Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Thumbtack Solved the Quality Problem Without a Centralized Review System
Transcript
- Lucas: So Thumbtack, the platform for finding local service pros — plumbers, wedding photographers, DJs, personal trainers — reached a hundred million dollars in revenue in 2024. But the really interesting part is how they built trust without a traditional review system. Luna: Wait, they don't have reviews? I thought every marketplace has reviews. Lucas: They do have reviews now, but early on, reviews were sparse. The problem is that when you hire a plumber, you might do it once every two years. So you can't build a review base the way Amazon or Yelp does. Thumbtack realized that the trust signal for services isn't 'what did other people say' — it's 'is this person vetted.' Luna: So they bet on background checks instead of user ratings. Lucas: Exactly. They introduced a 'background check' badge for pros who voluntarily submitted to a criminal records check and identity verification. The badge costs the pro maybe thirty or forty dollars, but Thumbtack covers part of it. And the data was clear: jobs that included a background-checked pro had a 40 percent higher conversion rate from quote to booking. Luna: That's huge. But doesn't that create a two-tier system where only pros who pay for the badge get the business? Lucas: It risks that. But Thumbtack positioned it as a premium trust signal, not a gate. And because the marketplace is request-based — you describe your job, and pros send you quotes — the customer sees a mix. The background-check badge just makes one stand out. It's opt-in, not required. Luna: Right, because on Thumbtack, the customer posts a job, not a search query. That's a different liquidity model than, say, Uber. Lucas: That's the second key design choice. Most marketplaces are inventory-based — you search for a product or a car ride and pick from what's available. Thumbtack is request-based. You say 'I need my kitchen sink fixed,' and pros come to you. That changes how you build liquidity on the supply side. Luna: Because instead of attracting a critical mass of listings, you need a critical mass of pros ready to respond to requests. Lucas: Right. And that's harder in some categories than others. For a common service like house cleaning, you get plenty of responses. For something niche like a petrochemical engineer consulting, you might get zero. So Thumbtack had to seed supply in the most demanded categories first. Luna: And they used a 'smart matching' algorithm to route the request to the most relevant pros, which also manages the customer experience — you don't get overwhelmed with 50 quotes. Lucas: Yeah, they limit the number of quotes per job, typically between three and five. That keeps the customer from suffering choice overload and keeps the pros from feeling like they're wasting time on a job they won't win. Luna: Which is a classic two-sided marketplace challenge: you have to balance the experience for both sides simultaneously. Lucas: And Thumbtack's take rate reflects that balance. They charge pros per quote, not a percentage of the job value. For a plumber, a quote might cost a few dollars. For a high-value wedding photographer, it might be twenty or thirty. The pro only pays if they respond. Luna: That's interesting because it aligns incentives — the pro only pays when they're actually pursuing the lead. But doesn't that lead to lower quality leads if pros just respond to everything? Lucas: That was a real problem early on. Pros were 'spraying and praying' — responding to every job regardless of fit. So Thumbtack added a feature where the customer sees how many jobs the pro has quoted recently. If a plumber has quoted 200 jobs this month, that's a red flag. Luna: So transparency on behavior, not just credentials. Lucas: Exactly. And that's the theme of Thumbtack's trust architecture: multiple signals layered together. Background check, response rate, number of quotes, reviews when they exist, and eventually a 'money-back guarantee' for jobs booked through the platform. Luna: The money-back guarantee is a big one. It shifts the risk from the customer to Thumbtack. If the plumber does a bad job, you're not out the cash. Lucas: Right. And that guarantee is only available if you book and pay through Thumbtack, which keeps the transaction on-platform. That's how they capture more data and more revenue, and also how they build trust. The guarantee is effectively an insurance product, and Thumbtack self-insures it. Luna: But self-insuring against bad jobs is risky. How do they keep the claims rate low? Lucas: They use the same signals to screen pros before they can offer the guarantee. A pro needs a certain number of completed jobs and a high rating to qualify. So the guarantee becomes a reward for good behavior, not a blanket offering. Luna: That's smart. It's like a carrot and stick built into the platform design. Lucas: And it works. Thumbtack's net promoter score is in the mid-60s, which is strong for a marketplace. The churn among pros who get consistent leads is low, and customer repeat rate is decent — about 30 percent of customers come back within a year. Luna: That repeat rate seems low for a marketplace. But I guess for services, you don't need a plumber every month. Lucas: Exactly. The purchase frequency is low, so the lifetime value of a customer is harder to capture. That's why Thumbtack has expanded into more frequent services like dog walking, tutoring, and personal training — to increase the stickiness. Luna: And they've also moved into larger projects like home renovation, where the job value is thousands of dollars. Lucas: Right. That brings higher take-rate dollars per job, but also more risk. A botched renovation is a bigger claim than a botched house cleaning. So the trust signals have to be even stronger for those categories. Luna: Before we get into the future of Thumbtack, I want to flag something that's been on my mind. If today was actually useful to you, the way these stay ad-free is listener support — buy me a coffee dot com slash fexingo. Lucas: Yeah, it's a simple way to keep the show going without any ads. We appreciate everyone who chips in. Luna: So back to Thumbtack: they're now testing ai powered job descriptions that automatically extract the key details from a customer's request and suggest the right category and price range. That should improve matching and reduce the number of bad quotes. Lucas: Which addresses the classic marketplace problem of asymmetric information. The customer doesn't know what the job should cost, and the pro doesn't know exactly what the customer needs. AI can bridge that gap. Luna: And if it works, it could increase conversion even further. Thumbtack's current conversion from request to booking is around 15 percent. That's room for improvement. Lucas: A 15 percent conversion means 85 percent of requests don't result in a booking. Some of that is just customers price-shopping, but a lot is mismatched expectations. If AI can reduce that mismatch, you get better liquidity without adding more supply. Luna: Exactly. And that's the holy grail for a marketplace: better matching without more spend. Lucas: So Thumbtack's story is really about how you build trust when the traditional signals don't work. They invented a new trust layer — the background-check badge — and layered on behavior signals, guarantees, and AI to create a system that works for low-frequency, high-stakes transactions. Luna: And the lesson for other marketplace builders is: don't copy Yelp. Understand your transaction's frequency and risk profile, then design the trust architecture from scratch. Lucas: Exactly. Thumbtack didn't try to be Yelp for services. They built something that fits the shape of the problem. And that's why they're still growing.