“We only lost $62 million!” I keep reading exciting headlines like this about iBuyer programs, headlines with uplifting celebrations about losing less than the year before. Some of these companies report significant increases in gross revenue, which sounds like a profit until you realize that overall there was a massive net loss. One article in Seeking Alpha describes iBuyer programs as “An intrinsically broken business model.”
iBuyer is short for “instant buyer.” These buyers are usually large companies that purchase massive numbers of homes as-is using algorithms and other technology to determine pricing, allowing them to make an “instant” cash offer. They then take care of the selling process and charge the seller fees to do so.
Zillow tried their hand at iBuying years ago, promptly lost billions, and then shuttered this program in 2021. Redfin’s iBuying business followed a similar trajectory and closed in 2022. Since then, there have been no profitable iBuyer companies to date (that I am aware of–please correct me if I’m wrong!). However, companies continue to spend billions trying to make iBuying work. The most encouraging statements regarding various iBuyers are usually less than thrilling, for example: “[This] isn’t a profitable business yet, but it is certainly starting to show a path to get there.” Inman’s latest article on this business model describes how “losses mount as iBuyer acquires and sells fewer homes,” though industry executives say they’re taking “big steps towards profitability.”
I can’t predict the distant future… perhaps iBuying will finally become profitable in a world nothing like this one? I could see this happening if somehow iBuyers controlled the majority of the real estate market due to a change in law that made it extremely risky to sell a home to an individual, causing sellers to prefer selling their homes to large corporations. In this scenario, the housing market could be much more controlled and predictable for iBuyer algorithms. Barring a drastic shift like this, it’s difficult to see a profitable future for iBuying in the near-term.
One major issue I see with this business model is that iBuying companies are trying to apply mass production techniques to a product that they only control for a short period of time, a product that is actually a bunch of homes owned by random individual humans. To complicate this further, these homes are products that can’t be moved or exchanged for one another. This sounds like what happens in eBay’s nightmares… eBay wakes up in a cold sweat after dreaming that it had to personally purchase every single item it sells, and couldn’t ship anything.
Like eBay, one could argue that in the case of iBuyers, it’s the tech platform that’s the actual product. However, a major difference between iBuyers and eBay is that the iBuyer has to purchase each home and sell it before it becomes a major drag on their balance sheet. eBay doesn’t care that much if an antique clock listed for $10,000 sits on their website for a year, but that’s because eBay itself at no point has to pay $10,000 for the clock.
Real estate cannot be moved and is a non-fungible asset, so it’s inherently a localized market. This makes it even more difficult to apply automated, mass-production techniques to selling homes. The problem with the Zestimate is a perfect illustration of this, as described by an article in Business Insider. This article uses the Austin, Texas market as an example, where 94% of homes sell for “within 10% of the last Zestimate before the deal goes through.” This sounds like a highly accurate algorithm if you don’t think about it at all. As described by Business Insider, when a home is listed, the Zestimate changes to account for the information provided by the seller, including new photos, recent updates… and the list price. From there, “the Zestimate keeps adjusting until the house actually sells.” Thus, the Zestimate is simply following all of the human-provided information and adjusting based on this info until the property is sold. The Zestimate’s real prediction is that most homes will sell for within 10% of the last listed price before their sale… which is one of the least compelling predictions I’ve ever heard. The average home sells very close to its list price; according to Redfin, “The typical home sold for exactly its list price one year ago, and roughly 2% above its list price two years ago.”
The problems with iBuying are remotely similar to the problems with self-driving cars. I remember having an argument in 2015 with a friend who insisted that within the next two years, all cars would be full autopilot. To be fair, many news articles back then were saying similar things. Needless to say, this reality has not come to pass, though I imagine we’ll all be in completely self-driving cars well before iBuyer programs become financially viable. I’m not saying that full autopilot for all cars on the road will never happen, and in fact various levels of autopilot have dramatically changed driving for the better. However, people tend to jump the gun when technology like this can do a number of dazzling things really well. Currently, autopilot cars in a controlled environment in perfect weather are quite functional. However, once you get into snowy weather driving at night along a gravel road, you suddenly run into a lot of highly unpredictable problems, some of which might result in your car having to make weird philosophical choices, like choosing between your life and the life of a young family in a split second, if it can even see through the snow to do so.
I’m getting off topic here, but not as much as it might seem. Some general theories about housing markets provide genuine insight in stable and predictable environments. However, to make money at predicting housing prices, there are an absolutely wild number of factors to predict on both a macro and micro level. An iBuyer’s AI model would have to predict what natural disasters will occur in the near-term, how home tastes will change, what politicians will be elected to office, what laws will change, how individuals across local markets will respond, and then the algorithm would have to assign thousands of highly accurate price adjustments to this information. A large-scale iBuyer model must essentially know how to predict the future of the world in order to make money. AI may reach that level of complexity eventually, but that is a really high bar to clear if you’re just trying to flip some houses.
https://www.businessinsider.com/is-my-zestimate-accurate-home-prices-obsession-zillow-algorithm-homeowner-2024-12
https://www.redfin.com/news/housing-market-update-homes-selling-below-asking-price-june-2024/
https://seekingalpha.com/article/4660959-opendoor-major-downside-expected-intrinsically-broken-business-model
https://www.inman.com/2025/02/24/offerpad-losses-mount-as-ibuyer-acquires-and-sells-fewer-homes/?utm_source=hotsheet&utm_term=&utm_medium=email&utm_campaign=HotSheet_20250225
https://finance.yahoo.com/news/opendoors-business-clearly-improving-buy-110800037.html
https://www.rubyhome.com/blog/ibuyer-stats/
https://www.mikedp.com/articles/2021/11/3/zillow-exits-ibuying-five-key-takeaways
