Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Thumbtack Solved the Lead Quality Problem
Transcript
- Lucas: You hire a plumber, they show up, do the job, you pay them. Simple enough. But if you've ever been on the other side — trying to find that plumber through a marketplace — you know it's not always that clean. Luna: Right. You submit a request, five different pros message you, three of them never follow up, and one quotes double what you expected. The whole thing feels broken. Lucas: Exactly. That's the lead quality problem. And it almost broke Thumbtack — one of the largest local services marketplaces in the US. Today I want to look at how they fixed it, because it's a really instructive case for any marketplace that deals with high-intent, infrequent transactions. Lucas: And before we dive deep — if these conversations are useful for what you're building or running, we want to mention something quickly. We deliberately don't run ads on these shows. It's a choice we made early on. If you want to support that choice, the link is buy me a coffee dot com slash fexingo. No pressure at all. Luna: Yeah, it's a small thing that keeps this ad-free. And honestly, it lets us talk about a topic like this without trying to sell you anything. So appreciate anyone who chips in. Lucas: Alright. So Thumbtack launched in 2008. The original model was straightforward: a customer posts a project — 'I need my gutters cleaned' — and nearby pros pay Thumbtack for the lead. Thumbtack charged per lead, fixed price, no matter the quality. Luna: That sounds like the classic lead-gen model. Angi — formerly Angie's List — did something similar for a while. But what went wrong? Lucas: A couple of things. First, because any pro could buy the lead, you'd get five responses — but maybe only one of them was actually good for the job. The customer had to sift through noise. Second, pros were paying for leads that often didn't convert — maybe the customer was just price shopping, or the job was too small. So pros churned because their cost per acquisition was too high. Luna: Classic marketplace failure. If the supply side doesn't get good leads, they leave. And if they leave, the demand side has fewer options. Death spiral. Lucas: Right. By 2015, Thumbtack had raised over $250 million, but the unit economics were shaky. They needed to fix the lead quality problem — not just for pros, but for customers too. So they made three big changes. Lucas: First, they introduced instant booking. Instead of posting a project and waiting for responses, customers could now see a pro's availability and book directly. That shifted the marketplace from a lead-gen model to a true transaction model. Thumbtack started processing payments and taking a percentage. Luna: That changes the incentive, doesn't it? When you take a cut of the transaction, you actually care that the transaction happens. In the lead-gen model, you get paid whether the customer hires anyone or not. Lucas: Exactly. So that was the first move. Second, they invested in pro reviews and background checks. They wanted to help customers feel confident booking instantly. But reviews alone weren't enough — because local services are so varied. A plumber might have 100 great reviews, but if they don't specialize in water heaters, the customer's still going to waste time. Luna: So how do you solve that? You can't have a review for every possible sub-category of plumbing. Lucas: You can't. So the third change was the most important: they rebuilt the matching algorithm to consider lead intent. They started asking customers more questions upfront — 'Is this an emergency?', 'What's your budget?', 'When do you need it done?' — and used that data to score leads. High-intent leads cost pros more, low-intent leads cost less. Dynamic pricing for leads. Luna: So the price of the lead itself signals quality. A pro sees a $20 lead and knows that customer is probably just browsing. A $50 lead means they're serious and likely to book. That's smart. Lucas: Exactly. And it self-selects. Pros who are good at converting high-intent leads will happily pay more, because their close rate is high. Pros who are less skilled can buy cheaper leads and still make it work. The marketplace becomes more efficient for everyone. Luna: I want to dig into the numbers a bit. What was the impact on Thumbtack's take rate and overall revenue? Lucas: So after these changes, Thumbtack's take rate — the percentage they keep per transaction — went from effectively zero in the old lead-gen model to around 15 to 20 percent for instant bookings. That's in line with other services marketplaces. More importantly, pro retention improved. The cost per acquired customer for pros dropped because they weren't wasting money on bad leads. Luna: And for customers, the experience got better. They didn't have to field five phone calls. They could just book. That's a classic network effect — the more high-quality pros you have, the more customers come, which attracts more pros. Lucas: Right. But here's the nuance: Thumbtack didn't just flip a switch. They had to manage the transition carefully. Some pros were used to buying cheap leads and resented the price increase. So Thumbtack introduced tiers — you could still buy leads the old way, but instant booking leads were priced higher. Over time, they shifted the mix. Luna: That's a good lesson for any marketplace making a structural change. You can't just cut off the old model overnight. You have to let the market adapt. Lucas: Yeah, and they also used data to prove the value. They showed pros that their close rate on instant booking leads was, say, 40 percent, versus 15 percent on standard leads. That made the higher price palatable. Luna: I'm curious how this compares with TaskRabbit. TaskRabbit also does local services, but they take a slice of the hourly rate. Do they have a lead quality problem? Lucas: TaskRabbit has a different structure. They're more like a labor marketplace — you book a tasker by the hour. The lead quality issue there is more about tasker reliability. Do they show up? Do they have the right skills? TaskRabbit solved that with a heavy vetting process and a rating system. But they don't have the same 'pro buys lead' dynamic because the platform controls the match. Luna: So Thumbtack's innovation was really about pricing the lead itself as a product. It's not just matching — it's pricing the match quality. Lucas: Exactly. And that's a model that can apply to other markets. Think about Upwork for freelancers, or even real estate marketplaces like Zillow. Anywhere there's a lead that someone pays for, you can tier the lead price based on intent signals. Luna: But it requires data. You need to know which questions to ask and how to predict conversion. Not every marketplace has that. Lucas: Right. Thumbtack had years of historical data from the old model. They could see which projects actually turned into booked jobs and which didn't. That gave them the training data to build the algorithm. A new marketplace starting from scratch would have to start simpler — maybe just a manual tier system based on project size or urgency. Luna: So what's the headline for someone building a marketplace today? If you're launching a local services platform, should you start with lead-gen or transaction? Lucas: I'd say start with transaction, if you can. The lead-gen model creates a misalignment of incentives from day one. You're making money even when the customer doesn't hire anyone — so you have less motivation to improve the match. But if you take a cut of the transaction, your success is tied to the quality of the match. That's a much healthier foundation. Luna: But transaction models are harder to set up. You need payments infrastructure, dispute resolution, maybe escrow. That's a higher upfront cost. Lucas: True. But the long-term network effects are stronger. And the trust is deeper. Thumbtack proved that you can migrate from lead-gen to transaction — but it's painful. They had to retrain both sides of the market. If you can start transaction-native, you avoid that pain. Luna: I want to circle back to the instant booking feature. That seems like the linchpin. Without it, the dynamic pricing of leads wouldn't work as well. Lucas: Absolutely. Instant booking gives the customer a frictionless experience, and it gives Thumbtack a guaranteed transaction to take a cut from. It also reduces the time to hire dramatically. In the old model, the average time from lead to booking was maybe two days. With instant booking, it can be minutes. Luna: That speed is a competitive advantage against traditional methods like calling around or using Angi's request a quote model. Lucas: Right. And it's one of those features that builds liquidity. When customers know they can get a plumber booked in 10 minutes, they're more likely to come back. And when pros know they'll get steady work through instant booking, they stay on the platform. Luna: So the lead quality problem is really a trust and efficiency problem. Thumbtack solved it by aligning incentives, using data to price leads, and making the transaction seamless. Lucas: Exactly. And they're not done. They're now experimenting with subscription models for pros — paying a monthly fee for a certain number of leads. That's another evolution. But the core lesson is: if you want a healthy marketplace, make the match quality the most important product you sell. Luna: Good place to leave it. Next episode we might look at how Thumbtack is handling the move into verticals like home security or commercial services. But that's for another day. Lucas: Yeah, lots more to unpack. For now, thanks for listening — and if you found this useful, you know where to find us.