Zillow CEO: ‘More than half of homebuyers cry during the process.’ AI can fix that
When Zillow launched 20 years ago, the home-buying process happened almost entirely offline. The company’s digital listings, combined with its innovative “Zestimate”—an estimate of a home’s value, based on the kind of data typically only available to real estate professionals—marked a turning point for the housing market. Zestimates weren’t exact representations of value, but they put power back in the hands of prospective buyers (to sellers’ and agents’ chagrin). Their near-instant popularity was an early “do your research” internet moment.
Fast-forward to the present day, and Zillow, which has a $13 billion market cap and reports earnings after the market close on February 10, is once again under pressure. In partnership with First Street, a provider of climate risk data, Zillow had been publishing climate risk scores for the homes on its platform. But some homeowners have balked at their scores, and some have even sued. A home that First Street labels as high-risk for flooding, for example, can quickly drop in value.
Zillow CEO Jeremy Wacksman, who joined the company in 2009 and became CEO in 2024, is walking a careful line. Zillow’s brand is associated with data transparency and consumer empowerment, but an increasing share of its revenue is tied to real estate agents who pay for software tools that help them virtually stage homes and message with open house leads.
On the eve of the company’s Q4 2025 earnings, which analysts estimate will be $650 million in revenue for the quarter, 17.4% growth, Wacksman sat down with Fast Company to talk about how Zillow is navigating the competing demands of its two-sided marketplace, the affordability crisis, and more.
This interview has been edited for length and clarity.
Zestimates weren’t perfect in Zillow’s early days, but even so, they were never taken down. Yet in the case of First Street’s climate risk scores, you’ve moved them to a separate place; people have to click through and leave the site. Walk me through the thinking there.
When you are focused on helping the consumer, the buyer, and seller, you want to try and provide as much information as you can with the right context. That’s what the Zestimate always has been. And we show our work, right? We show what our accuracy is. It’s not an appraisal, it’s not a professional opinion, but it’s a great starting point. Our philosophy is: Information needs context. And it’s hard for it to be perfect, but more information generally, when given the right context, is better than less.
Climate is the same way. We still have climate data on our site. We’ve tweaked what we show on Zillow versus what we link out to third parties, just given some of the accuracy data on some of the [climate] data sets. Get an official appraisal and figure out what you want to do, but almost every buyer and seller appreciates that data to start that conversation.
It feels like part of the reason people are so sensitive to potentially a bad climate score or a bad Zestimate is just because of the state of the housing market. Inventory is tight, sales are sluggish, prices are high. There is this growing consensus that we need to make owning a home more affordable. Where do you see that tension over affordability show up on Zillow?
The most stark change over the last decade has been the percentage of buyers who start with an affordability question versus start by window shopping. When I first joined, everyone just wanted to window shop. That’s how you got started. You fell in love with a home, then you hired an agent, then you figured out your financing. Now that’s flipped. About half as many buyers will say, I don’t even want to go get excited. I want to understand, can I do this? That massive shift, I think, is the best representation of the affordability crisis.
You’re right: We are in a period where homes are incredibly unaffordable. Home prices have run up nearly 100% in some markets from pre-pandemic levels. Average rates obviously ran up, too, so your mortgage per home gets even less affordable. That’s why you’re seeing transaction volumes at 30-year lows. In a normal housing market, 5.5 million to 6 million homes will trade every year. We’re at 4 million. Relative to the last couple of years, it’s getting a little better, but if you zoom out, it is still dramatically unaffordable.
What are the leading indicators you’re watching to see if the market is on the cusp of change?
One way to track the overall picture is [to look at] what you consider an affordable home for a median-income household. With a 20% down payment, 30% of your income [going toward] a mortgage payment is affordable. At the market’s worst, we were close to 40% on average, so almost nothing was affordable. Now it’s drifting back down. I think that’s what you watch for.
Zillow has been investing a lot in AI. One of the features that probably a lot of people have seen is virtual staging, or the use of software to add furniture and design elements to a listing. How do you go from conception to execution on something like that?
Virtual staging as a concept is not something we invented. What we invented was the ability to do it with really modest tools. We saw this technology coming; we saw 3D tours and interactive floor plans, all these things. But they were done with very high-end, very expensive cameras and media capture.
How do you get that to an average real estate agent? That’s the problem. The average real estate agent has maybe a $100 360-camera or an iPhone. So we built the technology ourselves. That’s why you now see virtual staging on 5% or 10% of listings in some markets. The agents can easily capture some media; we can generate an interactive walk-through.
Zillow is trying to go from being an ad-based business to supporting transactions directly. Do those tools generate the kind of leads that you’re interested in?
We are fortunate to have hundreds of millions of people on Zillow every month, shopping, dreaming, starting their transaction, using us as a companion. But we only help a very small share of those people actually buy, sell, or rent a home.
Virtual staging and interactive floor plans and 3D tours: Those are tools for a listing agent to go win more business. But we spend as much time on software for agents to run their business, whether that’s book their appointments, schedule showings, run their offer processes.
Sometimes, the best use cases of AI are just making that more efficient. I’ll give you an example. Agents are sending hundreds of messages a day to all their clients. Well, that’s something that generative AI can help with. We’re now sending millions of what we call smart messages. That’s something that sounds very simple, but think about the productivity gain. It helps them be a better guide, which helps them convert more business, which helps Zillow convert more business.
Zillow is a partner to ChatGPT, which has been starting to encourage e-commerce through its interface. Some shopping categories are pretty simple. It’s easy to imagine, for example, that someone is going to go to ChatGPT and say, “What are the best pliers?” But a home is a little bit different, right?
Buying a home and selling home, for sure, it’s a very different category than most verticals. The most interesting stat to me is that as technology gets more pervasive, and as data gets bigger, buyers and sellers feel even more paralyzed, and they want even more help. The percentage of people who say, “I want a full-service professional to help me” is higher now than it was before the internet, before the smartphone.
You could think about ChatGPT and Gemini as a source of traffic for a vertical like ours, or as a technology we can build into our experience. We think the latter is really where it’s going to go. The average buyer spends six to nine months shopping, sometimes longer, sometimes years on and off. In this category, you are ultimately making a set of trade-offs between what you can afford, what’s available at the time, and all of your lifestyle situations—the commute, the school, the neighborhood. You don’t get to just pick from a shelf of endless supply.
That’s why you end up hiring really good people to help you figure out how to make those trade-offs and figure out how to make sure you don’t make a financial decision that will cost you, because it’s the biggest financial decision you’ll ever make in your life. When it comes to figuring out the mortgage, figuring out which house to buy, figuring out how to sell your house, we think it will be this interweaving of professionals, professional software, buyers and sellers. And now you’re going to see LLMs [large language models] weave into all that.
Twenty years in the future, what’s the one thing that will be really different about the search and the one thing that will be really different about the actual transaction?
This category doesn’t work like any other category on your phone. You open your phone and you look at shopping and ride-hailing and travel and even wealth management—so much is digital. Real estate doesn’t work that way yet. We’ve made a bunch of progress, but it doesn’t. It’s very analog. It’s very paper based. It’s very asynchronous. We survey consumers every year, and the stat that always sticks out to me is that more than half of homebuyers cry during the process. I’m hopeful that in 20 years, hopefully in less than 20 years, we [can make Zillow into] more of a one-stop shop. It’s smart. It’s personal. It’s just all there for you.