Zillow: the problem with pricing
I found myself scrolling through Zillow as my wife and I drove through a neighborhood a few weeks ago. Unless you’re a total stoic, then, like me, you’ve probably wondered about things like:
- How much does that house cost?
- …Can I afford it?
- ……What could I sell my house for?
With an app like Zillow, you can answer these questions almost instantly. But as with any transformative product, “Zillow surfing” has become such a habit that it’s very easy to forget what life was like before it existed. As a pioneer, how did Zillow make these insights possible? And why, if Zillow had developed such a strong market position, did Zillow’s instant offer program (Zillow Offers) flop? I think the answer has a lot to do with algorithmically estimating home prices.
Home buying before Zillow
Back in the 1990s, detailed real estate data was inaccessible to the average shopper. There were two repositories:
- Public Record: Courthouses, tax assessors, planning / zoning, and other public offices housed valuable property data.
- The Multiple Listing Service (MLS): The MLS was the database of record, but was only accessible to real estate professionals.
Most can hardly stomach a trip to the DMV, so unless you were really motivated, you weren’t headed down to the courthouse to dig through tax records. Secondly, engaging with a real estate professional carried its own friction; it wasn’t something you just did unless you were fairly serious about buying. Agents also had an incentive to quote higher prices to pocket commissions.
Moreover, working with a real estate professional went something like this: you’d find one in a phonebook and tell them you were after a 3 bedroom, 2 bath house on the east side of town. Your agent would access the MLS on your behalf, and print out half a dozen listings for a ride along. Without Zillow surfing, you went through an intermediary; you never had a full picture of the market; you were not an empowered buyer.
Starting with price
Although Zillow consists of a whole suite of property management tools, the very first thing Rich Barton and his team started with was the price of a home. On the nascent site, there were almost no additional tools, only an estimate compiled from millions of property listings. Importantly, Zillow allowed shoppers to add details about their home’s listing, increasing the estimate’s accuracy. Over time, this exchange created a free, public alternative to the MLS and sites like Realtor.com (the latter did not offer an estimate of the home’s value in the early days).
Critics wondered whether a web service could accurately assess a property’s value, but the bar was so low in the 1990s — remember, you’d have to call up a local seller’s agent and consult on your home’s price and prevailing market conditions — that it did not need to be incredibly accurate. Even an upper or lower bound was better than the alternative, which was usually nothing at all.
Zillow aggregates
Zillow provided value to virtually any shopper in the US by growing its listing count and user base over a series of acquisitions.
- StreetEasy: StreetEasy unlocked New York City for Zillow, one of the most dynamic real estate markets in the world.
- Postlets: By acquiring Postlets, Zillow enabled property managers to syndicate property listings across multiple sites, furthering Zillow’s reach.
- Trulia: In its largest acquisition to date, Zillow acquired Trulia. This provided Zillow with more monthly users, bolstering its ad business.
With each acquisition, Zillow increased their breadth, creating an unmatched database of real estate prices quicker than RedFin, their competitor. But where Zillow went fast and wide, RedFin went slow and deliberate. The latter focused on expanding city-by-city, and to this day, lacks truly ubiquitous listings.
For example, my parents live in Elizabethtown, Kentucky, where I can readily search Zillow for listings and property estimates. But when I do the same thing in RedFin, I get the “notify me” message below. This is a key point — more on it in a moment.
Zillow dives into iBuying
One of the hottest new real estate trends is offering an instant buying (iBuying) service to sellers. Depending on where you live, you can use a service like OpenDoor to get an instant quote for your home and a six figure check in the same week. What a time to be alive.
Now if you’re Zillow, and you’ve spent over a decade amassing the largest property listing and pricing database in the United States, why not dive into iBuying? On the surface, it would seem like the obvious play:
- Zillow buys homes below their proprietary estimate, saving sellers time and money.
- Zillow sells homes at or above the estimate.
- Zillow uses this sales data to reinforce their estimate, creating a virtuous feedback loop.
Zillow launched Zillow Offers in select markets in 2019, but the service was short-lived. In 2021, Zillow scuttled the Offers program and laid off 25% of its staff, losing some $500MM in the process. In the wake of the failure, CEO Rich Barton stated:
We’ve determined the unpredictability in forecasting home prices far exceeds what we anticipated.
So, given Zillow’s enormous database and experience pricing homes, how could such a venture have failed?
Without a careful study of the inputs and outputs of Zillow’s proprietary algorithms, we cannot know for sure. Analysts have put forth several theories:
- The pandemic, inflation, and supply chain issues besieged the whole iBuying market, throwing off whatever the algorithms had trained against.
- The algorithms did not react quickly enough to market heating and cooling.
- The algorithms failed to account for renovation and other costs related to home flipping.
All of these are probably valid, but I wanted to offer another explanation: Zillow’s core identity as a property listing marketplace for the masses was never going to bode well for Zillow Offers.
The difficulty of iBuying
Digitally flipping homes is an outright brutal business. As of early 2021, OpenDoor was generating just $5,000 in profit for every home sold. Assuming a median home price of $375,000 across the US, that means margins are just over 1%. There is no room for error, period.
Given the existential pressure of accurately calculating a home price, I believe OpenDoor’s algorithms are uniquely optimized in ways that Zillow’s are not. For example, I cannot search for homes on OpenDoor in Elizabethtown (where my parents live) or even in large nearby cities, like Louisville or Indianapolis. This suggests that OpenDoor is only flipping homes where it has a sufficiently high degree of in-market confidence.
To be fair, Zillow rolled out its Offers program on a market-by-market basis, but the difference is that Zillow never had to wager its life on flipping houses. Its core business model a marketplace to connect home buyers and agents. For that, the winning formula involves ubiquitous home and agent listings, so that even a shopper in Elizabethtown has a sense of their home value, their equity and what they might be able to purchase.
In short, I believe that mass market coverage and perfect price predictions are inherently opposed — they’re very different business models — and that has as much to do with Zillow Offers’ failure as anything else.
***
By the way, I am not being critical of Zillow’s home estimates. Zillow reports a median error rate of 1.9%, which is pretty crazy if you ask me. I only want to underscore how difficult it is to be successful in the iBuying space. Being off 2% on a $400,000 home means you’re off by $8,000 — well in excess of what OpenDoor generates in profit on each home they flip.
Additionally, none of this is to say that OpenDoor will be successful. If anything, writing this piece has increased my skepticism in the iBuy real estate movement. The trend has a lot to prove in the future.