On Tuesday morning, August 4, 2026, just over five hundred people at Zillow Group learned they no longer had jobs. Their managers learned at the same moment they did. Nobody was pulled aside the day before, nobody got a conversation, nobody got a warning. In a company that has spent five years celebrating itself as a distributed, remote-first workplace it calls Cloud HQ, that is what a layoff looks like now. A message arrives on a laptop somewhere, and then access to everything goes away.
The cuts amounted to roughly seven percent of Zillow’s global workforce. They landed one day before the company reports second quarter earnings. And they came at the end of the most financially successful stretch in the company’s twenty year existence.
That last sentence is the whole story, so let me put the numbers behind it.
A Record Year
In 2025, Zillow generated $2.6 billion in revenue, up sixteen percent. It reported $622 million in adjusted EBITDA at a twenty four percent margin. It produced $420 million in free cash flow, up thirty six percent year over year. It ended the year with $1.3 billion in cash and investments.
And it posted $23 million in GAAP net income. That was Zillow’s first annual profit since 2012, when it was a much smaller pre-merger business, and it ended twelve consecutive years of losses running from 2013 through 2024, a stretch that includes the $528 million wipeout in 2021 when the company shut down its house flipping operation.
The first quarter of 2026 was better still. Revenue rose eighteen percent to $708 million against a residential real estate industry that grew two percent. Net income came in at $46 million, double what the entire prior year had produced, on diluted earnings per share of nineteen cents against three cents a year earlier. Adjusted EBITDA hit $182 million. Operating cash flow nearly doubled to $200 million.
This is not a company in distress. This is a company outgrowing its own industry by a factor of nine, generating cash at record rates, and telling investors every quarter that its strategy is working.
Then it cut seven hundred people in six months.
The Claim With No Evidence Behind It
Chief Executive Jeremy Wacksman explained the decision in a blog post. The cuts, he wrote, were about a disciplined cost structure and getting more efficient, with the right people in the right positions. Continuing to grow at scale, he added, requires the company to work differently than it does today.
Read that again and notice what is missing. There is no metric. There is no baseline. There is no target. There is no identification of which functions were redundant, which processes were consolidated, which tools replaced which work, or what the company expects output per employee to look like on the other side. There is no before and after. There is nothing a reader could check.
Getting more efficient is a testable claim. Efficiency has units. Revenue per employee is a number. Cycle time is a number. Support tickets resolved per head is a number. Listings processed, leads routed, code shipped, loans originated per loan officer, all of them numbers, all of them tracked internally with considerable sophistication, because Zillow is a data company that has spent twenty years building measurement infrastructure and reports operating metrics to Wall Street every ninety days.
Zillow did not publish any of them. It did not publish one.
I have spent close to thirty years shipping production software. I have run engineering organizations, sat in the meetings where headcount gets decided, and made the case both for and against specific roles. I know what a real efficiency argument looks like, because I have had to build them and defend them. It looks like this. Here is the work. Here is what it costs today. Here is the mechanism by which it costs less tomorrow. Here is the evidence that the mechanism works. Here is what we measure in ninety days to know whether we were wrong. Five sentences. Any competent operator can produce them in an afternoon.
When a company cuts seven percent of its workforce and will not produce those five sentences, the honest conclusion is not that the analysis is confidential. It is that the analysis was never the point.
The January Tell
There is a piece of evidence that settles this, and almost nobody has picked it up.
In late January 2026, Zillow cut approximately two hundred people. The company characterized those cuts as performance related, part of its normal annual review cycle. The framing was straightforward. These were underperformers, this is what a healthy company does, move along.
Zillow ended 2025 with 7,068 employees. It reported 7,058 employees as of March 31, 2026.
Ten. In the two months following a two hundred person performance reduction, Zillow’s total headcount fell by ten people. It had backfilled nearly the entire cut.
There are only two readings of that. Either the January terminations were not really about performance, and the company quietly replaced those roles because the work was real and needed doing. Or the company genuinely believed it had two hundred underperformers, removed them, and then immediately hired two hundred replacements at recruiting cost, onboarding cost, and ramp cost, which is a devastating indictment of its own hiring and management practices.
Either way, six months later it turned around and cut five hundred more.
That is not a workforce strategy. That is churn with a press release attached. And it tells you that the right people in the right positions language in Tuesday’s memo is not describing a considered organizational redesign, because the same organization ran the opposite experiment in January and reversed it inside of sixty days.
Six Hundred and Twenty Six Million Dollars
Here is the number that reframes the entire announcement.
In the first quarter of 2026 alone, Zillow Group repurchased 13.5 million shares of its own stock for $626 million. Not for the year. For the quarter. That single quarter’s buyback nearly matched the $670 million the company spent across the entirety of 2025.
Cash and investments fell from $1.3 billion at the end of 2025 to $788 million at the end of March.
Zillow spent down roughly forty percent of its cash reserve in ninety days buying its own shares off the open market, and four months later told five hundred people that the company needed to be disciplined about costs.
Zillow does not disclose average employee compensation, so what those five hundred jobs cost is an estimate rather than a fact, and I will show my work. Reported total compensation at Zillow clusters around $180,000, which fully loaded with payroll taxes and benefits puts five hundred people somewhere in the range of $100 to $125 million a year. Reasonable people can move that number around by twenty percent in either direction. It does not matter. At any defensible estimate, one quarter of share repurchases is roughly five years of payroll for every person terminated on Tuesday.
I want to be fair about the full capital picture, because the strongest version of this argument is the one that concedes everything it should. Zillow returned $1.1 billion to capital holders in 2025, of which $419 million was repayment of convertible debt. That is deleveraging, not extraction. The buybacks also retired real shares rather than merely offsetting grants, with diluted share count falling from 256 million a year ago to 240 million at the end of Q1. That is a genuine return to owners. And at a depressed share price, buying back stock is a defensible allocation decision that many serious investors would applaud.
I am not arguing the buyback was irrational. I am arguing something narrower and harder to dodge. A company that voluntarily spent forty percent of its cash reserve on share repurchases in a single quarter cannot credibly describe a layoff four months later as a matter of cost necessity. It was a choice about who gets the money. The choice is legible in the cash flow statement, and it was made long before anyone drafted Tuesday’s memo.
The Policy They Announced in February
Here is the part that turns this from an opinion into an argument, and it has been sitting in public since winter.
On the fourth quarter 2025 earnings call in February, Chief Financial Officer Jeremy Hofmann described the company’s cost architecture in plain language. Ninety percent of Zillow’s share based compensation, he said, is allocated to fixed employees, and the company was seeing increased leverage as it strives, in his words, to keep fixed headcount relatively flat. In the same set of materials, Zillow noted that its $670 million of 2025 repurchases more than offset share based compensation expense for the year, bringing ending share count down two million shares.
Read those two statements together and the entire machine becomes visible.
Equity compensation creates dilution. Buybacks retire that dilution with cash. Holding headcount flat holds the equity grant pool flat, which slows the dilution, which reduces the cash required to offset it, which frees capital for more buybacks. In this model an employee is not primarily a producer of work. An employee is a line item in a share count calculation.
Now put the numbers on it, because they are all disclosed.
Share based compensation in 2025 was $390 million, down thirteen percent year over year. Depreciation and amortization was $264 million. Together those two non-cash items account for nearly the entire gap between the $622 million of adjusted EBITDA the company celebrates and the $23 million of GAAP net income it actually earned.
Ninety percent of that $390 million, roughly $351 million, was allocated across a fixed employee base of about seven thousand people. That is approximately fifty thousand dollars per employee per year in equity expense alone.
Five hundred employees is therefore on the order of twenty five million dollars of annual share based compensation, removed permanently, in a single action.
And Zillow has told investors it will reduce share based compensation by more than fifteen percent in 2026, an outlook it raised in May from a prior commitment of more than ten percent. Fifteen percent of $390 million is roughly fifty nine million dollars. Tuesday’s cut plausibly delivers close to half of that target by itself, with a full quarter left in the year for it to show up in the reported numbers.
That is a coherent, defensible, entirely legible financial rationale. It is also completely different from getting more efficient. Efficiency means the same work gets done with fewer people. This is the work getting smaller, or getting absorbed by whoever is left, in service of an expense line the chief financial officer committed to on a conference call in February.
I would have considerably more respect for a memo that said so. Zillow could have written it. The company already told Wall Street the strategy out loud. It simply declined to repeat it to the people it applies to.
The AI Dodge
There is one more piece, and it is the part that will age worst.
Zillow has spent 2026 telling investors it is becoming an artificial intelligence company. In April, Wacksman told a room of real estate executives that Zillow employees were being retrained to use AI in their jobs, and that the gains, while small, were compounding. In May, on the first quarter earnings call, he told investors that Zillow is rapidly becoming an AI native company.
On Tuesday, reporters at GeekWire and Real Estate News both asked the obvious question. Did AI have anything to do with these cuts? Initially the company said nothing at all. Later in the day a spokesperson stated that the layoffs were not connected to Zillow’s adoption of AI, and were instead about better positioning the company for the path ahead.
So the company is rapidly becoming AI native, is actively retraining its workforce on AI, is reporting compounding efficiency gains from that retraining, and eliminated seven percent of its workforce in a record quarter. And none of these facts are related to one another.
I want to be fair here, because this is a genuinely difficult position for any executive. Saying that AI enabled five hundred job cuts invites regulatory attention, guts morale among the survivors, and hands every competitor a recruiting pitch. The incentive to deny is enormous and I understand it completely.
But notice what the denial costs them. AI driven productivity was the one explanation available that would have made this look like efficiency rather than arithmetic. By ruling it out, Zillow removed its own best defense. What is left is a company that cut five hundred jobs in a record quarter for reasons it declines to specify, having spent $626 million on buybacks in the same period, against a share based compensation target it announced in February.
If AI did drive the cuts, the company is not being straight with anyone. If AI did not drive the cuts, then the AI native transformation the chief executive has been selling to Wall Street all year has produced no measurable operational leverage whatsoever, and somebody on the next earnings call ought to ask about that.
Both branches are bad. That is what happens when you build a narrative for investors that you cannot repeat to reporters.
What They Are Actually Cutting Into
There is a strategic dimension here that deserves attention, because it is the part where I think Zillow has a real problem and this layoff makes it worse rather than better.
Look at where the growth is actually coming from. In the first quarter of 2026, Zillow’s Residential business, the legacy Premier Agent lead generation engine that has funded everything else for a decade, grew eight percent to $450 million. Mortgages grew fifty six percent. Rentals grew forty two percent to $183 million, with multifamily up fifty seven percent.
Both of the fast growing segments are more labor intensive, not less. Zillow told investors explicitly that costs were rising in the first half of 2026 because it was hiring rentals salespeople and loan officers for Zillow Home Loans. You cannot originate mortgages without loan officers. You cannot sell multifamily advertising packages without a sales force.
The company has not said which teams absorbed Tuesday’s cuts, and I am not going to pretend otherwise. But it is difficult to reconcile a five hundred person reduction with a stated hiring push in Rentals and Mortgages. The arithmetic points at the legacy Residential organization and the shared functions supporting it.
Which is precisely the part of the business that is under attack.
On June 11, Google took its home listing ads national, expanding enhanced Local Services Ads across all fifty states after a pilot in eight markets. Property details, pricing and photos render directly inside mobile search results, with a button to call, message or book an appointment with an agent without ever leaving Google.
I want to be precise about this, because the headline overstates it and the overstatement is how bad arguments get discredited. Eligibility in fifty states is not inventory in fifty states. The listing data is supplied by HouseCanary, and as of the national announcement only three multiple listing services were participating: the California Regional MLS, San Diego MLS, and My State MLS. The format is mobile only. And it is a paid advertising product, not an organic listings page. Google is not running a portal.
But look at what the product actually is, because that is the part that should keep Zillow awake.
Enhanced Local Services Ads for home listings is Premier Agent. It is the same business, feature for feature. An agent pays for placement, a buyer with intent sees a property, the agent gets the call. The only difference is that Google’s version sits one layer higher in the funnel, at the search box, where the buyer already is. Zillow’s twenty year franchise was being the place people went immediately after Google. If the connection completes before that click, the franchise is what is at risk, not the traffic figure in any particular quarter. The market understood this immediately, and Zillow fell nearly five percent intraday on the announcement.
Meanwhile Zillow’s own audience numbers are pulling in two directions, and both belong in this piece. Traffic to Zillow’s mobile apps and sites in the first quarter was down three percent year over year to 220 million average monthly unique users, with visits also down three percent. In the same press release, Zillow reported that Comscore measured its average monthly unique visitors up twelve percent year over year to 127 million, and stated that it was the only large company in its category to grow its share of the real estate audience over the previous six quarters.
Those are different methodologies measuring different populations, and you can argue about which one matters more. What is not arguable is that the first party number, the one Zillow controls and has reported for years, went negative. And it went negative in the first quarter, months before Google’s national rollout, which means whatever is happening to the top of Zillow’s funnel started before the obvious villain arrived.
So the company just reduced staffing in the division facing an existential competitive threat, in order to protect margins in the divisions that are growing but carry an active federal antitrust trial. The Federal Trade Commission and the attorneys general of Virginia, Arizona, New York, Connecticut and Washington go to trial on August 24 in Alexandria, over the $100 million payment Zillow made to Redfin to become the exclusive provider of multifamily rental listings on Redfin’s platforms. That is the Rentals business. That is the growth engine.
Maybe that is the right bet. It might be. But it is a bet, it is a large one, and it is not what getting more efficient means.
What the Room at the Top Costs
I do not believe executive compensation is inherently scandalous, and I have no interest in the reflexive version of this argument. Building a company is hard, running one at scale is harder, and the people who do it well should be paid well.
But numbers are numbers, and there is one structural point here that most coverage misses.
Remember that $390 million of share based compensation, ninety percent of it allocated to fixed employees. Executive equity comes out of that same pool. The layoff was, on the company’s own stated logic, an action to shrink it.
As reported in Zillow’s April 2026 proxy statement, Chief Executive Jeremy Wacksman’s 2025 total compensation was approximately $7.1 million, down from $15.7 million in 2024, the year he was elevated to the top job and received the accompanying grant. Chief Operating Officer Jun Choo received approximately $9.6 million. Co Executive Chairmen Richard Barton and Lloyd Frink received approximately $6.9 million and $6.8 million. Chief Financial Officer Jeremy Hofmann, the executive who described the headcount policy, received approximately $5.5 million.
Those five packages total roughly $35.9 million. Zillow’s entire full year 2025 GAAP net income was $23 million.
In the same filing, Zillow disclosed that it paid roughly $531,000 chartering aircraft from entities controlled by Barton and Frink, and that it employs both founders’ sons at combined compensation of approximately $583,000.
Now the caveats, because I want this argument to survive contact with a hostile reader. Most of that executive compensation is grant date fair value of equity vesting over four years, and Zillow’s stock is down roughly forty five to fifty percent this year depending on your measurement date, so what those executives ultimately realize will be materially less than the headline figures. The GAAP net income comparison is also unfair on its own terms, since that $23 million sits beneath $390 million of stock compensation and $264 million of depreciation and amortization. Measured against $420 million of free cash flow instead, five named executives at $35.9 million is roughly eight and a half percent, which is unremarkable for a company this size.
Take the most generous framing available, then. Zillow still awarded its five named executives more in a single year than it will save annually by terminating five hundred people. It chartered private aircraft from its own founders. It employed their children. And it drew that executive equity from the very pool it told the workforce had to shrink.
The word discipline is doing a great deal of work in that memo, and it is not being applied evenly.
The Part That Is Not About Numbers
Five hundred people. Mortgages, tuition payments, medical coverage. In a technology workforce this size there will be visa holders among them, and a terminated H-1B employee has sixty days to find a new sponsor or leave the country. These are the people who shipped the products that produced the eighteen percent growth rate the chief executive cited in the opening paragraph of the memo announcing their termination.
Their managers learned when they did. That detail is the one I keep returning to. It means no manager anywhere in that company was permitted to advocate for a single person on their team, because no manager knew the list existed. The decision was made entirely above the level of anyone who had ever watched these people work.
There is a version of this that is defensible. Companies do have to shrink sometimes. Markets change, bets fail, strategies get revised, and nobody serious believes employment is permanent. But the minimum obligation a company incurs when it does this is to be honest about why, and to demonstrate that somebody actually did the work of figuring out what could be removed without breaking the thing.
Zillow has not met that bar. It produced a paragraph of language that could be pasted verbatim into any layoff announcement at any company in any year. The real explanation exists, it is quantified, and it is in the February earnings transcript. They just would not say it to the five hundred.
How to Shop for a Home Without Zillow
Now the constructive part, because complaining is cheap and the alternatives are genuinely better than most people realize.
First, understand what Zillow actually is. It is not a listing service. It does not own inventory and it does not generate listings. Nearly everything you see on Zillow arrives from the local Multiple Listing Service, the broker controlled database that is the actual source of truth for what is for sale in your market. Zillow ingests that feed, wraps it in an interface, and monetizes your attention on top of it. When you click contact agent on a listing, you are usually not contacting the listing agent. You are being routed to a Premier Agent, meaning an agent who purchased advertising in that ZIP code. You are the product, and the price of you is set by auction.
That single fact reframes everything. If the underlying data is the MLS, then you can get the MLS without the toll booth.
The most direct route is a local brokerage website. Almost every real estate brokerage in America runs an IDX feed off the same MLS that Zillow pulls from, which means a small independent broker’s site in your town shows you the same inventory, often faster, without selling your contact information to the highest bidder. Search for brokerages operating in your target neighborhood and use their site directly. The data is identical. The incentive structure is not.
Realtor.com is the closest thing to a direct competitor and has a structural advantage worth knowing about. It operates under agreement with the National Association of Realtors and receives listing data through direct MLS relationships, which historically has meant faster status updates and higher listing accuracy than portals working from secondhand feeds. Homes.com, operated by CoStar, has built its entire pitch on the opposite of the Premier Agent model. It routes inquiries to the actual listing agent rather than auctioning them off, which if you are a buyer means you are talking to the person who has been inside the house. Redfin remains a serious search tool, though be aware that since Rocket Companies acquired it in 2025 it carries its own vertical integration incentives around mortgage.
Google’s new home listing ads are worth watching but not yet worth relying on. The format is live in all fifty states, but the actual listing inventory currently flows from only three participating MLSs, it is mobile only, and it is an advertising product rather than a search index. In a year it may be the best free tool available. Today it will show you a fraction of what is on the market in most of the country.
For anything the portals systematically underrepresent, go around them entirely. New construction is the clearest example. Builders pay Zillow for placement in its New Construction marketplace, which means you are seeing the builders who paid, not the builders who exist. Go directly to the national builders’ own sites, to your regional builders’ sites, and to your county or city permitting portal to see what is actually breaking ground. For distressed and foreclosed inventory, the government channels are public and free. HUD Home Store carries FHA foreclosures, Fannie Mae runs HomePath, Freddie Mac runs HomeSteps, and your county trustee or sheriff’s sale listings are a matter of public record. For sale by owner inventory lives on Facebook Marketplace, on Craigslist, and on dedicated FSBO sites, and much of it never touches an MLS at all.
Then there is the data itself, which is the part most people never think to check. Your county assessor and county recorder publish, for free, the ownership history, the tax assessment history, the recorded sale prices, the lot dimensions, and the permit history for every parcel in the jurisdiction. That is primary source material. It is more reliable than anything a portal will show you, because it is the record from which the portal’s data is derived.
Which brings me to the Zestimate, and this is worth internalizing before you make the largest purchase of your life. By Zillow’s own published data, the nationwide median error rate for the Zestimate is roughly two percent for homes actively on the market and roughly seven percent for homes that are not listed. The on market number sounds impressive until you read Zillow’s own methodology, which notes that once a home is listed the model incorporates the listing information, including the list price. In other words, the accurate version of the Zestimate is the version that has already seen the answer. The number you get when you look up a house that is not currently for sale, which is most houses most of the time, carries a median error around seven percent. Median means half of all such estimates are off by more than that, with substantially worse results in low transaction markets. On a $600,000 home, seven percent is $42,000, and half the time it is worse than that.
The replacement for a Zestimate is a comparative market analysis from an agent who has physically been inside comparable properties, or a licensed appraisal. Neither is free. Both are correct.
Finally, and most importantly, hire your own buyer’s agent, chosen by you, before you start clicking on listings anywhere. Interview three of them. Ask how many transactions they closed in your specific target area in the last twelve months. Ask them to walk you through a deal that fell apart and what they did about it. An agent you selected on the merits works for you. An agent assigned to you because they bought your ZIP code works for the portal that sold you.
Wednesday Afternoon
Zillow reports second quarter earnings on Wednesday afternoon. Watch three things.
Watch whether back half margin guidance gets raised on the strength of Tuesday’s cut, which would confirm the layoff was a guidance instrument rather than an operational decision. Watch whether the share based compensation outlook improves again, from down more than fifteen percent to something steeper, which is the single cleanest confirmation of everything above. And watch whether restructuring charges are deferred into the third quarter, which would confirm the timing was chosen to keep them out of this print.
Then watch whether a single analyst asks the chief executive to quantify the efficiency he claimed on Tuesday.
I would bet against that last one. Analysts do not usually ask, because the answer does not change the model. The margin improves whether or not the company can explain it.
That is the actual lesson here, and it has very little to do with Zillow specifically. We have arrived at a place where a public company can cut seven percent of its people in the best quarter it has ever had, offer no operational evidence of any kind, and face no consequence. Not from the market, not from the press, not from the analysts whose entire job is to ask hard questions. The word efficiency has been fully decoupled from measurement. It now means, simply, fewer people.
Five hundred families found that out on a Tuesday morning, by email, on a laptop in a spare bedroom.
About the Author
Gal Ratner is the founder and CTO of Inverted Software and WhiteStar Labs, and Chief Architect at Prana Entertainment. He has nearly thirty years of production software experience on the Microsoft and .NET stack, with clients including Microsoft, Sony, Rockstar Games, 2K Games, Best Buy, and Allegiant Air. He was employee number six at Break.com and a finalist for the Los Angeles Business Journal’s CTO of the Year. His current work centers on production agentic AI, including MCP servers, the Microsoft Agent Framework, RAG pipelines, SQL Server 2025 vector search, and the PLogger observability framework. He also builds and operates ShopSnap, a zero transaction fee e-commerce platform, alongside several other production properties. He writes about what actually ships, and what merely gets announced.
Ridiculous argument. The CEO is not obligated to show her math work to the public. No company does that.