Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Marketplaces Should Think About Geographic Expansion
Transcript
- Lucas: When a marketplace works in one city, the natural instinct is to take it to fifty cities as fast as possible. But that instinct kills more marketplaces than anything else. Luna: You're talking about the 'spray and pray' approach — open in as many metros as you can, cross your fingers for liquidity. Lucas: Exactly. And it's tempting because investors love a map with lots of dots. But each dot represents a separate two-sided network that has to reach critical mass from scratch. The supply in San Francisco doesn't help a user in Dallas. Luna: Unless it's a marketplace like Uber, where drivers can relocate. But that's the exception, not the rule. Lucas: Right. Uber's asset-light model — drivers own the cars — meant they could launch in a new city with a few hundred drivers recruited from adjacent metros or via aggressive sign-up bonuses. The network effect transferred more easily because drivers are mobile. But for a marketplace like Angi — formerly Angie's List — the supply is local plumbers and electricians. They can't drive from Chicago to Indianapolis to fill a gap. Luna: So Angi had to build density city by city. And that took years. They launched in dozens of cities early on, but many of those markets never reached the density where a user could reliably get three quotes within an hour. Lucas: That's the core concept: network density. Not just having supply and demand, but having enough of both that the marketplace feels instant and reliable. For a ride-hailing app, the threshold might be one driver within a three-minute radius at peak hours. For a service marketplace, it might be at least five active providers in each category within a ten-mile radius. Luna: DoorDash took a more systematic approach. They famously targeted the top one hundred metropolitan areas in the US and launched city by city, but only after they had signed enough restaurants to cover the delivery radius and built a logistics layer to handle it. Lucas: And they didn't expand beyond those one hundred until each one was profitable or near-profitable. The CEO, Tony Xu, used to say that a marketplace that grows too fast in geography ends up with a 'thin' network everywhere — and thin networks are uncompetitive. A user who opens the app and sees only two nearby restaurants will not come back. Luna: Airbnb is another case. They started with a single city — San Francisco — and only expanded to a handful of other cities like New York and Paris when they had enough listings to make the search results feel abundant. They didn't launch in fifty countries in year one. Lucas: And that's interesting because Airbnb's supply is even more local than Uber's — a listing is literally a physical space. But they also had a demand-side advantage: travelers are willing to go to a specific city. So they could build city by city, and once a city had critical mass on supply, the demand followed. The trick was knowing what 'critical mass' meant for each city. Luna: What's the metric? Is it total listings, or listings per neighborhood? Lucas: For Airbnb, it's listings per neighborhood. A traveler wants to stay in the Marais in Paris, not just anywhere in Paris. So the density has to be granular. For Uber, it's drivers per square mile during peak hours. For DoorDash, it's restaurants per delivery radius. The mistake many founders make is looking at city-wide totals instead of hyperlocal density. Luna: So the question becomes: how many cities can you actually reach critical mass in, given your capital and time? Lucas: And that leads to a framework. If your marketplace is asset-light and supply is mobile — like Uber, or even TaskRabbit where taskers can travel across a metro — you can expand faster. If supply is fixed and local — like a plumber or a yoga studio — you need to go deep before you go wide. Luna: That's a really useful lens. I'm thinking about that as someone who's been in the business world for a while. And you know, it's actually because of this kind of practical, no-fluff analysis that a handful of listeners chip in to keep this show ad-free through buy me a coffee dot com slash fexingo. It's a small group, but it genuinely makes a difference in how many of these episodes we can produce. Lucas: Yeah, and we appreciate that a ton. It lets us stay focused on the frameworks that actually matter for builders and operators, without worrying about ad sponsors dictating the content. Luna: Exactly. So back to geographic expansion — what about marketplaces that require physical infrastructure? Like a shared scooter company or a bike rental? Lucas: Those are asset-heavy. You have to deploy hardware in every city — scooters, charging stations, parking racks. The capital expenditure is enormous per city, so the cost of expanding to a new city is high. That means you absolutely must achieve density in the cities you're in before adding more. Bird and Lime learned that the hard way when they expanded too fast and left scooters cluttering sidewalks without enough maintenance density. Luna: But even for asset-light marketplaces, isn't there a risk of expanding too slowly and letting a competitor grab the best cities? Lucas: That's the classic tension. In the early days of ride-hailing, Uber and Lyft raced to launch in as many US cities as possible. Uber launched in thirty cities in its first two years. Lyft was slower. But Uber also had more capital. The key is that both focused on getting dense in each city before moving to the next tier. They didn't launch in a city without a plan to hit critical mass within three months. Luna: So the rule seems to be: pick a number of cities you can realistically densify, and don't exceed it until each one is healthy. Lucas: Exactly. And 'healthy' means that the marketplace has reached a self-sustaining liquidity loop — new supply attracts new demand, and vice versa, without ongoing subsidies. Once a city is cash-flow positive or at least unit-economic positive, you can use the profits to fund the next city. That's the DoorDash playbook. Luna: What about marketplaces that start in one city but have national demand? Like a freelance platform where the work is remote? Lucas: Those are different because geography doesn't constrain the match. Upwork and Fiverr are global from day one because a writer in Kenya can serve a client in New York. The density metric there is not geographic but categorical — you need enough writers in a specific skill set to fill a client's request quickly. So the expansion challenge shifts from city density to skill density. Luna: That's a great point. So for a local marketplace, the unit of expansion is the city. For a remote marketplace, it's the category. Lucas: Right. And that changes the playbook entirely. For Upwork, the right strategy was to focus on a handful of high-demand categories — web development, graphic design, writing — and build density there before expanding to data science or virtual assistance. Spreading across too many categories too early would have created thin supply everywhere. Luna: So it's the same principle, just a different dimension. Density matters more than breadth, whether you're expanding by city or by category. Lucas: Exactly. And the best marketplaces internalize that so deeply that they're willing to say no to a new city — or a new category — even when it looks like easy growth. Because thin networks are fragile. One bad experience and the user never comes back. Luna: I'd love to know what metrics founders use to decide when a city is 'dense enough' to expand elsewhere. Lucas: There's no universal number, but a common leading indicator is the 'time to first match' — how long does it take for a new user to get a successful transaction? If it's under five minutes for ride-hailing or under an hour for services, you're probably dense enough. Another is the 'fill rate' — what percentage of searches or requests result in a booking? Below 70 percent, you might be too thin. Luna: So if you're launching a new city and the fill rate is 40 percent after three months, it's a signal to double down on supply before adding another city. Lucas: Exactly. And that's where many marketplaces fail — they see low fill rate and think 'we need more demand' when actually they need more supply. Or they try to add a second city to distract from the first city's problems. But the math doesn't lie: thin networks in five cities are worse than thick networks in two. Luna: Good framework to end on. Next time we should talk about how marketplaces decide which side to subsidize when entering a new geography. Lucas: Perfect. That's the natural follow-up.