And I think that created a lot of trust on the demand side to continue working with grubhub even when something went south. And I think there's been a lot of research that proves if someone has a negative experience and you go above and beyond to make up for it, they're actually more loyal than if they if they didn't have one. We saw that in our data as well. If someone had a negative experience, we comped their meal, they actually ordered more often and were more loyal than people who had just had good experiences. And it's tough to do that as an entrepreneur because you basically have to scale customer service usually with people and people especially out here, they want to build software that solves problems. And you can build software that makes customer service agents more effective, but you still need people on the other end. And that was the largest team at.

Grubhub I think we also have a tendency in the Valley to get myopically focused on the product, and that is our product. But it's much more than that. These offline activities that happen, like the service is the end to end customer experience. It is, how good are the taskrabbits, for instance, are they qualified, do they have the right skill sets, et cetera? Are they cordial when they greet you, do they follow up, as you would expect? And. And for a long time when I was at TaskRabbit, I think we focused so much on the web experience, the native app experience, and not enough on the training and the offline end to end. And I think when the company shifted to do more of that, we saw better business results. So important not to forget that.

Yeah, I think that was a key issue that homejoy had in scaling its marketplace, was that most people that experienced it said they didn't actually do a very good job in cleaning my home. Right. And if that is an issue, nothing else is gonna fix that. Right. Until you get better at cleaning the home, because that's the value you're selling.

Turns out that impacts ltv and you can find yourself underwater on cacti LTV pretty quickly.

Yeah, which is exactly what happened and why they had to shut down. Last topic we'll talk about is, we've talked about the success of some famous marketplaces. We've mentioned Uber, Airbnb, Eventbrite, grubhub, even though there's been quite a few of those. I think we probably both expected more marketplaces to become successful in the last 10 years than there have been. Why do you think this golden age of marketplaces that perhaps you and I expected and a lot of investors expected didn't materialize as much? Do you have any theories?

I think the on demand economy is part of that. I think that these more fully managed services popped up that would have, I think, historically been more or less managed marketplaces. Examples being, well, just look at TaskRabbit's pivot. It was very much at first a bidded marketplace whereby it was just people posting a job and TaskRabbits deciding to bid. And over time it shifted to this managed approach where you have somebody at your house in 45 minutes if you want. And I think that is part of it. So it's not necessarily that services didn't pop up to service a vertical or a need, it's just how it was built out and less of a marketplace model. I think one learning for me is looking for markets that are either big enough in TAM and or have a high enough frequency of usage to make sense to enable a very large marketplace to be built are maybe not as many as we originally thought. And so a trend I think that we might start to see is it's hard to go head to head or impossible to go head to head with the large horizontal marketplaces. But there's likely to be much more verticalization and people will find verticals where there are very targeted passions and passionate communities that can be built that ultimately may increase to a big enough market to support a billion dollar valued marketplace. But they're not super obvious right now, at least to me.

Yeah, I don't think Airbnb was super obvious to most people and how big that business could be because it wasn't easy to do and Airbnb made it easy to do and now it's big. Yeah, I agree. I think a lot of things that would have been marketplaces entrepreneurs were able to raise a lot of money and say, why don't I just own supply entirely? And I think there was this analysis in Take a space. Should I just connect buyers and sellers? Should I own the delivery of the supply? In which case like, you know, in my experience with food you have like doordash and postmates come out and they're like, well, we'll actually like bring the food to the people. You don't need to have your own courier which expands like the market. And then some companies just saying like, oh, let's just vertically generate supply and we'll make the food ourselves.

That's right.

A lot of these companies grew really fast and and raised a lot of money. I think it's unclear if all of those were very sustainable businesses. So you might come out of that with more marketplace y approaches. But it certainly made people think maybe marketplace isn't the right go to market strategy for this space. I agree that there are going to be more spaces that are kind of quiet big but are less obvious. And I think in the portfolio we have a company called Convoy which connects truckers to people that need to ship things via trucks. And it's growing incredibly fast. Super boring market, but I think it's measured in like trillion dollar tam. Right. It's just no entrepreneur out here is thinking about truck drivers. Right. But this company in Seattle and they're doing really well. Self storage is a space that's heated up over the last couple years with clutter makespace trove. And you know, it turns out that's a multibillion dollar industry that you can connect buyers and sellers and create a better experience as well. So I think there are, like you said, these markets that are big, that a marketplace can help achieve better liquidity and a better experience for both sides. But it's not like the obvious stuff like food and travel anymore. That stuff's been taken. It's super competitive and you need to dig a little harder. And I think instead we saw food and travel and transport just get a lot more competitive with a lot more players in there. So I'm hoping that more people find these other either high market size or high market expansion opportunity spaces to create new marketplaces in.

Yeah, I totally agree with you. And I thought a very interesting point you made was some businesses started with the marketplace model in mind and have shifted over time to they can go to one or two suppliers for the supply side and it's actually much less work to manage the marketplace. Like, I think about Lending Club started very much peer to peer lending, which is effectively marketplace model. Renaud Lapanche, the founder of Lending Club, he's starting a new competitor. They are seeking all of their capital supply from a couple big investors. Like, he's learned that that's not the right approach for the market. And I think a number of other financial companies have. So that was pretty fascinating to me.

Yeah. The other thing that I thought about with marketplaces is that the huge ones got incredibly huge. Huge, right? Uber and Airbnb are the two most recent examples. If you look at other spaces, there have been similar issues happening. Right? Take online advertising. Facebook and Google got huge. It's hard. Like Twitter's trying to be a number three. It's not going super well. Snapchat's trying to be a number three. Their initial earnings did not go super well. It's just hard to create a lot of massive spaces. And now what you're seeing is it's easier for Uber to expand into an adjacent space, like food delivery or Airbnb to expand into adjacent space with, like, local experiences than it is for people to spin up those companies themselves and win. Because now Uber and Airbnb have a brand that they can trade against. So you're now seeing these verticalized experiences just get bigger by going more horizontal. Now that they have massive scale, it's going to be interesting to watch and see if that is most effective. And, you know, we're buying, you know, random stuff From Airbnb in 10 years newber or if there's all these new brands we've never heard of. But I think online advertising tells A tale that it could become increasingly like just a couple massive, massive players. Yeah. And you know, there's going to be some smaller opportunities around that, like hoping Pinterest does really well as an online.

Advertising crossing my fingers for you.

Right. But you know, maybe harder to break out into a household name.

On that topic of niche markets, I've heard the argument made that traditional market sizing may not apply as much to marketplaces if they are in fact winner take all, because the person or the company that does become a winner, the value of that enterprise may be much higher than if you're the second or third or fourth player in a much bigger market and or the point about expanding tangentially, there may be a bigger market than initially it looks on the outset. So I think that was an interesting perspective that I've read as well.

Yeah. And I think something you talked about was, I think there was an original thesis posted online. I can't remember who posted it around. Well, if you want to do social networking, you need to be a niche social network. Right. And potentially that being a path forward for marketplaces as well. Like find something perhaps super small at the beginning, but just own it and then see how big you can make it. I don't know if you have any other thoughts on that.

Yeah, it was M.G. siegler, who's a former tech crunchwriter, he's a partner at Google Ventures, he wrote last month about niche networks and he was referring to social networks and basically arguing, hey, if you're trying to build the next great social network, we only have so much time in our lives and there's only so much room on our home screen, on our iPhones. So it's not going to happen. Like you're not going to build the next Snapchat, Twitter, et cetera. And I thought that completely applied to marketplaces. Like you're not going to probably take, take down ebay, Amazon, Craigslist, although there are some exceptions there that I'm starting to see. So I might eat my words later. But again, finding these niche targeted passions either in a social networking context and, or for marketplace, I think could apply.

Let's talk about Craigslist for a minute because that's one of the historical big winners in marketplaces. I think we've started to see chinks in their armor finally after it's probably been predicted that they would fall for 10 years, but it now seems to be starting to happen. There are certain things that have been carved off and then attacking them more head on in some of their key areas or services like OfferUp and Let Go. What do you think has created that opportunity? What lessons are there for future entrepreneurs around how to tackle an incumbent?

If you'd asked me five years ago, can anyone take down Craigslist? I would have said, I really don't think so. But I think there's so much evidence with offer up up $1.2 billion valuation, I just assume something's working. My hypothesis and I have no information to support this other than my own observations, but I think the platform shift to mobile Craigslist, eventually the experience is so bad they don't have a native app. I think maybe a poor enough user experience can wear down consumers over time. And if a solution that comes up that is mobile first or native to the next platform, that that's emerging, maybe that's enough to take down an incumbent. But I don't know. I think we'll see.

Yeah, I think platform shift is a really interesting thesis for can incumbents succeed? I think the Facebook bear thesis, when they went public was like, oh, mobile's going to crush them. And then it turns out they're an amazing business on mobile. But I think for something like Craigslist, which has been a much slower mover, it's created a window for services like OfferUp to compete. You know, I use OfferUp and I don't think it's particularly that much better than Craigslist, but I think it's enough. It's actually mobile first, which is where the users are, that it's allowed them to grow. And I think Letgo was founded by OLEX founders, right. They applied that similar model and that's worked internationally for them as well. And so being really smart about building supply and demand, being really smart about platform shifts, and then maybe getting some luck if the incumbent is asleep at that wheel, might be the way to move forward. So look for things like that, I guess could be a learning for entrepreneurs.

Agreed.

All right, well, that's all the time we had for today. I want to thank Brian for coming in. I think we learned a ton of interesting things and yeah, thanks for listening, everyone.

Thanks so much, Casey.