Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Marketplaces Use Cross-Side Subsidies to Solve the Cold Start
Transcript
- Lucas: Let's talk about a tactic that almost every successful marketplace has used at some point, but rarely admits to publicly, at least not in those terms. Luna: You mean the one where you basically pay people to use your platform, even though there's no revenue yet from that side. Lucas: Exactly. Cross-side subsidies. The concept is simple: you deliberately lose money on one side of the marketplace in order to attract the other side. Uber Eats, for example, spent years subsidizing delivery fees for consumers. They were paying part of the delivery cost out of their own pocket, sometimes the entire fee. Luna: Which is essentially giving free money to diners so that restaurants see enough orders to stay on the platform, and drivers see enough demand to keep driving. Lucas: Right. The cold start problem is the classic chicken and egg: you need supply to attract demand, and demand to attract supply. A cross-side subsidy is a way to break that cycle by effectively paying one side to show up before the other side is large enough to make it worth their while. Luna: But there's a real risk here. You're burning cash. If you don't reach the liquidity threshold — that point where the platform becomes valuable enough without subsidies — you're just subsidizing a party that never gets going. Lucas: Which brings us to the difference between a smart subsidy and a dumb one. A smart subsidy is designed to be temporary and to build habits. DoorDash, when they launched, guaranteed Dashers a minimum hourly rate — something like $15 an hour even if there were no orders. That subsidy made drivers available, which meant delivery times were fast, which meant customers stayed. Luna: And once there were enough customers, the order volume became high enough that the minimum guarantee kicked in less and less often. Eventually, Dashers earned well above that floor without DoorDash having to top them up. Lucas: That's the ideal. The subsidy creates the initial liquidity, and then the liquidity makes the subsidy unnecessary. Contrast that with Groupon, which subsidized deeply discounted deals for consumers. The problem was that the subsidy became the entire value proposition. Luna: Once Groupon raised its take rate or reduced the discount, merchants saw less value, and consumers went elsewhere. There was no liquidity moat. Lucas: Groupon's cross-side subsidy was effectively a Ponzi-like dynamic: you need ever-deeper discounts to keep people coming. But because the subsidy didn't build any habit or network effect that lasted beyond the deal, it collapsed when the funding dried up. Luna: So what's the right way to structure a cross-side subsidy? Are there rules of thumb? Lucas: I think there are three principles. First, subsidize the side that is more price-sensitive and has lower switching costs. For most marketplaces, that's the demand side. Consumers will switch for a dollar off. Supply-side professionals — contractors, drivers, hosts — often care more about reliability and volume. Luna: Makes sense. Uber Eats subsidized consumers with fee discounts, not restaurants. Airbnb, interestingly, did the opposite early on: they subsidized high-end hosts by sending professional photographers to take listing photos for free. Lucas: Which is a great example of the second principle: subsidize the side that creates the most value for the other side. A great photo of a rental doesn't just help that host; it helps every guest searching on the platform, and it makes the whole marketplace look more trustworthy. So subsidizing supply quality improves the experience for demand. Luna: Third principle: the subsidy must have a clear off-ramp. You need to know, from day one, what metric signals the subsidy can stop. For DoorDash, it was average earnings per hour exceeding the minimum. For Uber Eats, it was order density per neighborhood reaching a threshold where delivery fees could cover driver pay. Lucas: If you don't define that off-ramp, you end up like many failed on-demand startups: subsidizing indefinitely, never reaching unit economics that work. I think of Homejoy, the home cleaning marketplace. They subsidized heavily — first clean for $20. That brought in tons of customers, but cleaners were paid poorly, and quality suffered. Customers churned, and the subsidy never built loyalty. Luna: So the subsidy actually attracted the wrong kind of customers: bargain hunters who would leave as soon as the price went up. Lucas: Exactly. That's why the most effective cross-side subsidies are targeted. Thumbtack, for instance, now uses machine learning to identify exactly which homeowners are most likely to hire a pro after receiving quotes. Instead of subsidizing all quotes, they only subsidize the ones with a high predicted conversion rate. Luna: So the subsidy is a loss leader, but a very precise one. You're not spraying money; you're placing bets where the expected lifetime value is highest. Lucas: Precisely. The other modern twist is using subsidies to solve seasonal liquidity imbalances. TaskRabbit, for example, will sometimes subsidize taskers in certain cities during slow months to keep supply available for when demand spikes. That's a cross-side subsidy with a time dimension. Luna: Speaking of time, I want to bring this back to something you said earlier about the cold start. Is there a case where a marketplace can launch without any subsidy at all? Lucas: Rarely, but it happens when the supply side has an intrinsic motivation to be there. Early eBay didn't need to subsidize sellers; people already wanted to sell their collectibles. The platform just needed to create a way for buyers to find them. But for most modern marketplaces — especially on-demand services — there's a gap between what supply is willing to do for free and what demand is willing to pay. The subsidy bridges that gap. Luna: It's basically venture capital building a bridge: you spend money now to create the network that generates returns later. But the bridge has to be built to the right destination. Lucas: And you have to know when the bridge is complete enough that you can tear down the tollbooth. By the way, if conversations like this are useful for what you're building or running, the show stays ad-free because of listener support. You can find us at buy me a coffee dot com slash fexingo. It's a small thing that makes a big difference for us. Luna: Absolutely. And it keeps the focus on the substance, which is what we care about. Lucas: So back to subsidies. I think the biggest mistake founders make is treating the subsidy as a marketing expense rather than a liquidity investment. Marketing is about awareness; a cross-side subsidy is about behavior change. You're trying to create a habit, not just a transaction. Luna: That's a good lens. So if I'm starting a marketplace today, what's the first thing I should do? Lucas: Identify which side has the highest willingness to pay and which side has the highest price sensitivity. Subsidize the price-sensitive side, but only enough to get them to transact. Then measure the retention and referral rate of those subsidized users. If they don't return without the subsidy, your money is wasted. Luna: And if they do return, you've hit the liquidity threshold. Then you can gradually reduce the subsidy and start optimizing take rates. Lucas: Right. And remember that the subsidy doesn't have to be cash. It could be time, convenience, or quality. Airbnb's free photography was a non-cash subsidy that created huge value. So think creatively about what your supply side values that costs you little to provide. Luna: Good point. That's a much more sustainable subsidy model because it doesn't directly burn cash. Lucas: Exactly. So to sum up: cross-side subsidies are a powerful tool for solving the cold start, but they must be temporary, targeted, and tied to a clear liquidity goal. Use them to build a habit, not to buy a transaction. Luna: And have an exit plan from day one. I think that's the key takeaway. Lucas: Absolutely. Next episode we'll look at how the best marketplaces use data to personalize the subsidy — essentially pricing the subsidy per user based on lifetime value. Luna: Looking forward to that.