Compass Claims A $5,590 Zillow Tax, But The Maths Not Mathing

Compass put out research this week saying homes shown on Zillow sold for 1.3% less than homes Zillow had banned. Within two days, three different groups of economists went after it — on the size of the sample, on missing controls, on the fact that every listing came from Compass’ own business. They called it the ‘Zillow tax.’

Every one of those complaints is fair. Not one of them found the real problem. The problem isn’t how they built the sample. It’s what they measured.

The caption on that post is flat wrong. Homes shown on Zillow did not sell for 1.3% less. A misstatement that big has another name, and plenty of people would use it.

Here’s why.

What they actually measured

Per Compass’ own research release, the company looked at 296,966 of its listings from January 2025 through May 2026. Of those, 806 were banned by Zillow. The banned homes came in at 100% of their asking price. The homes on Zillow came in at 98.7%. Compass took that gap, multiplied it against the price of a typical American home, and called it a “$5,590 Zillow tax.”

Now here’s the part that matters.

It does not measure how much money the seller walked away with. It can’t. Because the asking price isn’t a fact about the house. It’s a number the agent chose.

Let’s look at two identical houses, side by side

Picture two identical homes. Both are worth $1 million.

The first is listed at $1 million and sells for $987,000. That’s 98.7% of asking.

The second is listed with Compass as stated in their study, not on Zillow, at $950,000. Buyers pile in, and it sells for $950,000. That’s 100% of asking.

Compass’s measurement says the second house did better, however, Compass’s seller LOST $37,000.

The person being graded is holding the red pen

This is the whole problem, and no amount of extra data fixes it. Compass agents set the asking prices on both sides of this comparison. So Compass is reporting a score on a test Compass wrote. That’s a high school student writing his own final exam, taking it and then grading it. Of course he got an A.

And it gets worse, because the pressure only runs one direction.

When a home is marketed quietly, one agent sets the price with almost nothing coming back. No portal traffic. No first weekend of showings. No competing offers hinting the number is low. When you price in the dark, you price carefully — and careful pricing produces great-looking percentages.

The number Compass picked is the one a smaller pool of buyers would be expected to win all by itself — whether or not the seller came out ahead.

Underpricing and outperforming look identical on this measurement, and nothing in the study can tell them apart. It’s the same worry Gary Keller raised about listing exposure, coming at it from a different angle.

What a real study looks like

The question the industry needs answered is: Does limiting who can see a listing change what the seller gets, compared to what that home would have brought with full exposure?

Answering it takes an outside yardstick — an estimated value, an appraisal, a set of matched comparable sales. Something the company marketing the home doesn’t control.

Several studies do exactly that, and they don’t agree with one another. Zillow’s off-MLS analysis measured sales against estimated values, and so did Bright MLS’ comparison of on- and off-MLS results. The University of Georgia’s Dallas-Fort Worth study used a pricing formula and found an edge that mostly disappeared in the later years. You can argue about which one is built better, and people in the business already do.

Compass’s study isn’t a weaker version of those. It isn’t in the same category.

What leaders should demand

Brokerage owners, MLS boards and association leaders are making real decisions right now — on listing rules, on portal deals, on lawsuits — partly on research like this. That’s the world the unwinding of Clear Cooperation created, and there’s more of this research every month while the quality keeps slipping.

The standard should be simple. If a study claims to show what sellers gained or lost, it has to say what it compared against. If the answer is “the seller’s own asking price,” then it doesn’t show what sellers gained or lost. Any dollar figure has to come from an estimate of what the home was worth, not from a percentage multiplied against an unrelated number. And any comparison should say plainly what kind of homes it’s describing.

None of this requires picking a side. Some sellers genuinely want privacy, and the industry should be able to say so out loud. But it should say it with research that measures what it claims to measure — not by turning the MLS into a proxy for a business fight.

Right now, the most widely shared number in this entire argument measures the distance between a house and a number an agent made up. Whatever that is, it isn’t a tax.

Darryl Davis, CSP, is a national speaker, real estate coach, and the bestselling author of How to Become a Power Agent in Real Estate. Don’t miss this month’s free webinar series at PowerAgentWebinar.com. Through his POWER AGENT® Coaching Program, he helps real estate professionals build thriving businesses and lives at the Next Level®. Learn more at darrylspeaks.com.

This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners.

To contact the editor responsible for this piece: tracey@hwmedia.com