Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Uber Separated Liquidity From Driver Supply
Transcript
- Lucas: There is a moment every marketplace founder hits — you recruit a hundred suppliers, you open the platform, and it still feels empty. No transactions happen. And you think: I just need more supply. Luna: Right, the classic cold start. More supply means more liquidity, more transactions. That's the intuition. Lucas: But Uber figured out early that liquidity isn't just about how many drivers you have. It's about whether those drivers are where riders actually are, at the time riders actually need them. And that is a fundamentally different problem. Luna: So you're saying you could have ten thousand drivers across a hundred cities, and still feel empty in the one neighborhood that matters. Lucas: Exactly. And that's exactly what Uber saw in 2014. They had signed up drivers in most major US cities. But in any given city, the supply was concentrated in certain areas — usually near airports or downtown — and sparse everywhere else. On a Friday night in a residential neighborhood, wait times were fifteen, twenty minutes. Luna: Which defeats the whole value prop of ride-hailing. If I have to wait that long, I might as well call a taxi. Lucas: Exactly. So Uber's insight was: you don't need more drivers. You need drivers to move to where demand is highest at any given moment. And the way to get them to move is not by pleading with them in a newsletter. It's by changing the economics in real time. Luna: Surge pricing. Lucas: Surge pricing. But not just as a way to charge riders more. Think about it from the driver's perspective. If you're a driver sitting in a quiet part of town, you have no incentive to drive ten minutes into a busy area. But if fares in that busy area are two times, three times normal, suddenly that ten-minute deadhead makes sense. Luna: So surge pricing is actually a supply relocation mechanism. Lucas: Exactly. It's a real-time price signal that moves drivers from low-demand areas to high-demand areas. Uber's data from 2014 showed that during a typical Friday night surge in San Francisco, driver density in the Mission District increased by about 40 percent within twenty minutes of the surge multiplier hitting 2.0x. Luna: Forty percent. That's huge. And that means the marketplace felt liquid in the places that mattered most. Lucas: Right. And here's the key — Uber didn't need to recruit more drivers to achieve that. They just needed to redistribute the drivers they already had. That's a much cheaper and faster solution than trying to grow the supply base. Luna: So what does that mean for a marketplace founder who can't use surge pricing? Like, if you're building a platform for freelancers or rental items, you can't just dynamically adjust the price every five minutes. Lucas: Sure. But the principle applies. The question is: how do you concentrate supply where demand is concentrated? For some marketplaces, that means geographic targeting in your supply acquisition. Don't try to recruit suppliers in all fifty states. Find the three neighborhoods where you see early demand and recruit there aggressively. Luna: So it's a density-first strategy. Thiel talks about this in 'Zero to One' — dominate a small market first. Lucas: Exactly. And that's what Uber did in their early days. They launched city by city. They didn't try to be everywhere at once. But what's interesting is that even within a city, you need density at the neighborhood level. Uber's early metrics showed that if you had ten drivers in a two-square-mile area, the average wait time was under five minutes. If those same ten drivers were spread across twenty square miles, wait time jumped to twelve minutes. Luna: Same number of drivers, completely different experience. Lucas: Exactly. So the liquidity metric that matters isn't 'total suppliers.' It's 'suppliers per unit of demand density.' If you can't measure that, you're flying blind. Luna: And this is where a lot of marketplace founders go wrong. They celebrate hitting a thousand suppliers, but those suppliers are scattered across fifty cities. The platform still feels empty. Lucas: Yeah. And then they think they need more marketing to attract more users. But the real problem is supply distribution. Uber solved that with surge pricing. Other marketplaces solve it with different tools — like scheduling, or minimum commitments, or just really aggressive geographic targeting in the early days. Luna: Let's talk about one of those tools. You mentioned minimum commitments in a previous episode, but I'm thinking about something like OpenTable. They didn't have surge pricing. How did they solve the density problem? Lucas: OpenTable's approach was to sign up entire neighborhoods of restaurants at once. They'd go to a district like the Mission in San Francisco and recruit twenty restaurants in a two-block radius. That created a dense cluster where diners knew they could always find a reservation within a short walk. And once that cluster was liquid, they expanded to the next neighborhood. Luna: So it's the same principle — concentrate supply where demand is — but the mechanism is different. Instead of a price signal, it's a sales strategy. Lucas: Right. And that's the lesson for episode 93: liquidity is a function of density, not just volume. And the best marketplace builders design their entire go to market around that insight. Luna: I think this ties into something we haven't talked about yet, which is the 'phantom driver' problem. Uber had a specific issue where drivers would log in but not actually move. They'd sit in their driveway waiting for a fare. That created the illusion of supply without actual liquidity. Lucas: Great point. Uber's data showed that in some cities, up to 30 percent of drivers who were 'online' were actually stationary for more than thirty minutes. They were technically available, but they weren't where the demand was. So the marketplace looked liquid on the dashboard — drivers available! — but the customer experience was terrible. Luna: How did they fix that? Lucas: Partly through surge pricing, which incentivizes drivers to move. But also through what they called 'forward dispatch' — sending drivers a fare before they'd finished their current trip, so they'd be directed toward the next high-demand area. That kept drivers flowing through the system rather than parking. Luna: So the algorithm was actively shaping driver behavior, not just reacting to it. Lucas: Exactly. And that's the next level of marketplace design. It's not enough to have supply. You have to have supply that is responsive to demand signals in real time. If your suppliers are passive — they list an item and wait — you have to think about whether that's actually going to create liquidity. Luna: This is making me rethink how I evaluate marketplace startups. I usually look at total suppliers and total transactions. But maybe I should be looking at concentration ratios. Lucas: I think that's exactly right. For a marketplace with geographic dynamics, the Herfindahl index of supply concentration might be more predictive of success than raw supplier count. Or you could look at the ratio of suppliers in the top 10 percent of demand zones. Luna: So if you're a founder listening, what's the one thing you should do differently after this episode? Lucas: Map your demand density. If you have early users, plot their locations. Then figure out whether your supply is concentrated in those same areas. If it's not, you have a distribution problem, not a supply problem. And don't try to solve a distribution problem by recruiting more suppliers — that just makes the scatter worse. Luna: That's a good place to land. Speaking of things worth the effort — if this episode helped you think about marketplace liquidity in a new way, honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. It's the smallest possible ask, and it keeps these conversations ad-free. Lucas: Yeah, and listener support is literally the only reason we can spend an entire episode on a single marketplace mechanic without worrying about ad breaks. So if that's valuable to you, the link is there. Luna: Alright, but speaking of density — let's bring it back to a specific counterexample. What about marketplaces where supply is digital, not physical? Like a stock photo platform or a freelance coding marketplace. Does the density principle still apply? Lucas: It applies differently. Digital supply has zero marginal cost of reproduction, so you can have infinite supply without worrying about geography. But you still have the problem of matching supply to demand. On a stock photo site, if you have ten million photos of sunsets but only ten photos of industrial machinery, then for the customer searching for industrial machinery, the marketplace feels empty. So you need density in the demand-specific category, not in geographic space. Luna: So it's about categorisation, not location. Lucas: Exactly. The principle generalises: liquidity is about having the right supply where the demand is, whether that's a neighborhood, a category, a price point, or a time slot. The medium changes, but the structural problem is the same. Luna: I think that's a strong note to end on. Next episode we'll look at how marketplaces handle the opposite problem — too much supply concentrated in one place, and how they prevent congestion. Lucas: Looking forward to it.