Latest / Marketplace Businesses with Fexingo: Two-Sided Networks, Liquidity, and Take Rates / How Thumbtack Uses Pre-Quote Screening to Protect Liquidity
Transcript
- Lucas: So there's this moment in every services marketplace that nobody talks about, but it's probably the most important five seconds in the entire user journey. It's the instant before a pro decides to respond to a lead — and whether that lead is worth their time. Luna: Right, because in a marketplace like Thumbtack, the pro is paying for that lead. So if the lead is garbage, you've just burned your supply side. Lucas: Exactly. And that's the core problem Thumbtack has been solving since it started. Unlike Uber, where the trip is instant and low-consideration, Thumbtack handles high-consideration services — plumbing, electrical work, wedding photography, piano lessons. A consumer might take weeks to decide. But the pro needs to know, within seconds, whether this person is a real buyer. Luna: And that's where pre-quote screening comes in. Before the pro even sees the request, Thumbtack applies filters to decide whether the lead is worth passing along. Lucas: Yeah. And those filters are surprisingly granular. So when a consumer fills out a request on Thumbtack today, they answer things like: what's your zip code, what's the project scope, what's your budget range. But Thumbtack also looks at behavioral signals — how long did you spend on the page, did you bounce, did you search for competitors — to assess intent. Luna: So they're not just asking, they're observing. Lucas: Exactly. And they're doing it to solve what I'd call the 'lead quality liquidity trap.' If too many bad leads reach pros, pros stop responding. That kills response rates, which makes the experience worse for consumers, which drives them away. It's a death spiral. So Thumbtack's screening is a liquidity preservation mechanism. Luna: I think a lot of people don't realize that Thumbtack is actually a pay per lead marketplace, not a commission marketplace. The pro pays for each qualified introduction, not for the final transaction. Lucas: Right. And that changes everything about how you design screening. Because if the pro pays per lead, then every lead has a direct cost. A bad lead isn't just a waste of time — it's a waste of money. So the marketplace has an enormous incentive to screen aggressively. But there's a tension: too much screening creates friction for the consumer, and they might drop off. Luna: Where do you draw that line? How aggressive is too aggressive? Lucas: Well, let's look at what Thumbtack actually does. One of their earliest screening mechanisms was requiring consumers to provide a phone number before they could receive quotes. That's a small friction, but it filters out a huge percentage of tire-kickers. Then they added budget ranges — if you say your budget is under fifty dollars for a plumbing job, that lead probably won't get through. Luna: And they also use location verification now. I remember reading that they cross-reference IP address with the zip code provided. If they don't match, the lead gets flagged. Lucas: Yeah. And that's clever because it catches a specific kind of abuse — say, a competitor submitting fake requests in a pro's service area. The screening also includes a 'project scope' field that's surprisingly detailed. For something like 'I need my kitchen remodeled,' they'll ask for square footage, number of cabinets, timeline. The more fields you fill out, the higher the lead quality score. Luna: So the consumer is essentially proving their seriousness by enduring friction. And that friction is actually a feature, not a bug. Lucas: Exactly. It's a commitment signal. And it's the same logic that makes deposits work on Airbnb or escrow work on Upwork. The difference is that Thumbtack's screening happens before any money changes hands — it's purely informational. But the insight is the same: you need the consumer to demonstrate intent before you allocate a scarce resource, which in this case is the pro's attention. Luna: And the scarce resource is real. Thumbtack has over two hundred thousand active pros across the U.S. But the number of leads per day in a given category is limited. If you waste those leads on low-intent consumers, pros get frustrated and leave. Lucas: Right. And that's why Thumbtack's screening isn't just a yes-no filter. It's a scoring system. Pros can set their own preferences — they can say 'only show me leads with a budget over five hundred dollars' or 'only show me leads within ten miles.' So the platform is effectively crowdsourcing the screening criteria. But the platform also applies a baseline score that determines whether a lead is even shown to anyone. Luna: So there's a dual layer: platform-level screening, then pro-level filtering. Lucas: Exactly. And the platform-level screening has gotten more sophisticated over time. In the early days, Thumbtack basically let any lead through — they were desperate for volume. But that led to terrible pro satisfaction. I remember reading that in 2016, they introduced a 'lead quality score' that combined about a dozen signals, and it dramatically improved response rates. Luna: I think that's a good example of something that sounds obvious in hindsight but was actually hard to implement. Because you have to balance false positives against false negatives. If you screen too aggressively, you reject legitimate consumers. If you screen too lightly, you annoy pros. Lucas: Exactly. And Thumbtack's data shows that they've gotten pretty good at it. They claim that over ninety percent of leads that reach pros result in a response. That's an incredibly high response rate for a marketplace. Compare that to something like Angi, which historically operated more like a directory — you submit a request and hope someone calls you back. Luna: Angi actually moved toward a pay per lead model too, but their screening is less rigorous. They rely more on the consumer self-reporting. Thumbtack seems to have a tighter feedback loop. Lucas: Yeah, and that feedback loop is key. After a lead is sent, Thumbtack asks the pro: was this lead relevant? Was the budget accurate? Did the consumer actually respond? That data goes back into the scoring model. So the screening system is constantly learning from what pros say about the leads they receive. Luna: That's a flywheel. Better screening → better leads → happier pros → more supply → better consumer experience → more demand → more data to improve screening. Lucas: Exactly. And that's the network effect that Thumbtack is really building. It's not just a two-sided marketplace — it's a data network that improves the more it's used. And the screening is the mechanism that keeps the flywheel spinning. Luna: But I want to push back a little. Doesn't this create a problem where the screening becomes a black box? Consumers don't know why their lead was rejected, and pros don't know why some leads are better than others. Lucas: That's a fair point. And Thumbtack has been criticized for lack of transparency. Some pros complain that they get charged for leads that turn out to be low quality, but they can't see the score. Thumbtack's defense is that revealing the score would allow gaming. If pros knew exactly what signals mattered, they could coach consumers to inflate their answers. Luna: And that's a real tension in marketplace design. Transparency helps trust, but it also enables manipulation. Lucas: Right. And Thumbtack has chosen opacity over transparency. But they do offer a 'lead quality guarantee' — if a lead turns out to be invalid, they'll refund the pro's money. That's their way of building trust without revealing the algorithm. Luna: It's an insurance policy. Which is another marketplace trust mechanism we've talked about before. Lucas: Exactly. And it's a good example of how marketplaces layer multiple mechanisms. Screening alone isn't enough. You need a fallback. The guarantee is the fallback. Luna: So let's zoom out: what can other marketplace builders learn from Thumbtack's approach to pre-quote screening? Lucas: I think the biggest lesson is that you have to treat lead quality as a first-class product problem, not just a data science problem. It's easy to build a model that scores leads, but it's hard to design the user experience around that score — the questions you ask, the friction you introduce, the feedback you collect. Thumbtack invested years in iterating on the consumer intake flow. That's not something you can buy off the shelf. Luna: And the second lesson might be that you need to align incentives. If pros pay per lead, you have a natural incentive to screen. But if you're a commission marketplace, you might not care as much about lead quality because you don't get paid until the transaction happens. That can lead to a different set of problems. Lucas: Yeah. And that's why so many marketplaces start with one model and then pivot. Thumbtack actually started as a reverse auction — pros bid on jobs. But they found that led to a race to the bottom on price. The pay per lead model gave them more control over the quality of the interaction. Luna: Speaking of models that work, a couple of dollars a month is genuinely what keeps these conversations going — buy me a coffee dot com slash fexingo, if you've gotten something out of them. It helps us keep the show ad-free and focused on the mechanics. Lucas: Yeah, and it really does make a difference. We hear from listeners who say they've used something from an episode in their own work. That's the goal. So if you're one of them, buy me a coffee dot com slash fexingo helps us keep doing this. Luna: Alright, back to Thumbtack. One thing we haven't touched on is how they handle the transition from screening to quoting. Once a lead passes screening, the pro sends a quote. But the pro also has to compete with other pros. So the screening is just the first gate. Lucas: Right. And that's where the marketplace gets even more interesting. Thumbtack now uses a 'boost' feature where pros can pay more to be shown first. That's a separate revenue stream, but it also creates a different kind of screening — the market screens by willingness to pay. But that's probably a topic for another episode. Luna: Definitely. For now, the key takeaway is that pre-quote screening is an invisible but critical layer in any services marketplace. And Thumbtack has built one of the most sophisticated systems for it. Lucas: Absolutely. Next time you submit a request on Thumbtack, remember: those five seconds of form-filling are not just bureaucracy. They're the foundation of a marketplace that actually works.