iBuyers were a scam and Zillow proved it

For a few years in the late 2010s, the iBuyer pitch dominated real estate tech. Companies like Zillow Offers, Opendoor, Offerpad, and RedfinNow would buy your house with an algorithm-generated cash offer in days, do light renovations, and resell it for a small spread. The story was that machine learning had cracked home valuation, that the friction of traditional sales was unnecessary, and that scale would bend the economics of real estate.

Then Zillow shut down its iBuyer arm in late 2021, took a half-billion-dollar write-down, and laid off 25% of its workforce. The model didn’t fail because of an unlucky market. It failed because the underlying premise was wrong.

The valuation problem

Zillow’s iBuyer thesis depended on Zestimate-style algorithms pricing houses with enough accuracy to systematically buy below true market value. In aggregate, the algorithms did okay. On any specific house, they often missed by enough to wipe out the entire spread.

Houses aren’t fungible. Two homes on the same block can differ in value by 15% based on layout, condition, finishes, neighbors, lot grading, and dozens of other factors that don’t show up in tax records or MLS photos. An algorithm trained on transaction history can capture averages but can’t see the leak in the basement, the dated kitchen, or the noisy intersection that a human walkthrough would catch in two minutes.

The companies tried to compensate with conservative offers, but conservative offers lost bids to traditional buyers. Aggressive offers lost money on the resale. There was no clean middle.

The market timing trap

iBuyers also took directional bets on home prices without realizing it. Buying inventory in volume means you’re long the housing market for the holding period. In a steady market, the model can squeeze a small spread. In a falling market, every house you bought last quarter is worth less when you sell. In a fast-rising market, sellers figure out they’re being underpaid and stop using you.

Zillow’s collapse coincided with rising costs and tightening price growth. Inventory they had paid algorithmic prices for in spring couldn’t be resold profitably in fall. The company had built what amounted to a leveraged real estate fund without admitting it, and when the trade went the wrong way, the loss was structural.

What survived

Opendoor still exists, but its market cap is a shadow of its peak, and its model has shifted toward agent partnerships and listing services rather than pure principal buying. The original iBuyer thesis, that pure algorithmic principal trading would replace traditional brokerage, is effectively dead.

What survived is the iBuyer-as-marketing-funnel idea: offer a quick cash bid as a way to capture a lead, then route most of those leads into a traditional listing relationship. That’s a real business, but it’s not what was sold to investors at the peak.

The takeaway

The iBuyer collapse wasn’t a market accident. It was the predictable result of a model that mispriced individual homes, ignored its own directional risk, and confused scale with skill. The next proptech promise that depends on algorithmic pricing of unique assets deserves the same skepticism.