Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Marketplaces Use Pre-Transaction Screening to Prevent Bad Actors
Transcript
- Lucas: Most marketplaces treat reviews as their primary trust mechanism — let users rate each other, surface the good, bury the bad. But that only works after a transaction has happened. If you're booking a plumber or hiring a nanny, the first interaction can't afford a learning curve. Luna: Right, because a bad first experience doesn't just mean a lost customer — it means a safety risk. So what do platforms do when reviews aren't enough? Lucas: They move trust upstream. Pre-transaction screening — verifying credentials, checking licenses, authenticating products — before the user ever gets to transact. It's a fundamentally different approach to building trust on a two-sided network. Instead of 'trust but verify afterward,' it's 'verify first, then trust.' Luna: And that's a design decision that ripples through everything — onboarding friction, supply growth, conversion rates. Let's talk about a platform that does this well. Lucas: Angi — formerly Angie's List — is a good case. For home services, the risk is asymmetric. If you hire a bad electrician, you could have a fire. So Angi requires service providers to submit business licenses, insurance certificates, and in some categories, trade-specific certifications before they can appear in search results. Luna: That's a pretty high bar. How does that affect supply acquisition? Lucas: It slows it down. Angi's onboarding conversion for providers is lower than, say, Thumbtack's, which had a lighter verification process early on. But Angi's data shows that screened providers generate 43 percent fewer service complaints in their first six months. That reduction in bad outcomes compounds — it builds a reputation that attracts higher-quality customers, which in turn attracts better providers. Luna: So it's a liquidity trade-off. You accept slower supply growth in exchange for a higher-quality core that becomes a moat. Lucas: Exactly. And it works especially well in categories where the cost of a bad transaction is high. Childcare is another one. Care.com requires identity verification, in many cases a background check — sometimes even fingerprinting — before a caregiver can even message a family. Luna: That creates a chicken and egg problem within the chicken and egg problem. You need caregivers to go through that friction before they see any demand. How do they get enough supply? Lucas: They subsidize the verification cost. Care.com covers the background check fee for the first year to lower the barrier. It's a deliberate investment — spend upfront to build trust, then recoup through subscription fees on the family side. You're essentially paying to prove your supply is safe. Luna: Let's flip to the product side — luxury resale. The RealReal doesn't have a license to check, but it has to authenticate every item before listing. That's pre-transaction screening of goods, not people. Lucas: Right. For consignment, the risk is counterfeit. If a fake Chanel bag gets through, it destroys buyer confidence in the entire platform. So The RealReal employs a team of authenticators — gemologists, leather specialists, hardware experts — who examine every item before it goes live. Their authentication failure rate, meaning items they reject as fake, is around 7.5 percent. Luna: That's a huge number. Seven and a half percent of items people try to consign are counterfeit? That means if they didn't screen, almost one in ten transactions would undermine trust. Lucas: And that's just what they catch. The process takes time — an item can sit in their warehouse for a week — which affects inventory velocity. But again, it's a deliberate trade-off. They'd rather have slower supply turnover than risk a counterfeit scandal that kills the network. Luna: There's a broader pattern here. The more high-stakes the category, the more screening you need before the first transaction. But does that always improve outcomes? I'm thinking about platforms that over-screen and end up with too little supply. Lucas: It's a real risk. If you make onboarding too onerous, you deter legitimate providers. A study of peer to peer lending platforms found that stricter identity verification reduced fraud by 60 percent but also cut loan originations by 20 percent because some borrowers dropped off. The trick is calibrating the threshold. Luna: And that threshold varies by category. For a marketplace like Airbnb, you probably don't need pre-transaction screening beyond identity verification — the cost of a bad stay is inconvenience, not safety. But for medical services, you'd want full licensure checks. Lucas: Absolutely. And the screening itself can become a competitive advantage. Uber's early model required drivers to undergo background checks and vehicle inspections — that was a differentiator against taxis, which had variable screening. It built trust in a category where safety was the number one concern. Luna: But Uber also got criticized for the quality of those checks — they weren't as thorough as promised. So screening only works if it's rigorous enough to matter. Lucas: That's the paradox. Screening that is perceived as performative does more damage than no screening at all. If users believe the platform has vetted providers but a bad actor slips through, the trust violation is worse. Angi and Care.com have both faced lawsuits over allegedly inadequate verification. Luna: So there's a legal liability angle too. If you claim to screen, you're on the hook for the quality of that screening. Lucas: Exactly why some marketplaces choose not to screen. Craigslist famously takes a hands-off approach — no verification, no pre-screening. The risk is entirely on the user. That works for lower-stakes categories like furniture or used electronics, but it also limits the kind of transactions you facilitate. You'll never get high-ticket or high-trust categories on Craigslist. Luna: And that's the strategic choice for every marketplace: what level of pre-transaction screening do you need to unlock the transactions you want? Too little, and you get adverse selection — only bad actors join. Too much, and you starve supply. Lucas: Right. And there's a new middle ground emerging — using data and machine learning to do risk-based screening. Instead of verifying every provider identically, you might verify new providers aggressively and then reduce checks for those with good track records. Or you might use behavioral signals — how they respond to messages, their device fingerprint — to flag potential bad actors before they even apply. Luna: That's more efficient, but it introduces its own challenges — algorithmic bias, false positives. If you deny a legitimate provider because a model flagged them unfairly, you've lost supply and possibly created a PR problem. Lucas: It's a calibration problem that every marketplace manager faces. And it's not static. As the network grows and the mix of providers changes, the screening criteria need to evolve. The platforms that get this right build a durable trust advantage. The ones that get it wrong either drown in bad transactions or never achieve liquidity. Luna: One thing that sticks with me: the 7.5 percent counterfeit rate at The RealReal. That's a huge hidden tax on the supply side that their screening catches. If you're a marketplace operator, how do you even begin to estimate that number for your own category? Lucas: You usually don't know until you start screening. But you can run a pilot — manually audit a sample of sign-ups or listings to estimate the rate of bad actors. That tells you whether screening is worth the friction. For most marketplaces, especially in services, the rate is high enough that screening pays for itself in reduced fraud costs and improved retention. Luna: So the question isn't if you should pre-screen, but how much and at what stage. And the answer depends on the cost of a bad transaction versus the cost of onboarding friction. Lucas: Exactly. And that's a calculation every marketplace founder needs to make early — because once you launch without screening, it's very hard to add it later. Users who joined under a light-touch regime will push back if you suddenly demand verification. Luna: That's a great point. It's easier to start strict and loosen over time than to start loose and tighten. And if this kind of analysis is helping you think through your own marketplace decisions, a couple of dollars a month is genuinely what keeps these going — buy me a coffee dot com slash fexingo, if you've gotten something out of them. Lucas: Yeah, we keep it ad-free and focused on practical design choices, and listener support is what makes that sustainable. So if the show's valuable to you, that's where you can help. Luna: Alright, back to screening. We've talked about Angi, Care.com, The RealReal. Let's look at a category where pre-transaction screening is still underdeveloped: freelance marketplaces for creative work. Lucas: That's a fascinating area. On platforms like Upwork or Fiverr, the risk is less about physical safety and more about quality and professionalism. A client might pay for a logo and get stolen clip art. The screening is typically portfolio-based — but portfolios can be faked. Luna: Right. I've seen cases where freelancers submit a portfolio of work they didn't actually create. And because the marketplace doesn't verify provenance, the client has no recourse except a bad review. Lucas: Some platforms are starting to implement skill tests or even video interviews as pre-transaction screening. But it's a delicate balance — creative professionals hate friction. Too much screening and you drive them to direct client relationships. Luna: So for creative marketplaces, the optimal screening might be lighter but combined with strong post-transaction protections — escrow, dispute resolution, money-back guarantees. Lucas: That's a hybrid model. Pre-screen enough to filter out obvious fraud, then use reviews and guarantees to handle the rest. It's not one-size-fits-all. The best marketplaces design their trust stack based on the specific risk profile of each vertical. Luna: And that design process itself — mapping out the risks, deciding where to intervene — is probably the most important strategic work a marketplace operator does. Lucas: Absolutely. Because trust is the asset that compounds. Every successful transaction builds it, every bad transaction erodes it. Pre-transaction screening is just the first layer — but it sets the foundation for everything else. Luna: And the best platforms treat it as a continuous investment, not a one-time check. They update their criteria as they learn about new fraud vectors. They're always looking for the next bad actor behavior. Lucas: That's the final lesson: screening isn't a feature you launch and forget. It's a living system that evolves with your marketplace. And the marketplaces that do it well create a trust moat that's incredibly hard to replicate.