Zillow Lost Traffic and Grew Revenue 18%. Here’s What That Tells Us About the Future of Real Estate

RealEstateAPI -

Last August, at Inman Connect, Clelia Warburg Peters said something that stopped me cold: it doesn't make sense to run a brokerage without ancillary services anymore.

I wrote it up the same week. The argument was simple. Commission margin was collapsing under volume pressure, agent splits, and NAR settlement fallout. Rocket's CEO had just put a number on the endgame — cut the average cost of a transaction in half, from $40,000 to $20,000.

And that number raised a question nobody at the conference seemed eager to answer out loud:

Where does that lost revenue get recaptured?

My answer was that brokerages would become holding companies. Title, mortgage, insurance, home services, data. Multiple profit centers, because the math on a $20,000 transaction doesn't work if commission is your only one.

Own more of the consumer relationship, or risk becoming a commodity onboarding service for agents.

Q2 2026 earnings came out last week. The recapture question has an answer now.

It's in the segment tables.

And it's more aggressive than what I described.

Zillow lost audience and grew revenue 18%

Start with the number that shouldn't be possible in a flat market.

Zillow's Q2 2026 revenue hit $772 million, up 18% year over year. Average monthly unique users fell 2%, to 239 million. Total visits fell 2%.

Fewer people showed up.

On a rough revenue-per-visitor basis, each one was worth about 20% more.

The composition is where the story lives.

For Sale revenue grew 14% — and the Residential piece of it, the core listings-and-leads business everyone in this industry means when they say "Zillow," grew 7%.

Mortgages grew 75%, to $84 million. Rentals grew 31%, to $209 million, with multifamily up 42%.

Seven percent in the legacy business. Seventy-five in mortgages. Same company, same quarter.

Mortgages and rentals together are now about 38% of company revenue. A year ago they were about 32%.

Six points of revenue mix in twelve months.

The housing market may be flat. But that almost misses the point.

Zillow isn't just outperforming the housing market. It's outperforming its own dependence on the housing market.

That is the ancillary thesis executed at scale — except Zillow arrived there without starting as a brokerage at all.

Rocket bought the demand instead of building the stack

Rocket ran the same play from the other end, and it's the cleaner proof.

It spent $1.75 billion on Redfin.

Not simply for the brand.

For the shoppers.

A year in, the mortgage attach rate on Redfin-originated purchase business is 47%, against a 50% synergy target management set at close. Redfin-driven mortgage leads doubled year over year in June. Rocket posted $2.78 billion in net revenue and its most profitable quarter in four years.

Read that attach rate again.

A buyer who comes in through the portal is now close to a coin flip to finance through the company that owns the portal.

That's the $20,000 transaction being funded.

Not by defending every dollar of brokerage economics, but by owning enough of the surrounding stack that you can afford to compress one part of the transaction because you have more places to make money across the customer relationship.

I said the math required multiple profit centers.

It does.

What I didn't say is who would be standing in the middle of them.

What I got wrong

I put the vertical integration at the brokerage layer.

It's happening one layer up.

The holdco isn't the brokerage.

It's the platform that owns demand.

And that is a materially harder problem for everyone downstream, because a brokerage that vertically integrates still has to go buy its customers — increasingly from a company that has already integrated vertically, owns the demand, and decides how that demand gets routed.

Vertical integration is powerful. Vertical integration plus distribution is something else entirely.

If I own title but you own the consumer relationship, I'm downstream from you.

If I offer mortgages but you decide which lender the buyer sees first, I'm downstream from you.

If I run a brokerage but my buyers increasingly originate inside somebody else's portal, I'm renting my distribution.

And whoever owns that distribution can keep adding ways to monetize the relationship.

The companies controlling the funnel are optimizing revenue per visitor, not visitor count.

Zillow's numbers show it working.

And nothing about a 47% mortgage attach rate suggests portal-sourced demand stays neutral.

The second signal nobody's reading

Everyone is going to write about the mortgage number.

The one I'd stare at longer is rentals.

$209 million in revenue, up 31%. Multifamily up 42%.

That makes rentals more than twice the size of Zillow's mortgage segment.

Why does that matter?

Mortgage still needs someone to buy a house.

Rentals don't.

Rental marketplaces, property-management workflows, resident services, and the data generated around them create recurring economic relationships that don't depend on transaction volume.

When a $209 million quarterly business growing 31% is also the one least dependent on home-sale transaction volume, that tells you where the stack goes after title and mortgage saturate.

A year ago, I listed data and software as the underrated pieces of the ancillary stack.

I'd revise that upward. As these platforms move from one-time transactions into recurring relationships, the data and workflow layer underneath stops being ancillary and starts being the business.

That deserves its own post, and it'll get one.

Where I land, a year later

The core call holds:

You either own the relationship or you rent access to it.

What changed is the scale of who's doing the owning.

I thought brokerages would respond to deteriorating transaction economics by becoming vertically integrated holding companies.

They are.

But the more important race is happening above them.

Zillow is moving downstream from search into financial services.

Rocket moved upstream from financial services into search.

They are approaching the same destination from opposite directions:

Own the demand. Own more of the stack. Monetize the relationship several times.

And if you're building downstream from those platforms, your strategic response can't simply be to add more features.

You need to think carefully about what you actually own.

Your customer relationship.

Your workflow.

Your data.

Because those are the things somebody upstream can't take from you — unless you architect your business in a way that gives them away.

That's obviously not a neutral observation coming from me. We built RealEstateAPI around the idea that companies should control the property-data layer underneath their products. But the case was on the record a year before these earnings releases confirmed it, which is the only reason I'm comfortable restating it now.

One more called shot, so this is checkable next August:

By Zillow's Q2 2027 release, mortgages plus rentals will represent at least 45% of total revenue.

If I'm wrong, it'll be right here, with the numbers.

Where does the $20,000 go?

It goes to whoever owns the customer after the click.

A year ago, that was a question.

It isn't anymore.

Sources