Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Marketplaces Use First-Party Data to Raise Take Rates
Transcript
- Lucas: Luna, I want to talk about something that keeps marketplace operators up at night: how do you raise your take rate without destroying your liquidity? Luna: Right, the eternal tension. Take rate goes up, supply side gets grumpy, demand side may leave. What's the secret move here? Lucas: The secret might be data. Specifically, first-party data that the marketplace collects from its own operations. If you can use that data to prove you're driving more value, you can justify a higher fee. Luna: So it's not just about taking a bigger slice — it's about making the pie bigger first? Lucas: Exactly. Let me give you a concrete example. Uber's upfront pricing model. Before 2016, Uber showed you the estimated fare range, and the final price was based on time and distance. Now they show you a fixed price before you book. That fixed price is calculated using millions of historical trips, traffic patterns, weather data, even the likelihood that you'll accept a higher fare. Luna: And that lets them capture more consumer surplus. If they know you're willing to pay $25 for that ride, they don't leave money on the table by charging $18. Lucas: Precisely. Uber's take rate went from around 20 percent in 2016 to roughly 27 percent by 2024. Not all of that is upfront pricing, but a big chunk is. They're using data to price closer to what each rider will pay, and they're doing it algorithmically so drivers still get a competitive cut. Luna: But doesn't that risk alienating riders who feel like they're being charged more than the person next to them? Lucas: It does. And Uber has faced backlash. But they've been smart about it. They frame it as 'personalized pricing' based on demand, not on your personal data. The key is that the price is still a function of market conditions, not your past behavior. That's the line they walk. Luna: So the data isn't about who you are, it's about what the market will bear at that moment. That's an important distinction. Lucas: Right. Compare that to something like Amazon's marketplace. Amazon uses its first-party data — what sells, at what price, with what margins — to set fees for third-party sellers. They charge a referral fee that varies by category, from 8 percent for furniture to 15 percent for electronics. But they also charge for advertising, fulfillment, and premium placement. The total take rate can exceed 30 percent. Luna: And Amazon justifies that by saying, 'Look, we know what works. If you pay for our data-driven recommendations, you'll sell more.' Lucas: Exactly. Amazon's first-party sales give them an unmatched dataset. They know the optimal price for every product, the best keywords, the seasonality. Sellers pay for that insight. It's a take rate that's tied to value creation. Luna: But what about marketplaces that don't have a first-party sales arm? Like Airbnb. They don't own any listings. How do they use data to raise take rates? Lucas: Airbnb introduced smart pricing in 2015. It's a tool that suggests prices to hosts based on demand, local events, seasonality, and booking patterns. Hosts who use smart pricing see, on average, a 10 to 15 percent increase in bookings. And Airbnb's take rate is around 14 to 16 percent, which is higher than most hotel booking sites. Luna: So the data helps hosts make more money, and in return, Airbnb can charge a higher commission. Lucas: Exactly. It's a value for fee swap. The marketplace says, 'We'll use our data to help you earn more. In exchange, we keep a slightly bigger cut.' And it works because the host's net income still goes up. Luna: But if a host doesn't use smart pricing, they still pay the same take rate, right? That feels a bit unfair. Lucas: It is, and that's a criticism. But Airbnb's argument is that the platform's data benefits everyone, even passive hosts, because the pricing tool helps clear the market. If too many hosts price too high, guests don't book, and the whole marketplace suffers. Luna: So it's a collective good argument. Interesting. Let me ask you about the flip side. If data is so valuable, why don't smaller marketplaces just copy these strategies? Lucas: Because they lack the data. It's a classic data moat. Uber has billions of trips. Amazon has decades of purchase history. A new marketplace for, say, dog walking might have a few thousand transactions. Their pricing models are much noisier. If they try to raise take rates based on weak data, they risk killing the marketplace. Luna: So incumbents have a structural advantage. They can use data to optimize take rates, and new entrants can't compete on fee alone. Lucas: Right. And that's why we see so many marketplaces offering free tiers or low initial take rates. They need to accumulate data first. Once they have enough, they can start optimizing pricing and raising fees. Luna: That's actually a really useful insight for someone building a marketplace today. Don't obsess over take rate early on. Focus on transaction volume and data collection. Lucas: Absolutely. And there's a parallel here to our show. If these conversations are useful for what you're building or running, that's the whole point. Honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. Luna: It really helps keep the show ad-free and focused on the stuff that matters. So thanks if you've ever thrown a few bucks our way. Lucas: All right, let's get back to data moats. There's another angle I want to explore: the ethical boundary of using data to set prices. Luna: Where is that line? When does optimizing take rate cross into price gouging? Lucas: I think it comes down to transparency and choice. If users know that the marketplace is using data to set prices, and they can opt out or see alternatives, it's fine. The trouble starts when the data is used in ways that feel manipulative. Luna: Like that time when it was revealed that Uber's 'upfront pricing' sometimes showed higher fares to users with low battery, because they're more likely to accept a ride quickly. Lucas: Exactly. That's a creepy use of data. It exploits a moment of vulnerability. Uber said they stopped that practice, but it shows how easy it is to cross the line. Luna: So the smart play is to use data to improve the marketplace for both sides, not just to extract more from one side. Lucas: Right. And that's what the best marketplaces do. They use data to reduce friction, improve matching, and increase the total value created. Then they take a cut of that larger pie. Luna: Let's bring in another example: Upwork. The freelance marketplace. Lucas: Good one. Upwork charges a sliding scale take rate: 20 percent for the first $500 with a client, then 5 percent after $10,000. They use data to identify high-value relationships and lower the fee to keep them on platform. Luna: So the data tells them which relationships are worth subsidizing. That's a form of price discrimination, but it's pro-market. Lucas: Exactly. And it works. Upwork's take rate has been stable around 14 percent, but their gross services volume has grown 20 percent year over year. The data lets them optimize for long-term value, not just immediate fee extraction. Luna: That's a lesson for any marketplace. Don't just use data to squeeze more out of today. Use it to build a healthier ecosystem for tomorrow. Lucas: And that's really the takeaway. First-party data is the most powerful tool a marketplace has for raising take rate without breaking trust. But it has to be used responsibly. The moment users feel exploited, the data moat becomes a liability. Luna: So the next time you're on a marketplace and you see a fee, think about the data behind it. It's not just a number — it's a reflection of how much value the platform thinks it's creating for you. Lucas: And maybe ask yourself: is that fee backed by real data and real value, or is it just a tax on convenience? That distinction matters. Luna: Great episode. Thanks, Lucas. Lucas: Thanks, Luna. See you next time.