Part IThe Story
Seven Thousand Roofs
On the afternoon of November 2, 2021, the company Rich Barton had co-founded owned roughly seven thousand houses it no longer wanted. They had been bought one at a time, each priced by an algorithm whose ancestors had launched fifteen years earlier on a single promise: that a computer could look at any house in America and tell you what it was worth. Now many of those houses were being relisted at a loss. The division that bought them, Zillow Offers, had lost $420 million in the third quarter alone. That afternoon Barton told shareholders and employees that Zillow would stop buying homes, sell the inventory, and cut about a quarter of its workforce. The stock fell nearly 25 percent.
The letter he posted that day is the most honest document a public-company CEO of his generation has produced about his own mistake. It is also a strange document, because of what it concedes. Zillow had gone into home-buying, Barton wrote, intending "to become a market maker, not a market risk-taker," and the whole enterprise rested on one capability: forecasting the price of a home accurately three to six months into the future. The company had run "countless simulations." It had set itself unit-economics targets within plus or minus 200 basis points of breakeven and promised investors it would stay inside them. It had not stayed inside them. "Put simply," he wrote, "our observed error rate has been far more volatile than we ever expected possible."
Error rate. Fifteen years earlier, sitting in a borrowed office in Seattle, Barton and his co-founder had arrived at an image that would make them famous: an aerial map of an American neighborhood, zooming and panning the way Google Maps had just taught people to zoom and pan, and floating above every house, a number. "We wanted prices on every roof," Barton would say. "Because everything should have a price."
Nobody in the history of consumer technology has more consistently bet on that idea, and very few people have won more often with it. Expedia showed people what plane tickets cost. Zillow showed them what houses were worth. Glassdoor showed them what their coworkers were paid. Each time an industry had hoarded a number, Barton pried it loose and handed it to the public, and each time the public rewarded him for it. Then he tried to buy the roofs himself, and the number turned out to be a different kind of thing when it was your own money underneath it.
The piece that follows orbits that November afternoon. It moves backward to a phone call with a travel agent and a man painting houses in Connecticut; forward to a Saturday Night Live sketch, a lawsuit over a twenty-four-hour rule, and a climate score that lasted about fourteen months. It keeps returning to the number on the roof: who gets to see it, who gets to set it, and what happens to the person who has spent his life insisting that everything should have one.
By the Numbers
The Barton Ledger
$3.6BPrice Barry Diller's company paid for Expedia, completed in 2003
1MZillow users in the first three days after its February 8, 2006 launch
79%Zillow's first-day stock gain on Nasdaq, July 20, 2011
$2.5BZillow's acquisition of rival Trulia, closed February 2015
~7,000Homes Zillow owned when Barton shut down iBuying in November 2021
$420MZillow Offers' loss in Q3 2021
~10BTimes Zillow's apps and sites were accessed in 2020
The Voice on the Other End of the Phone
The origin story Barton tells most often doesn't start with a technology. It starts with the sound of one. In the early 1990s he was a young Microsoft product manager who traveled constantly, and every trip meant calling a travel agent. On the other end of the line he could hear the agent typing, querying a system he knew existed and could not see, translating his preferences into keystrokes and the keystrokes into whatever the screen said back.
— Rich Barton, to Wired, 2013I remember I wanted to jump through the phone and look at the screen myself, turn it towards me and just take control. And I knew that I would spend more time and do a better job searching than this person who was doing something on my behalf.
Turn it towards me. Put a geometry to Barton's entire career and it looks like this: a screen facing a professional, slowly rotated until it faces the customer. He didn't build the reservation systems, or the multiple listing services, or the salary databases. He built the swivel.
The swivel arrived in 1994, and the way it arrived tells you nearly everything about the man. Microsoft was then in its Encarta phase, convinced that the CD-ROM would become the encyclopedia, the atlas, the field guide and the reference shelf of the American home. Barton had recently moved from the Windows 95 team into the small consumer division, where he inherited what he later called a portfolio of "rando ideas." One was a multimedia travel guide: a whole shelf of guidebooks pressed onto one disc, with pictures and audio. By one account the project was already called Expedia.
He went into his first product review with Bill Gates at twenty-six and told the chairman of Microsoft that the idea in front of him was, as Barton put it to Fortune, "dumb, small, and uninteresting." He had numbers. The entire U.S. travel-book market was maybe $100 million. Coming from the operating-systems group, where business plans ran into the billions, it looked like a rounding error. And there was a deeper flaw: a travel guide is most useful while you're traveling, and in 1994 the laptop was, in his words, "this compact suitcase that weighed 20 pounds." Nobody was going to haul one through the Uffizi.
Then he did the thing that separates the critic from the founder. He demoed something better. On Prodigy, the early online service, he had found easySABRE, a tool American Airlines' Sabre unit had built so travel agents could reach airline reservation systems from home. It wasn't meant for consumers. Barton, a self-described geeky online guy, had gotten in anyway. He showed it to Gates, who, it turned out, already knew about it. "This is a change-of-the-world thing," Barton told him. If consumers could do this themselves, Microsoft could become the largest seller of travel in the world. Nobody yet called this e-commerce.
The pitch had a second half, and it was personal. His fiancée, Sarah, was a medical student at Northwestern applying for OB/GYN residencies, and Seattle had exactly one program, six or seven residents a year, among the most competitive in the country. Barton assumed she wouldn't match there. He assumed they'd end up in New York. So he asked Gates to fund the travel idea outside Microsoft, as a venture, and argued that it didn't belong inside the company anyway: "It's travel first, software second. Not software first, travel second."
Gates and Ballmer, Barton recalled, "kind of laughed at me." Then they said yes, with an amendment. Do it here. We'll build a team. "I'm sure Sarah will match her residency at University of Washington." She did. Barton still doesn't know whether anyone put a thumb on the scale, and you can hear him enjoying not knowing. He stayed in Seattle and became general manager of a thing that didn't exist yet. Gates, he likes to say, was his first venture capitalist.
There is a version of this story that is about vision, and there is a truer version that is about nerve, and the truer version is mostly about the size of a market. The CD-ROM was a real product solving a real problem. It was just a small pond. Barton's later formula — "Is it a big pond, and are there good fishermen?" — fits on a cocktail napkin, and he arrived at it by telling the richest man in the world to his face that the pond he'd been handed was a puddle.
A Client, Not a Server
He had been running small businesses since he was old enough to drive. Barton was raised in New Canaan, Connecticut, the son of a mother who taught school before she married and, as he puts it, "became the CEO of the family," and a father who was a mechanical engineer with big sales jobs at Union Carbide and who liked to remind his son that in his day plastics and chemicals were high tech. Barton was a self-described "closet geek," one of the first kids in the neighborhood with a Radio Shack TRS-80. The family line ran back to the Midwest; two ancestors, John Barton and his son Horace, sit in the South Dakota Tennis Hall of Fame. These details don't predict anything. Suburban Connecticut in the late 1970s produced a great many tidy boys with early computers. What it produced less often was a sixteen-year-old ice cream man.
He drove the truck himself. He priced the merchandise, worked out the daily route, and figured out how much inventory of ice cream and Popsicles to carry on a hot afternoon, which is a working course in demand forecasting for a teenager. He caddied at the local country club for two years. Then he joined a buddy's house-painting crew for a summer, looked at the economics, and concluded that he could bid the jobs himself. For the next couple of summers during college he ran his own painting company with his own crew and cleared $15,000 to $20,000 a summer, which in the late 1980s, he says, "was a boatload of money." His parents paid his tuition. Everything else was on him.
"It taught me how to run a business, hire people, manage clients, do quality work," he told the New York Times in 2001. "I fell in love with running my own thing. And painting made me feel really good, especially when I was finished."
That last clause is easy to skip, and it shouldn't be skipped. Especially when I was finished. The pleasure was in the completed wall.
At Stanford he started out in industrial engineering, a discipline his mechanical and electrical classmates dismissed as "imaginary engineering" because its students cared too much about business. He chose it, he says, because it was a bachelor of science, and because it was the most business-y of the engineering tracks. Junior year he left for Stanford's program in Florence, having fallen for Italy on a high school trip with his Latin teacher: "The food, the wine, the girls, just the family culture, the work is not as important." When he came back, the department told him he'd fallen off the track for an accredited degree. "I don't care," he said. A friend's father, a professor, pointed out that Stanford let students design their own engineering degrees. So Barton designed one. His diploma reads general engineering: industrial economics, a major with exactly one graduate in its intended form.
The friend was Nina Roberts, and she matters to this story more than her brief appearances suggest. She was an engineer in Barton's class, and the two of them interviewed for the same jobs senior year: strategic planning at Disney, the investment-banking analyst programs, the strategy consulting firms. Barton, who calls himself the "success kid," went for whatever was hardest to get, and landed at Alliance Consulting Group, a Boston Consulting Group spinoff in Cambridge, just as the 1989 recession began eating consulting jobs. Roberts didn't get the big job. She got an offer from a little tech company in Seattle called Microsoft.
He knew within months that consulting was wrong for him. The trite explanation, he says, would be the suit and the uncomfortable shoes ("I like to go barefoot"), and that wasn't it. He liked presenting to CEOs. He got what amounted to an MBA in Michael Porter's competitive strategy in one year, a stint another profile called his "business boot camp." What he discovered was a taxonomy. The world, he decided, divides along something like a barbell of clients and servers, a software-architecture metaphor he applies to human temperament. Lawyers, doctors, consultants and academics serve. They get variety and curiosity, and they rarely see anything through to the end. "I actually loved what the clients were doing more than what we were doing," he said. He was a client. He wanted the finished wall.
By one account he interviewed at Patagonia and Orvis, the fly-fishing outfitter, trying to fold his love of the outdoors into a paycheck. Meanwhile Roberts called from Seattle every week. You've got to come out here. It took about a year to get the flight and the offer. In 1991 he joined Microsoft, then about 3,000 people. "It was like I was a kid in a candy store," he said. "I never left the place."
Except that he did, twice, and the second time he took 150 people with him.
His first boss there was Brad Chase, a group product manager in what Barton remembers as "this elite little group of really smart people." Barton's job on MS-DOS 5 was unglamorous and physical: design the box, run the manufacturing, get the product into Egghead, the software chain. Then he had an idea. DOS for Dummies, he noticed, was selling millions of copies in Barnes & Noble and Borders, the great bookstore palaces of the era. Why not bundle the upgrade with the book, make the book the manual, and sell the package where the readers already were?
He met with one of the Riggio brothers at Barnes & Noble and with the creator of the Dummies series. He designed the product and built inventory at eight to ten dollars a unit, heavy for cost of goods. He felt, he admits, "pretty like a big deal." The package looked too much like a book. It sat on bookstore shelves next to twelve-dollar paperbacks with a price tag of $49 or $54, and customers didn't realize there was software inside. He grades it "at best a C. Maybe a D." Microsoft ate the inventory. He reckons the mistake at ten to twenty million dollars.
He went into his performance review expecting, at best, a lecture. Chase looked at him and said: "All right, what's your next big idea?"
Barton still has one of the boxes.
The Spouses Were the Toughest Investors
Expedia went live on the web in 1996, from an office for 143 people down the road from Microsoft headquarters. On the other side of the continent, Steve Kaufer and Langley Steinert's first TripAdvisor office sat above Kosta's Pizza & Seafood. The contrast is the point. Barton was building a startup with a monopolist's checkbook, and the checkbook came with a monopolist's reputation.
The hard part, he said later, was never the consumer. "We were so cool and interesting that consumer adoption and word of mouth were not a big challenge." The hard part was the suppliers. "Microsoft was powerful, and having young kids from the company go into offices to negotiate with old guys at American Airlines didn't sit well with them." So Barton did what a painter does on a client's porch. "I wore a suit, tried not to be cocky. I built relationships based on demonstrated expertise. If you challenge people to think about business in a new way, they tend to forget you don't look like" — and here the quote trails into the familiar shape of a young man who has learned to make older men feel they discovered him.
Ballmer was his boss by then, and Barton's method with Ballmer was theater. Early on he'd walked into Ballmer's office carrying a homemade business card: a travel-agent card with Ballmer's face on it. "Steve, do you want to be the largest travel agency in the world?" Barton still does the impression of the answer — Nooooooo! — in the gravelly bellow. Ballmer kept the card anyway. Barton's explanation: he had cropped the photo so that "a little bit of my hair was coming over the top, so he had some hair."
By 1999 the bubble had made the theater unnecessary. Barton went back to Ballmer and asked for $100 million for a television campaign. Expedia was in the pole position, he argued, and could build the biggest brand in travel. Ballmer laughed. We don't do that. Barton pointed at the public markets, where, as he remembers it, dogshit.com was going public at a billion-dollar valuation. (It was more like $500 million, he concedes, and the company was probably called something else.) The markets would give Expedia the money for nearly nothing. Let me spin it out.
He didn't pitch the spin-out as finance, though the finance was real: the conglomerate discount meant the market wasn't valuing Expedia properly as one more line inside Microsoft. He pitched it as an HR experiment. Microsoft was getting so big, he told Ballmer, that people could "hide out in random corners," collecting option gains as long as Windows NT and Office kept winning, a "compensation-accountability disconnect." Meanwhile the most entrepreneurial people in the building, people like Barton, were looking at the Field of Dreams outside and would leave. Spin out a company and you keep them by giving them a stake in something they could actually move. Ballmer understood. "Yeah, we'll take a flyer on it."
It was, Barton has said, a feat that had never happened at Microsoft and hasn't happened since. "Microsoft doesn't divest things," he told Wired. "One of its great strengths, and possibly a great weakness, is it has a monolithic strategic world, nothing isn't strategic, everything fits the grand plan."
Greg Maffei, Microsoft's CFO, became Expedia's chairman. Maffei runs through the rest of Barton's career as a mentor, board member and source of maxims; his best-known contribution is the three-question test for joining a board ("Is it local? Is it fun? Is it lucrative?"), of which any good opportunity must satisfy at least two. Expedia offered its Microsoft team the choice to come or stay. Of the 150 people asked, all but about two came.
What persuaded them wasn't the employees' appetite. It was their marriages. "The hardest sell was not the people, it was the spouses," Barton said. He remembers several dinners at Wild Ginger in Seattle with a draft S-1 on the table, highlighted and annotated by a skeptical husband or wife walking him through the risk factors. "Those were my toughest investors, actually."
The roadshow ran in November 1999: fifteen cities in three weeks, five or six meetings a day, a chartered plane carrying the bankers, the CFO and a thirty-two-year-old CEO, "exhausting, and exhilarating, and repetitive, and kind of boring." The book was filled thirty times over. Sarah, pregnant with their first child, wasn't due until December. With two and a half weeks done and the deal plainly working, Barton called the roadshow off a day early and flew home. He got into bed at 1 a.m. At 3 a.m. his wife, an obstetrician who knew exactly what was happening, tapped him on the shoulder.
"Honey," she said, "if our baby is born on the IPO day, do I really have to name him Expedia?"
He had promised the team. They went to the hospital. While Expedia traded for the first time, their son Will was born. Asked years later to name the high point of the whole adventure, Barton didn't pick the sale or the money. He picked that night.
The HR experiment, he notes with relish, "kind of failed." Expedia was a real business, profitable and growing, one of the few recent internet IPOs that survived the crash; investors fled into quality, and Expedia was quality. Microsoft went over the cliff with everyone else, and it took seventeen years for its stock to get back to its November 1999 price. The best people at Microsoft, it turned out, would have been better off at Expedia.
Then the conglomerate he'd escaped was replaced by a different one. Microsoft owned roughly two-thirds of Expedia and didn't even sit on its board. When someone new arrived in Redmond preaching focus, the easiest focusing move available was a big offer from Barry Diller, who was assembling an interactive empire that became IAC. Barton credits the strategy to Diller's young deal lieutenant, Dara Khosrowshahi, later CEO of Expedia and later still of Uber, who saw that online travel was already large and about to get much larger and wanted to consolidate it. Diller bought Microsoft's stake, then bid for the rest. Expedia was folded into IAC alongside Hotels.com; Wired puts the completed deal at $3.6 billion in 2003. Barton stayed as CEO until March 2003 and left about a month before the merger closed.
"I didn't control the company," he said. "I learned that henceforth I would control the companies that I started."
Twenty-Seven Weeks
Something else happened in those years, and he tends to tell it quietly.
He was thirty-four or thirty-five, a very young public-company CEO who had also just had to steer a travel business through September 11. He'd been, in his words, "not quite sleep under the desk, but work all the time." The Bartons socialized with Microsoft people and then Expedia people and talked about work at dinner. He played weekend basketball and tennis and snowboarded, and had not yet noticed that a body has to be maintained to keep doing those things.
Sarah was pregnant with twins, which is high-risk by definition. On a drive north toward Whistler, at about twenty-seven weeks into a forty-week pregnancy, she said she thought something was going on and asked to stop by the hospital so her OB could take a look. She was partly dilated, with small contractions. She was ready to get back in the car and recline the seat for the rest of the drive (the shoemaker's children, Barton says, have no shoes). Her doctor, Edith, told her she wasn't going to Whistler and she wasn't going home. She was being admitted.
For six weeks Sarah stayed in the hospital on muscle relaxants while Barton looked after their three-year-old, Will. She carried the babies to roughly thirty-five or thirty-six weeks. Josie and Russell were born healthy. "In the course of that period of time," Barton said, "it got me to just reassess my life and how I live my life and what my priorities were." He decided to quit.
He didn't start another company. He moved his family to Florence.
For a little more than a year they lived the life most founders only describe on podcasts. He took Italian classes in the mornings and charcoal life drawing, with the model changing poses every ten minutes. He took up road biking — "little cap, of course. No helmet" — which he found was also the easiest way to make Italian friends. Nina Roberts lived there by then, married to a European, and the woman who'd run Stanford's Florence program when Barton was a junior was still around. The family spent a few months skiing in the mountains. He thought he might write books, or make art or music. He had enough money for the rest of his life.
He didn't find his next calling. That's the plain truth of the Italian year, and he says it plainly. He was still on boards: IAC's, and Netflix's, which he had joined before its IPO and where Reed Hastings had once answered his question about mailing DVDs with "Rich, it's called Netflix, not DVDFlix." He kept watching the business world from a distance and kept finding it interesting. "I knew that we weren't going to stay in Italy," he said. That was, in his phrase, "great to learn."
The sabbatical became a creed. "I'm a big believer in the sabbatical," he says, and years later at Zillow, which didn't track vacation days, he praised Spencer Rascoff's emails announcing he'd be "literally off the grid" for forty-eight hours as a signal that it was "healthier for the company when you do wring the sponge out." But the Florence year deserves a less sentimental reading. A man who loved Italy for its indifference to work went there and found that he couldn't stop working, at least not in his head. He came home in 2004 with a road bike and an open question, and within months he had answered it with the most work-intensive thing he could have chosen.
Prices on Every Roof
The question got answered because somebody else got fired.
Lloyd Frink had been two years ahead of Barton at Stanford and was with him at Microsoft and Expedia. Frink stayed at Expedia after Barton left, until, as Barton tells it with obvious delight, he was fired by "our good friend Eric, who was running Expedia at the time, probably for good reason." (He declined to elaborate. "It's a fun story.") Barton was drifting toward venture capital and possibly California, and had been getting to know Bruce Dunlevie and Bill Gurley at Benchmark. Frink intercepted him. Wait, wait, wait, don't move to California. Let's just sit in an office and brainstorm for a while.
Frink's father was a stockbroker with a couple of spare rooms in his Seattle office, and he gave them to the two of them. Frink went off for two weeks to study something that amounted to Dropbox before Dropbox, came back and reported that it would definitely work and would make no money, because Microsoft and Google would give it away. They moved on.
Then Barton dug out an old file. During the Expedia years, a small team inside Microsoft had written plans for how the web would change one industry after another: an electronic stockbroker, a matchmaking service, all the classifieds categories and all the agent categories. Expedia had been one of those plans. Another was a digital real estate marketplace. Both men happened to be house hunting with growing families. Barton couldn't believe what they were up against. "This is 2003 and I can't get the price of a home online," he said. "Are you kidding? I can't even get the address." To the Seattle Times he described the experience as having "to hire a professional and get a little drip-drip of information."
There it was again: the professional, the keyboard, the screen pointed the wrong way.
Barton agreed to be CEO on one condition, that he wouldn't work very hard. Frink said fine, he'd do the work, he just needed Barton's name for raising money. Barton added that he'd also be starting other companies and doing venture capital. Frink said fine to that too. "He totally managed me," Barton said, laughing. "I think I'm in charge now. Well, he was in charge." Zillow was incorporated in Washington in December 2004. Barton and Frink and a couple of friends put in the first $5 million themselves, a move he recommends to any second-time founder who can afford it, because "you end up with more of the company down the road."
Barton described what they started with in a single sentence: "We knew 100 percent that there would be a leading digital real estate marketplace in the US." The business model was a different matter: "Who knows, who cares? It's a giant. It's a big pond."
They spent the next year in secret, which Barton turned into a joke when Benchmark announced him as a venture partner. "We're trying to recruit people," he told the Seattle Times, "but we don't want to tell them what we're doing."
What they were doing, at first, was wrong. Like everyone in 2004 they were dazzled by Google, specifically by its auction-based ad marketplace, and the founders were what Barton calls "mathy idealists," for whom auctions have tremendous geek appeal. The auction is the most efficient mechanism for price discovery. So obviously, they reasoned, American houses should be auctioned. A friend named Gordon gave them a house on something like consignment, and they tried.
They learned two things. An auction needs all the bidders present at once, a real-time liquid market, and housing works over weeks, showing to buyers one at a time. And people resist being taught a radically new way to do something they've done the same way for generations. A new way, Barton concluded, has to be "super-duper simple, obvious, and 10 times better than the current way. And just, we didn't check any of those boxes."
But they had been chasing price discovery, and in the wreckage of the auction they found the price itself. The picture that formed in their heads was the aerial map with a number on every roof, a "real time estimated value, algorithmically driven, as you're Google Map zooming over neighborhoods." Barton added a second image. A house was most American families' largest asset. Then plot it like one: give every house a value history that looks like a stock chart. "I knew that was catnip," he said. "The team knew it was candy."
They called the number a Zestimate. When Zillow went live on February 8, 2006, it offered free valuations on some 60 million homes. Barton told the Los Angeles Times that the site's national margin of error was 7.2 percent and that Zillow could become real estate's Kelley Blue Book. "We think that a key component to a fundamentally healthy marketplace is price transparency," he said. "People want to be empowered with information so they can take control of what is scary and difficult and fraught with frustrations." He predicted that commissions could come down, as they had in travel. An industry skeptic in the same story said automated valuations could be off by as much as 30 percent, because "every property is an illiquid, unique asset."
Walt Mossberg of the Wall Street Journal, then the most influential consumer-technology critic in the country, loved the site, and millions of people showed up on day one. The servers were literally in a closet; Amazon Web Services didn't exist yet. They fell over. Company lore holds that CTO David Beitel, who had been CTO at Expedia too, banned Christmas lights to save power for the machines. Barton says he doesn't remember that. He allows that it might be true. The engineers wrote a launch blog post riffing on Doc Brown's gigawatts from Back to the Future. Amy Bohutinsky, the head of marketing, told everyone not to worry: they'd make lemonade out of lemons. The San Francisco Chronicle ran the headline: "Zillow: Pornography for the real estate-obsessed." Bohutinsky was delighted. Within three days the site had a million users.
The house-porn line has followed Zillow for two decades, and Barton has never fought it. Years later, Saturday Night Live ran a fake ad with Dan Levy in paisley pajamas, sighing that the pleasure he used to get from sex now came from looking at other people's houses. Barton tweeted: "Wait. Have we been marketing @zillow wrong all these years???" His own version is more decorous. "People's capacity for spending time thinking about and dreaming about homes is not infinite," he told Stanford magazine, "but effectively it is."
Bill Gurley, by then on the board after Benchmark and TCV led the first round, had already asked the question that shaped what came next: What if we didn't have any marketing budget? Expedia had bought its brand on television. Barton's first reaction was no way. Then the constraint started generating ideas. Bohutinsky noticed that local newspapers had what Barton calls "an infinite newshole for housing data," always room for a story about prices on Elm Street, and she built a pipeline to feed it a constant stream of Zillow numbers. Zillow grew without paying for its audience because the product itself was the story. People looked up their own houses, then their neighbors', then their high school boyfriend's.
A month after the Chronicle headline, real estate agents were complaining that the numbers were wrong. In October 2006 the National Community Reinvestment Coalition asked the Federal Trade Commission to look at whether Zillow was misleading consumers about the accuracy of its valuations; the FTC declined to investigate, and the complaint was later withdrawn. In 2007 the Wall Street Journal checked and found the estimates "often are very good, frequently within a few percentage points of the actual price paid. But when Zillow is bad, it can be terrible."
That sentence could serve as Zillow's epitaph or its creed. The company kept refining the algorithm. Barton described it to Wired in 2007, from his 46th-floor office, as "a complex piece of AI" that picks the best-performing models neighborhood by neighborhood, and noted that Zillow was careful with major updates because "we are literally rewriting history." On June 14, 2011, it did rewrite history: an algorithm change revised current Zestimates on millions of homes along with historical values going back to 2006. The stock chart of the American house could be redrawn retroactively. Nobody stopped looking at it.
I'll Give You the FBI
Barton started seeing the same pattern everywhere.
In 2007, while still running Zillow, he was brainstorming with Robert Hohman, an Expedia colleague, and they got onto a near-disaster from the old days. Barton had once sent the raw results of an Expedia employee-satisfaction survey to an office printer and left them there. His assistant picked them up before anyone else saw them. Hohman and Barton asked what would have happened if the pages had gotten out, and decided the answer was: something good. Employees would know what their colleagues thought, and job seekers would know what they were getting into. The information that HR departments guarded so carefully was, in their view, information the public needed.
They named the company Glassdoor. Its provocative datum was salary, which Barton calls "a little bit taboo for a lot of people, so it was inherently secret and provocative." Its hardest problem was getting people to share. After hand-cranking surveys, Hohman's team hit on what Barton calls give to get: you see a little for free, and to see more you contribute your own salary, title and company, anonymously, plus a review of your employer. "You show me yours, I'll show you mine." Users could also rate their CEO. "We knew some CEOs would go crazy," Barton said. By 2014 Glassdoor had more than 22 million members.
In 2008 Fortune caught Barton at his most sprawling. He was CEO of a 150-person, not-yet-profitable Zillow that had raised $87 million and was selling ads to Deere targeted only at homes with enough acreage to justify a riding mower. He sat on the boards of Netflix and Common Sense Media. He was a venture partner at Benchmark. With some Seattle friends he ran a $10 million quant hedge fund. And he was a founder of two more startups, Avvo and the still-unlaunched Glassdoor. The writer called him "disgustingly relaxed."
Avvo was his most openly combative company. Its founder and CEO, Mark Britton, had been Expedia's general counsel and before that had worked at the Securities and Exchange Commission: a real lawyer, in Barton's phrase, which mattered because Avvo's plan was to rate lawyers systematically, something nobody had done. On the fundraising trip down Sand Hill Road, investors kept asking whether it was legal. You know you're going to get sued. Barton answered with Die Hard. Hans Gruber needs the power cut to open the last lock on the Nakatomi vault, and he knows the FBI's playbook calls for cutting the power, so he waits. Ladies and gentlemen, I'll give you the FBI.
The lawsuits were the launch plan. Avvo knew its legal ground was solid, figured the suits would be cheap, and could do much of the legal work itself. It won the first few, after which, Barton said, "the lawyers lost steam on suing." Every suit was also free publicity.
— Nick Hanauer, Seattle venture capitalist and early Amazon investor, to the New York Times, 2014You can name people who are richer than Rich, but you can't name very many people who have his track record. You will find very few people in this country who have as many times created something from nothing.
Barton calls the method provocation marketing, and when Tim Ferriss pressed him to lay it out as a seminar, the syllabus came out in about three sentences. Start in a "seven deadly sin zone," a subject people feel strongly about: money, envy, status, home. Go after a taboo or sacred cow in that zone. Then throw out every idea that works by frightening people. "I do believe a cheap way to get attention is to scare people, but I think it's cheap," he said. "Effective but cheap." You want people provoked, "but feel good, or tickled, or entertained."
The resistance was part of the method too. In many places, real estate associations lobbied to have Zillow outlawed, or argued that an unlicensed company couldn't produce something resembling an appraisal. Barton was untroubled, partly because his lawyers were confident and partly because he'd noticed something about the legislators: they used Zillow. The job was simply to "activate that latent love for the product." Elected officials don't like taking popular things away from voters. "Power to the people, baby," he said. He has explained to Barry Ritholtz that the phrase isn't a political slogan. It's "a technological one."
He also admits it can go too far, and he says so with a candor that will matter later: "Of course you can go too far, and in fact it may be that you need to go too far to establish where the frontier is."
Not everything worked. A social-media game startup he co-founded, King of the Web, went nowhere. A travel-photo app called Trover was bought in 2016 by Expedia, the company he'd started inside Microsoft twenty-two years earlier, which meant that a Barton startup was effectively sold back to the original Barton startup.
And through all of it he kept naming things. In a private email in 2006, later refreshed as one of perhaps ten posts on a blog named after a fly-fishing rig, Hopper and Dropper, he set out his rules. Literal names like Travel.com are easy to explain and impossible to own. Words given a new meaning, like Amazon or Apple, can work. The best path is to make up a word, because then "you own it." The cost is patience: Expedia took eight years to reach 50 percent unaided awareness among online travelers, and years after launch Zillow's awareness was still in the low single digits. The method uses Scrabble. Z, Q and X are worth ten points because they're rare, and rare letters jump off the page. Two syllables are ideal; Expedia, he concedes, was a syllable too long. The name should be easy to turn into a verb. Double letters help. So does one more test: "If the word could be a good dog name, you're onto something, like you can call for it. Zillow."
He gives Glassdoor a B. Not much of a dog's name.
The Coach Who Took Back the Field
Barton doesn't like running his companies for long. The Seattle Times once observed that he "rarely serves as CEO of the companies he creates" and instead "has a knack for finding leaders to pass them on to." He prefers founding, recruiting, sitting on the board and coaching. He describes a board seat as "you're instinctively a coach," and he likes his own executives to sit on outside boards ("if you love it, set it free").
Spencer Rascoff came to the Barton orbit when Expedia acquired Hotwire, the company Rascoff had co-founded. He had started in investment banking, where, he once recalled, you left your sport jacket on your chair at midnight and went down the fire stairs to a different floor so nobody would see you leave. He became Zillow's VP of marketing, then CFO, then COO. Zillow turned its first profit in 2010, and in September of that year Barton handed Rascoff the CEO job and became executive chairman.
Rascoff wanted to take the company public, and he did. Zillow filed on April 18, 2011, priced at $20 a share, valued at nearly $540 million, and rose 79 percent on its first day on Nasdaq under the ticker Z. Then it started buying: Postlets in 2011, RentJuice for $40 million and HotPads for $16 million in 2012, StreetEasy for $50 million in 2013, and in February 2015 its main rival, Trulia, for $2.5 billion, after which the combined company handled about 70 percent of online real estate searches and cut 350 jobs. Zillow became Zillow Group. By its tenth anniversary it filled twelve floors of the Russell Investments Center in Seattle, with nearly 1,000 employees there and 2,200 nationwide.
The money came overwhelmingly from agents, through Premier Agent, Zillow's program for buying exposure to the buyers who'd been browsing the site. That's the open secret of the "power to the people" years. The industry Zillow embarrassed became the industry that paid for it. An early agent described Zillow as "a place where you checked out your friend's house." A decade later it was where the agent's leads came from. The light Barton turned on had become a building the agents paid rent in.
He watched from the chairman's seat as a new generation of companies, led by Opendoor, started doing what Zillow only displayed. They bought houses directly from sellers with cash, fixed them up and resold them. The industry called it iBuying. In April 2018 Zillow announced it would do the same through a new business, Zillow Offers.
In February 2019 Rascoff left, and Barton took the CEO job back. Forbes summed up the moment in a headline: Barton had a $700 million stake in Zillow "and plans to turn it into a home-flipping machine." The tech analyst Kevin Kwok compared it to Michael Jordan returning from his first retirement, and noted that hardly anyone in Silicon Valley paid attention. Barton told The Real Deal the goal was to make homebuying a "one-click trade-in nirvana." Analysts and the company called the new direction Zillow 2.0: move from search to transaction, from showing you the house to selling it to you. In the Ferriss interview Barton explained the defensive logic, which comes out of Michael Porter's work. Any middleman that depends on Google for its traffic is exposed, because eventually Google comes looking for the same business. The defense is to build a brand people come to directly, and then to go "down into the workflow" of the industry, all the way to the transaction.
Owning the house was the furthest down you could go.
The pandemic first looked like a test, then like a vindication. In March 2020 Barton cut expenses 25 percent, froze hiring and paused home-buying. In July Zillow became one of the first companies to let about 90 percent of its employees work from home indefinitely, and Barton started posting about "The Great Reshuffling": remote workers looking for cheaper houses, many of them on his app. Zillow was accessed nearly 10 billion times that year. The stock more than quadrupled in a little over a year, and in February 2020 Forbes had counted Barton among the billionaires. In March 2021 the company said it had hired about 1,500 people over the previous twelve months and planned to add 2,000 more by year's end, a 40 percent increase in headcount. It had also spent at least $600 million on acquisitions to fill out the product.
One of the people in the building was Jeremy Wacksman, who had joined from Microsoft in 2009, run marketing, and become COO. The other is Barton's son Russell, the twin born after the Whistler scare, about whom Barton once wrote that his elementary school's principal, giving a tour of a new building, pointed at the high fence on a third-floor bridge and called it an example of how they had "Russell-proofed" the school. Barton's point in that essay was that he didn't want to teach his son fear that wasn't already there. I wanted him to be comfortable taking big swings.
Leveraged Housing Traders
The swing failed, in the end, on the one thing he had spent his career assuming: that the number on the roof was knowable.
Zillow Offers had been designed as a market maker, earning a thin spread on volume, paying sellers a fair price and selling quickly. That works only if you know what the house will sell for months from now, after repairs and holding time and whatever the market does in between. In 2020 and 2021 the market did things no model had seen: a pandemic, a brief freeze, and then prices rising faster than ever before. Zillow's forecasts missed, as Barton put it, "at different times in both directions by much more than we modeled as possible." Contractors were scarce and supply chains were jammed, so renovations slowed, holding periods lengthened, and every extra week of ownership increased the exposure. In October Zillow announced it had paused new purchases because of a processing backlog. Two weeks later it shut the whole business down.
— Rich Barton, letter to shareholders and employees, November 2, 2021Put simply, our observed error rate has been far more volatile than we ever expected possible and makes us look far more like leveraged housing traders than the market makers we set out to be.
The most important paragraph in the letter is the one where he declines the obvious excuse. "We could blame this outsized volatility on exogenous, black swan events, tweak our models based on what we have learned, and press on," he wrote. "But based on our experience to date, it would be naive to assume unpredictable price forecasting and disruption events will not happen in the future." Getting to scale would require "too much equity capital," would make earnings and the balance sheet too volatile, and would produce "a far lower return on equity than we imagined."
The decision cost about 25 percent of the workforce, people hired during the boom, many of them in the operations he was shutting. Ben Thompson of Stratechery wrote the fairest judgment. Zillow Offers had been Barton's responsibility, Thompson noted, "so he deserves blame, but at the same time, that makes the fact he looked at the business logically instead of emotionally that much more impressive." The best time to avoid a strategic mistake is at the beginning, he went on, and "the next best time — and the far more difficult time, given sunk costs — is the moment you realize the decision was wrong."
Bloomberg's Patrick Clark made the comparison to Jim Collins's "big hairy audacious goal," the idea from Built to Last that companies make progress by setting an outrageous objective and chasing it relentlessly. Zillow had found a goal that turned out to be big, hairy, audacious and wrong. The question the episode raises isn't whether Barton swings too hard. He has said, approvingly, that you sometimes have to go too far to find the frontier. The question is what a man who built three companies on displaying prices learned when he tried to bet on them.
Part of the answer is in the gap between two kinds of numbers. A Zestimate is a published estimate. If it's off by seven percent, a homeowner is annoyed and an agent has something to argue about over coffee. Zillow can correct it, or rewrite its history, and nobody's balance sheet suffers. A purchase price isn't an estimate. It's a commitment. As a publisher, Zillow had benefited from its error; the gap between the Zestimate and the sale price kept people talking and coming back. As a buyer, the same error was a liability. The Journal's 2007 verdict ("when Zillow is bad, it can be terrible") was harmless when Zillow only described houses and costly once Zillow owned them.
Nine months later, in August 2022, Zillow announced a partnership with Opendoor, the company it had chased into iBuying. Zillow users who wanted a cash offer would be sent to its rival.
The Folder Called The Breakfast Club
What Zillow became after 2021 its executives called a "housing super app": search, touring, financing, the agents' tools, the rental application and payment, one company handling every step from first browse to moving day, all part of what the company described as "five pillars of growth." In August 2024, after its sixth straight quarterly loss, Zillow named Wacksman CEO, and Barton joined Frink as co-executive chairman. "I'm not going anywhere," he said. "I'm not doing anything else, but I'm really excited to be moving to a role of coach and adviser rather than the field general." In 2024 the company had revenue of $2.24 billion, a net loss of $112 million, 6,856 employees, and a database of about 160 million American homes.
The company that turned the screen around has spent the years since learning what it's like to sit behind one.
In September 2024 Zillow partnered with First Street, a climate-risk modeling firm, to show a Climate Risk Score on listings, rating flood, wildfire and extreme-heat exposure: a provocative number about a subject people care about intensely. By November 2025 it had removed the scores, after real estate agents struggled to sell high-risk homes and owners watched their values drop. In 2007 Barton had told Wired that "once you open the information doors, it's difficult to close them." In 2025 Zillow closed one.
In June 2025 the brokerage Compass sued Zillow under federal antitrust law. Zillow had a rule that any home marketed elsewhere and not listed on Zillow within twenty-four hours would be barred from the site permanently, and Compass argued the rule targeted its private-exclusive listings. The once-unlicensed upstart that lobbyists had tried to outlaw was now the gatekeeper being accused of keeping the gate. That September a proposed class action in Seattle alleged that Zillow steered buyers to agents it was affiliated with, from whom it takes up to 40 percent of commissions, rather than to the listing agent. Five states sued over a $100 million agreement under which Redfin stopped competing for multifamily rental listings and moved its customers to Zillow. A company built on the image of a professional hiding a screen from a customer was now accused of arranging which professional the customer met.
None of this has been decided. All of it rhymes.
Barton watches some of it from a kitchen table. He's up at 6:30, hours before Sarah, with one or two cups of coffee and a smoothie of oat milk, two-thirds of an apple, pistachios, macadamias, blueberries, a can of the Japanese electrolyte drink Pocari Sweat, and a prune. He wears a heated, vibrating back wrap. CNBC plays on mute on an iPad next to his inbox, captions scrolling, because "business news is generally happy" and "regular news makes me feel bad." He doesn't take longevity supplements. "My observation is the harder you push against something," he says, "the harder it pushes back."
The kids are grown now. For years, though, he was the parent who got them to school, and since he'd worked as a short-order cook as a young man, he made breakfast to order for whoever came downstairs half-asleep. When his younger son started an ADHD medication that killed his appetite for the rest of the day, Barton made him twelve-egg frittatas with potatoes and sausage and watched him finish them. Most mornings he quietly took a picture. He never posed them.
The folder on his phone is called The Breakfast Club. It holds about a thousand photographs, in order, "a time series of these kids growing up," he says, which is the same structure as a house's price history on Zillow, a value plotted over time. "It's really a virtual possession," he says, "but it's my most prized possession."
None of the pictures has a number on it.
Part IIThe Playbook
Barton's career looks like one idea repeated, and in a sense it is: find a market where a professional hides a number from the customer, and show the customer the number. But the idea alone explains very little, because many people had it and few built anything durable. The principles below come from the specifics: the dinners at Wild Ginger, the house on consignment, the empty marketing budget, the shutdown letter. They include where the method broke, which teaches as much as where it worked.
The Barton Playbook
- 1.Listen for the keyboard on the other end of the line.
- 2.Kill your own assignment, then replace it with something bigger.
- 3.Pitch the institution's fear, not your dream.
- 4.Pay for control early.
- 5.Let the failed product hand you the killer feature.
- 6.Run as if you had no marketing budget.
- 7.Pick fights you have already won on paper.
- 8.Name it like a dog you'd call across a park.
- 9.Own the demand, then go down into the workflow.
- 10.Protect the people who rock the boat.
- 11.Know whether you are publishing a number or betting on it.
- 12.Hand off the field; keep the whistle.
Principle 1
Listen for the keyboard on the other end of the line
Each Barton company started from the same sensory clue: a professional on the phone, typing into a system the customer knows exists and isn't allowed to see. Travel agents had Sabre. Real estate agents had the MLS, split across what Barton described to Ritholtz as some 900 regional offices. HR departments had salary bands and survey results, one of which once sat unattended on an Expedia printer. Barton's question, as the New York Times summarized it, was always "What piece of marketplace information do people crave and do not have?"
The useful part is how specific the test is. He isn't looking for inefficiency in general. He's looking for an intermediary whose value depends mostly on access to information, in a category emotional enough that people will spend hours with the information once they have it. Houses qualified twice over: a family's biggest asset, and, as he told Stanford magazine, a subject people's capacity to dream about is "effectively" infinite.
The weakness of the test is that it tells you where demand will come from, not how you'll get paid. Barton said of Zillow's model: "Who knows, who cares?" That was fine for a company that grew rich on agent advertising. It was less fine when Zillow later tried to become the counterparty itself.
Tactic: List every moment this year when you were on the phone with a professional while they looked at a screen you couldn't see, and ask which of those screens millions of people would happily spend their evenings staring at.
Principle 2
Kill your own assignment, then replace it with something bigger
At twenty-six, Barton walked into a review with Bill Gates and called his own project "dumb, small, and uninteresting." The criticism alone would have been career suicide. What saved it was the replacement he had ready, easySABRE on Prodigy, which he could demo on the spot. He had sized the market (about $100 million for U.S. travel books in total), found the fatal product flaw (no one carries a twenty-pound laptop on vacation), and shown a bigger pond.
The heuristic he took from it, "big pond, good fishermen," leaves out a nuance he's careful about elsewhere. He warns that many founders "identify a real problem, but it's just too small," and he uses Uber to make the opposite point: what looked like a small black-car market was really all of transportation. So the skill isn't just picking big markets. It's telling apart a problem that's genuinely small from one that only looks small because nobody has defined the market correctly yet.
Tactic: Before your next project review, write one paragraph arguing that your current mandate is too small, and walk in with a working demo of what should replace it.
Principle 3
Pitch the institution's fear, not your dream
Barton pitched the same company three ways. To Ballmer in the early days he pitched a fantasy, with a homemade travel-agent card bearing Ballmer's face. To Ballmer in 1999 he pitched the spin-out as an HR experiment, because Microsoft's real worry wasn't Expedia's valuation. It was losing its most entrepreneurial people to the startups outside and leaving the lazy ones inside, collecting option gains. To the spouses of his 150 recruits, across a dinner table with a marked-up S-1, he pitched risk management.
The lesson isn't to tailor your message, which every sales training says. It's that institutions approve unusual things when the unusual thing solves a problem they already have. Microsoft didn't divest anything, as Barton told Wired, because "nothing isn't strategic." The only way to get something out was to make the spin-out serve something Microsoft considered strategic, which was its talent.
He's also honest that the stated rationale failed: Microsoft's stock took seventeen years to recover, and Expedia's people were the ones who made money. Arguments made in order to persuade don't have to be correct predictions. They have to be true enough, and in the listener's interest.
Tactic: Before your next big internal ask, write down the decision-maker's biggest worry this quarter that has nothing to do with your project, and rebuild the pitch around solving it.
Principle 4
Pay for control early
"I didn't control the company," Barton said of the Diller sale. "I learned that henceforth I would control the companies that I started." Expedia was sold in two steps by a parent company that owned about two-thirds of it, didn't sit on its board, and wanted a simpler story for Wall Street. The founder found out his company was for sale the same way an employee would.
At Zillow he did it differently. He and Frink put in the first $5 million with a couple of friends before raising venture money, and he recommends this to any second-time founder who can afford it: "you end up with more of the company down the road." When the time came to take the CEO job back in 2019, he had a stake Forbes valued at about $700 million and the standing to make it happen.
The cost of control is that nobody can stop you. The same structure that let him return to Zillow let him bet the company on iBuying. A founder who pays for control is also paying for the right to be wrong without much resistance.
Tactic: If you have the means, fund the exploratory phase yourself and treat the equity you save as the price of being in the room when your company's fate is decided.
Principle 5
Let the failed product hand you the killer feature
Zillow began as a housing auction. It failed in a few weeks, on one house borrowed from a friend, because housing isn't a real-time liquid market and because customers resist being retaught how to buy something they've bought the same way for generations. Barton's rule from that experience is strict: a new behavior has to be "super-duper simple, obvious, and 10 times better than the current way."
The auction failed and its purpose survived. They had been after price discovery, and the Zestimate is price discovery with the transaction removed: what an auction might say, available instantly for every house, with nothing to sell. The same pattern, an abandoned business model leaving behind its most appealing component, shows up in Barton's own Microsoft history: the DOS-for-Dummies bundle failed, and the next big idea is what Brad Chase asked about.
Z
From auction to Zestimate
How the abandoned model produced the feature that defined the company.
| What the auction required | What the Zestimate kept | Verdict |
|---|---|---|
| All bidders present at once | An instant number, no bidders needed | Kept the goal |
| Customers learning a new way to buy | Customers looking at a map they already understood | Removed the friction |
| A transaction to monetize | Attention to monetize (advertising) | Deferred the model |
| Price set by the market | Price estimated by an algorithm | Error became visible |
Tactic: When an experiment fails, write down the single outcome it was trying to produce and ask whether that outcome can be delivered by itself, without the business model you built around it.
Principle 6
Run as if you had no marketing budget
At Expedia, Barton wanted $100 million for television ads. At Zillow, Bill Gurley asked what would happen with no marketing budget at all, and the question changed the company. Barton's answer, built mostly by Amy Bohutinsky, was to make the product the marketing: a feature so provocative people would talk about it on the sideline of a soccer game, and a constant stream of housing data sent to local newspapers with what he called "an infinite newshole."
His seminar fits on an index card. Pick a "seven deadly sin zone." Find the taboo inside it. Throw out every idea that relies on fear, because fear is "effective but cheap." Keep the ideas that leave people "provoked, but feel good, or tickled, or entertained." Glassdoor's salaries, Zillow's neighbor's house and Avvo's lawyer ratings all fit. So did the Chronicle's house-porn headline, which Bohutinsky correctly read as good press.
P
Provocation, company by company
Each Barton company had one number that made people talk.
| Company | The provocative datum | Who was provoked |
|---|---|---|
| Expedia | Fares and availability, visible without an agent | Travel agents; airline executives |
| Zillow | An estimated value on every house, sold or not | Real estate agents and their associations |
| Glassdoor | Salaries by title and company; CEO ratings | HR departments; CEOs |
| Avvo | Systematic ratings of attorneys | Attorneys, some of whom sued |
Tactic: Take your marketing budget to zero on paper, then list the features you'd need to build for customers to tell other people about you unprompted.
Principle 7
Pick fights you have already won on paper
Barton's appetite for conflict is real, but it's calculated. Avvo's founder had been an SEC lawyer and Expedia's general counsel; the company knew its legal position and planned for the lawsuits as publicity ("I'll give you the FBI"). At Zillow, when real estate associations lobbied to have the site banned or argued it was practicing appraisal without a license, Barton didn't fight on their terms. He relied on the fact that legislators were users and voters were users, and "activated that latent love for the product."
What makes this replicable is its order. Get the legal footing settled first. Then make sure enough of the public likes the product that the officials who decide your fate are among them. A provoked incumbent then ends up arguing against its own regulators' constituents.
Barton adds a limit: you can go too far, and you may have to go too far to find out where the line is. Zillow found one of its lines in 2025, when it removed climate-risk scores after agents and homeowners objected. Popular support protects you only while the people providing it believe the number is helping them.
Tactic: Before launching anything that will anger an incumbent, get a written legal opinion and a count of how many of the relevant decision-makers already use your product.
Principle 8
Name it like a dog you'd call across a park
Barton's naming rules sound like a joke until you notice the companies they produced. Behind them is a real strategic claim: a literal name like Travel.com comes with meaning you can never own, and a made-up word becomes your property once people learn it. The price is time. Eight years for Expedia to reach 50 percent unaided awareness. Barton argues the price is worth it only if you're building "a brand that your grandchildren will know and love."
Q
The Barton Naming Rules
As described on Hopper and Dropper and to Tim Ferriss.
| Rule | Rationale | Example |
|---|---|---|
| Use high-point Scrabble letters (Z, X, Q) | Rare letters are distinctive and stick in memory | Zillow, Expedia |
| Prefer two syllables | Fewer syllables are easier to say and share | Expedia: one too many |
| Evoke something positive | Connotation does free marketing | Expedia: speed, expedition |
| Pass the dog-name test | You should be able to call it across a park | Zillow |
| Make it easy to turn into a verb | Verbs work their way into everyday speech | Usually ends in a vowel sound |
| Use double letters or palindromes | Unusual word shapes jump off the page | Glassdoor: a "B" |
The deeper connection is with provocation marketing. A made-up word is only affordable if you can bring in a large audience early without buying it. Barton chooses the harder kind of name because he's confident he can get attention for free.
Tactic: If you can't explain how you'll get your first million users without paid advertising, use a literal name; if you can, invent a word and score it against the six rules.
Principle 9
Own the demand, then go down into the workflow
Barton's strategic analysis of his own industry is plain Porter. A company that gets its customers from Google depends on Google, and eventually Google moves into its business, as it did with reviews, maps, flights and hotels. Many of the "TripAdvisor for X" companies of the 2000s, Avvo among them, were squeezed this way. Barton has two defenses. Build a brand strong enough that customers come to you directly. Then "look down your funnel," into the industry's workflow, and become a necessary part of the transaction.
That explains Zillow's acquisitions after 2011: the rental tools, the agents' software, ShowingTime, Follow Up Boss, the mortgage businesses, and finally the "housing super app." It also explains Zillow Offers, which was the same logic taken as far as it could go. The strategy is sound. How far down you go is what determines the risk. Software and services in the workflow are cheap, while owning the asset puts the whole balance sheet behind your forecast.
T
The descent into the workflow
Zillow's move from information to transaction.
2006
Zillow launches with Zestimates on ~60 million homes; revenue from advertising.
2008
Mortgage marketplace and iPhone app launch.
2011
IPO; first acquisition (Postlets), a listing tool for agents.
2015
Trulia acquired for $2.5 billion; Zillow Group formed.
2018
Zillow Offers begins buying and selling homes; rental applications and payments added.
2019
Barton returns as CEO to scale "Zillow 2.0."
2021
Zillow Offers shut down with ~7,000 homes in inventory.
2022
"Housing super app" strategy announced; partnership with Opendoor.
2024
Barton becomes co-executive chairman; Wacksman named CEO.
Tactic: Map the steps your customer takes after leaving your product, and integrate into the next step that requires software and trust, stopping before the step that requires you to own the asset.
Principle 10
Protect the people who rock the boat
After the DOS-for-Dummies bundle lost Microsoft as much as $20 million, Brad Chase asked Barton for his next idea, and Barton has never forgotten it. He kept one of the boxes. His view of culture follows from that moment: "Mainline corporate culture sometimes rejects, often rejects the innovators," so leaders have to "hire, cultivate, protect, and invest in those people." He adds a limit: if the failures keep coming, "maybe there's a pattern and somebody should go find another job."
Look at his co-founders and the pattern shows: Frink, Hohman, Britton, Rascoff, Bohutinsky, Beitel, nearly all people from earlier Barton companies who went with him into the next one. He runs his career as a long-running internal startup program. He also encourages his executives to sit on outside boards, because a company too afraid of losing people "is too insular."
Tactic: The next time someone on your team loses money on a well-reasoned bet, open the review by asking about their next idea before you discuss what went wrong.
Principle 11
Know whether you are publishing a number or betting on it
This is the principle Barton learned at the highest price. For fifteen years, the Zestimate's errors cost Zillow very little and, in a sense, helped it, because disputes over the number kept people coming back. The Journal's 2007 verdict ("when Zillow is bad, it can be terrible") was something to manage. Once Zillow started buying houses, the error rate became financial exposure, and the business could survive only within 200 basis points of breakeven.
The way Barton shut it down is the part to copy. He named the specific flaw (forecast volatility), refused the black-swan excuse, and assessed what the business would look like at scale rather than what it had cost so far. Ben Thompson's point applies: the hardest time to fix a strategic mistake, and the next best one, is the moment you realize it, with all the sunk cost still in view.
!
Market maker vs. market risk-taker
The distinction at the center of the Zillow Offers letter.
| Dimension | Publishing a number | Betting on a number |
|---|---|---|
| Cost of a 7% error | A complaint, a correction | Losses on the balance sheet |
| Effect of volatility | More traffic, more stories | Wide swings in unit economics |
| Capital required | Servers and data | Equity and debt that grow with inventory |
| Response to error | Revise the model, even retroactively | Sell the houses, cut 25% of staff |
Tactic: Before turning an information product into a principal business, calculate what your historical forecast error would have cost if you had paid for every prediction with your own capital.
Principle 12
Hand off the field; keep the whistle
Barton rarely stays CEO for long, and he says so openly. He handed Zillow to Rascoff in 2010 and to Wacksman in 2024, Glassdoor to Hohman, and Avvo to Britton. He describes the role he prefers as coach and adviser rather than "the field general." Between assignments he takes sabbaticals: the Florence year, and the morning routine he guards now. Maffei's test for board seats ("local, fun, lucrative," two out of three) is a way of saying that time spent coaching should be chosen as carefully as time spent operating.
The coach's role has its own discipline. Barton returned in 2019 because, as chairman, he could see a strategic shift that he believed needed a founder's authority. Taking the job back cost a successful CEO his position and led to Barton's biggest public failure. A founder's ability to take back the field is valuable only if the founder also knows when to give it back, which Barton did in 2024 after the company had found its next strategy.
Tactic: Write down the specific conditions under which you would return to a role you've handed off, and the conditions under which you would leave it again, before you hand it off.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Rich Barton, Stanford talk, 2017 (as reported by The Real Deal)Putting barriers between people and information is unsustainable in the world of technology.
— Rich Barton, to Wired, 2007Once you open the information doors, it's difficult to close them. I don't think it's characteristic of just the current market. I'm of the opinion that there is no hiding, period. Everybody is a reporter, a blogger, a rater of everything. Fighting that force is like fighting gravity.
— Rich Barton, to the New York Times, 2014If we're doing things for regular folks that make their lives better and save them money and give them transparency, we're on the side of the angels.
— Rich Barton, The Tim Ferriss Show, 2025Great organizations encourage innovation, encourage big idea people to take big swings and do not punish them when it doesn't work out according to plan.
— Rich Barton, The Tim Ferriss Show, 2025Of course you can go too far, and in fact it may be that you need to go too far to establish where the frontier is.
Maxims
- Listen for the keyboard. The best opportunities are wherever a professional is looking at a screen the customer can't see.
- Big pond, good fishermen. A real problem in a small market is still a small business.
- Make the product the marketing. Build the feature people will bring up on the sideline of a soccer game, and you won't need to buy their attention.
- Provoke, don't frighten. Fear gets attention cheaply and leaves a bad feeling; delight and transgression build a brand.
- Settle the legal question first, then pick the fight. Lawsuits you're confident of winning are publicity.
- Invent the word. A made-up name costs years of patience and becomes something you own outright.
- Control what you start. Self-funding the early search is the price of being in the room when your company is sold.
- Ask for the next big idea. How you respond to a smart failure determines whether you'll get the next smart idea.
- Separate publishing a number from betting on it. An error that brings in readers can bankrupt you once it's your money.
- Wring the sponge out. The sabbatical, the bike ride and the breakfast at home are part of how a long career keeps working.
