Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How the Best Marketplaces Use Cancellation Policies to Build Trust
Transcript
- Lucas: So there's this moment every marketplace founder eventually hits — you've got supply, you've got demand, but bookings aren't clearing. People browse, they add to cart, and then they don't pull the trigger. And the fix often isn't better SEO or a cheaper price. It's something much more subtle. Luna: Let me guess — you're going to say cancellation policy. Lucas: Exactly. Cancellation policy is one of the most under-discussed levers in marketplace design. It sits right at the intersection of trust, liquidity, and unit economics. And the best marketplaces tune it obsessively. Luna: I think most users see it as a footnote. You know, 'Free cancellation until 24 hours before.' But for the platform, that sentence is doing a lot of work. Lucas: Absolutely. Let's start with Airbnb. Before 2020, Airbnb had basically two cancellation options: moderate and strict. Moderate meant you could cancel up to five days before for a full refund. Strict meant 48 hours. After the pandemic, they introduced 'Flexible' — cancel up to 24 hours before, full refund. And they added 'Long-term' for stays over 28 days. Luna: And that changed behavior significantly. I remember reading that after they introduced Flexible, listings that adopted it saw about a 15 percent increase in bookings, but also a measurable uptick in last-minute cancellations. Lucas: Right — the classic trade-off. Lenient policies reduce the risk for the guest, which boosts conversion. But they increase uncertainty for the host. A host who gets a last-minute cancellation loses that night's revenue entirely, and they can't rebook it easily. So Airbnb had to balance that by letting hosts choose their own policy tier. Luna: It's a trust signal. When I see a listing with 'Strict' cancellation — 50 percent refund only — I know that host is serious about protecting their income. It signals quality. Lucas: That's a really important point. The policy itself becomes a sorting mechanism. Guests who book a strict-cancellation listing are more committed. They're less likely to flake. So hosts who want reliable guests choose strict, and guests who want flexibility pay a premium. The platform just facilitates the matching. Luna: And then there's OpenTable, which went the opposite direction. They don't charge the diner a cancellation fee — unless the restaurant enables the 'Late Cancellation Fee' feature. That's typically a $10 to $20 charge if you cancel within an hour of the reservation or no-show. Lucas: Right. OpenTable's data showed that a no-show costs a restaurant about $75 in lost revenue on average. So the fee isn't about making money — it's about reducing friction. When restaurants started using it, no-shows dropped by over 40 percent in some cases. Luna: And that made reservation slots more reliable. Which means restaurants were willing to offer more prime-time slots on the platform. So OpenTable's liquidity improved on both sides. Lucas: This is where it gets really interesting from an economic perspective. The cancellation policy isn't just a trust tool — it's a liquidity management tool. If you make cancellations too easy, you get ghosting. If you make them too hard, you suppress demand. The optimal point depends on the vertical. Luna: Peerspace, the venue rental marketplace, has a particularly clever model. They let hosts set their own cancellation policy from a menu of three tiers — Flexible, Moderate, and Strict. But they also automatically charge a service fee on cancellations that's non-refundable. That fee goes to Peerspace, not the host. Lucas: So Peerspace is essentially saying, 'We'll help you get the booking, but if the guest cancels, we keep our cut.' That aligns incentives. The host still gets some compensation, the platform doesn't lose its transaction fee, and the guest knows there's a real cost to canceling. Luna: It also prevents a perverse incentive: if the host kept the full cancellation fee, they might be tempted to cancel the guest themselves and rebook at a higher price. Peerspace's design removes that. Lucas: Good point. That's a classic marketplace problem — the 'host-induced cancellation.' On Airbnb, if a host cancels on a guest, they get penalized heavily. But on Peerspace, because the cancellation fee goes to the platform, the host has no incentive to cancel. Luna: I want to bring up another angle: the refund window as a competitive moat. In the early days of Getaround, the car-sharing marketplace, they offered free cancellation up to one hour before the trip. That was way more lenient than traditional rental car companies. It helped them convert first-time users who were nervous about the peer to peer model. Lucas: Right. For a new marketplace, lenient cancellation is a customer acquisition tool. You're saying, 'We trust you. If you change your mind, no hard feelings.' That lowers the barrier to the first transaction. Then, as the marketplace matures, you can tighten policies to improve supply reliability. Luna: And that's exactly what Uber did. In the beginning, you could cancel a trip within five minutes with no fee. Now, in many cities, you get charged a fee if you cancel after the driver has been matched for more than two minutes. That change happened as supply got tighter and driver utilization became critical. Lucas: The data is pretty clear: every marketplace goes through a policy lifecycle. Early stage: flexible, to build trust and volume. Growth stage: balanced, to optimize conversion and reliability. Mature stage: data-driven tiers, to maximize lifetime value. Luna: Let's talk about the numbers. Do you have any sense of what the optimal cancellation rate is for a marketplace? Lucas: It varies wildly. For Airbnb, the average cancellation rate across all listings is around 3 to 5 percent. But for flexible-cancellation listings, it can be as high as 8 percent. For strict, it's under 2 percent. The platform doesn't want cancellation to be zero — that would mean people are too afraid to book. But above 8 percent, supply starts to suffer. Luna: So there's a Goldilocks zone. And the policy is how you hit it. Lucas: Exactly. And the really sophisticated marketplaces use machine learning to recommend policies to individual suppliers. Airbnb does this now — when you list a property, it suggests a cancellation tier based on your property type, location, and historical booking patterns. It nudges hosts toward the policy that maximizes total bookings for their specific situation. Luna: That's the kind of micro-optimization most people never see. But it's happening under the hood on every booking page. Lucas: If these conversations are useful for what you're building or running, we should mention that listener support is what keeps the show ad-free. A couple of dollars a month genuinely makes a difference — buy me a coffee dot com slash fexingo, if you've gotten something out of these episodes. Luna: Yeah, it's a small way to keep us independent and focused on the details that actually matter. We really appreciate it. Lucas: Back to the policy lifecycle — I want to talk about something that few marketplaces get right: the refund speed. On Vrbo, when a guest cancels under a flexible policy, the refund can take up to 10 business days. On Airbnb, it's usually within 24 hours. That difference affects guest satisfaction and repeat booking rates. Luna: Fast refunds signal financial stability. It tells the guest, 'We have the cash to pay you back immediately.' That builds trust. Lucas: And it's a competitive advantage. A study from a few years ago found that marketplaces that processed refunds within 48 hours had 15 percent higher guest retention than those that took a week or more. The policy itself is important, but the execution of that policy is just as important. Luna: What about the other side? How do marketplaces handle cancellations initiated by the supplier? Lucas: That's a whole other topic, but briefly — the best marketplaces make supplier cancellations very painful. On Airbnb, if a host cancels a confirmed booking, they get charged a penalty fee — usually 10 percent of the booking total — and those dates get blocked so they can't rebook them. On some platforms, repeated cancellations can get you de-listed. Luna: That's because one supplier cancellation can destroy trust for the entire platform. A guest who gets canceled on might never come back. Lucas: Exactly. That's why marketplaces invest so heavily in predicting which hosts are likely to cancel. Some use historical data to flag high-risk listings and either flag them or require a deposit. The goal is to prevent the cancellation before it happens. Luna: Which brings us full circle. The cancellation policy is not just a set of rules — it's a system that shapes behavior on both sides. It's one of those invisible architecture decisions that defines the marketplace experience. Lucas: And it's a decision you can't change without consequences. When Uber tightened its cancellation window from five minutes to two minutes in some markets, it saw a small drop in ride requests initially — about 2 to 3 percent — but driver satisfaction went up significantly. They measured the trade-off and decided the supply-side benefit was worth it. Luna: So the next time I see a 'Free cancellation until 24 hours before,' I'll know it's not just fine print. It's the result of a lot of data and strategy. Lucas: It really is. And I think that's the takeaway: if you're building a marketplace, spend serious time on your cancellation policy. Test it like you'd test your pricing. Because in a two-sided network, trust is the currency — and the cancellation policy is one of the most direct ways you mint it. Luna: Great episode. Next time, I want to talk about how marketplaces handle disputes between users. That feels like the next layer. Lucas: That's a perfect follow-up. We'll dig into that in Episode 23.