The Web's local mogul

The company grew at bubble-era speed. Horowitz began making acquisitions, cobbling together a network of websites that included everything from the tech-stock tout Silicon Investor to the search engines Metacrawler and DogPile. His aggressive approach was richly rewarded: The company's market value peaked at more than $4 billion, and Go2Net became a top-20 Web property in terms of traffic. By mid-1999, Paul Allen's Vulcan Ventures had plowed $300 million into the business. In addition, Horowitz sold Allen a chunk of his shares, netting himself roughly $100 million.

When Go2Net sold out to Infospace in October 2000, it was cash-rich and profitable - rare for any dotcom at the time. The merger was a disaster, however, and infighting grew intense as the tech market imploded. After just 90 days, Horowitz, who had become Infospace's president, bolted. He spent the next year and a half entertaining job offers, concluding that the same trends that drove Go2Net's growth - the inevitable expansion of the Web reaching ever narrower markets - were firmly intact, despite the fact that investors had run for cover.

During the worst of the postbubble malaise, in January 2003, Horowitz and four partners launched Marchex with $20 million of their own money, eventually setting up in the same building in downtown Seattle that housed the original Nordstrom where Horowitz had sold women's shoes as a teenager. The plan was to try to build a network of sites and, most important, amass traffic without having to rely on the likes of Google and Yahoo.

Within months the team made a couple of key acquisitions. It snapped up a pay-per-click company called Enhance Interactive, which gave it partnerships with dozens of traffic sources and a base of advertisers. And it bought TrafficLeader, which ran paid-search ad campaigns for firms such as Office Depot (Charts, Fortune 500) and REI, companies that remain Marchex customers.

The biggest challenge for the company was building its own source of high-quality traffic - and gobbling up domain names was the first step. Go2Net had already taught the Marchex gang about the power of domain names.

They knew, for instance, that roughly 25 percent of all search inquiries had the top-level domain already attached to them; that is, people would enter a search for, say, "cars.com" or "personalloans.com," instead of simply typing in "cars" or "personal loans." They also knew that about 15 percent of Web surfers bypassed search engines entirely and typed their queries directly into the browser. It was a powerful phenomenon that few people - except for prescient players like Frank Schilling - understood at the time.

So Horowitz and his team went to work, shopping for domains related to hundreds of cities around the country. At first it was all done by hand, with execs almost randomly checking for names that would make sense for their local strategy. "I'd think, 'What about Seattle and insurance?'" recalls Peter Christothoulou, a co-founder and the company's chief strategy officer. "I'd send a note to Russ saying, 'Can you believe no one has this name?' I was buying 20, then 30 or 40 at a time, just as one-offs. I kept thinking, there has to be a better way."

There was. They began employing a method long used by some of the shrewdest domainers - writing software scripts to help them register in bulk. They took scores of cities and states that they knew they wanted to cover and wrote programs that would mash up words and compare them with the list of available domains in the database at VeriSign, which manages the domain system for dotcoms.

The program could hunt for every city combined with the word "insurance"; it would do the same with "auto repair," "doctors," and so on. With this method, Marchex snagged some 100,000 locally oriented names and registered 80,000 zip code domains, such as www.90210.com, on which it's building sites for local neighborhoods.

Yet even at that point, in mid-2004, most of the best names had long been taken by big domainers like Schilling. Mutual acquaintances arranged the meeting with Schilling, who introduced Marchex to a domain pioneer named Yun Ye, whose portfolio was stuffed with names that Marchex coveted. A Chinese citizen who lived at the time in Vancouver, British Columbia, Ye was intensely private and dealt with the Marchex people only under the agreement that they never mention his name. To this day, Marchex execs refer to the deal by the company name Ye adopted for the transaction: Name Development.

Horowitz ultimately persuaded Ye to unload his portfolio of more than 100,000 names and signed off on the $160 million deal in November 2004. But he still needed to hit up Wall Street to raise money.

Domain names were unknown as investment vehicles on the Street, and to help sell the idea, the Marchex crew brainstormed for ways to describe the practice of typing a name directly into a browser. Someone threw out the phrase "direct navigation." It stuck - and Horowitz and Christothoulou made the rounds of investment bankers to explain that direct navigation was a crucial component of the search economy. They raised $241 million with a secondary offering and closed the deal with Ye in February 2005.

It was a watershed moment: Wall Street analysts quickly caught on to direct navigation, giving domain investing an air of legitimacy it had lacked. And other well-heeled players entered the fray. Since the Marchex-Name Development deal, institutional investors and private funds - including Highland Capital and funds backed by Starbucks founder Howard Schultz and Ross Perot - have poured more than $1 billion into domain name ventures.

Until recently, Marchex did what every other domainer does with all its domains: It "parked" them, meaning that it filled the pages with ads (in this case, served up from Yahoo, which had a deal with Ye) and made money when anyone clicked on one of the links. If you type "debts.com" into a browser, a page of ads comes up. Click on an ad, and Marchex gets paid. Sites like those that are packed with nothing but ads generally don't show up in search results, but thanks to direct navigation, they're highly profitable.

When Marchex bought Ye's portfolio, his names were making $18 million a year on revenue of $20 million, margins that would make any CEO drool. But the Horowitz-led team believed that the parking model would eventually fade and that they could make a far bigger business building out the sites into more useful properties. To them, the $160 million - eight times annual sales - was a bargain. "We looked at these properties and saw them as so much more than parking lots," Horowitz says. "It was important to do the landgrab before people started to understand the value."

Horowitz is stabbing the air with his hands and imploring this reporter not to label him a "domainer." Sure, he admits, he sits on one of the biggest domain portfolios around. "But we are not domainers," he says. "That's not who we are. That's not what we're building. We want to develop consumer-friendly websites."

Marchex is having a hard time selling its vision. Since so many of its sites sat idle for so long, packed with nothing but ads, Marchex looked like a giant domain play except with much higher overhead. But the company has been developing new technologies. And in June, Marchex lit up 100,000 of its sites - with another 150,000 or so to go - changing them into destinations with a smattering of content and reviews. The goal is to create sites that, as Horowitz puts it, "have a soul."

In May 2006, for example, Marchex bought a review site called OpenList, a local guide that pulls together reviews for restaurants, hotels, and local attractions. The company then developed software that crawls the Web, sorts out duplicate content, and then generates a review. Look up San Francisco's Hotel Triton on BayAreaHotels.com, for instance, and the software-generated write-up reads like a Zagat guide: "What travelers said they loved: 'The location,' 'the staff,' and 'the room.' Guests can enjoy yoga and other local activities." Users add their own reviews too.

The longer the sites are around, Horowitz says, the deeper the content will get. There will be more user reviews, more content pulled from other sources on the Web. And Horowitz isn't ruling out the idea of human involvement. To help inject these sites with soul, Horowitz recently hired Bill Day, co-founder of About.com, now owned by the New York Times Co.

When Day launched About.com in 1996, he set out to collect consumer information based on topics and locales. The site became wildly popular, with people contributing items corresponding to their interests. The local approach never worked, however, because the audience just wasn't big enough. "We had this broad plan to publish sites on every locality in the world," says Day, now Marchex's media officer. "It was just too early."

Today, however, far more people are on the Web, high-speed connections are nearly ubiquitous, and local businesses are eager to find customers via the Internet - something not easy to do through Google or Yahoo. Moreover, people are more and more likely to surf for services in their neighborhoods. Day says he's considering hiring people around the country to write reviews and blog about specific locales. He doesn't envision hiring a blogger to write about something as narrow as, say, plumbers in Pasadena, Calif., but he could see getting someone to write about Pasadena overall.

Such efforts should boost traffic and attract more local advertisers. Right now, each site has plenty of highly targeted ads - some supplied by Yahoo (Charts), others by Marchex, which has 50 salespeople and thousands of advertisers through deals with companies like AT&T's YellowPages.com. When AT&T (Charts)'s 4,000 salespeople sell ad space to businesses around the country, they also offer packages promising clicks to a customer's site. Marchex runs that part, buying placement on search engines and putting the ads on its own domains.

Over the next couple of years, the company plans to rely more on its own sites - and help people search in a more practical way than the search engines allow.

If you live in New York City and need an orthopedist, for example, you can type "new york city orthopedist" into Google and get back a smattering of references - some ads for doctors, some articles that quote an orthopedist from New York City. Or you can go to Marchex's NewYorkDoctors.com - either by typing the name in the browser or searching on Google (Charts, Fortune 500) - and refine your search in countless ways, by neighborhood or by hours open or by whether the doctor speaks Spanish. "We're trying to let people make decisions online the way they do in everyday life," Horowitz says.

Since Marchex did the deal with Ye, the value of domains has only gone up. And Horowitz is perfectly aware that guys like Schilling have a far more lucrative business model. Schilling, who ended up getting a Gulfstream, makes his annual $20 million with only marginal overhead and no full-time employees.

Marchex, on the other hand, brought in revenue of $134 million in the past year, and had more than $31 million in operating cash flow, yet it booked just $220,000 in profit. In fact, with the stock trading so low, some on the Street have been asking whether the breakup value of the company is higher than the market value, which now stands at about $400 million - an idea Horowitz won't argue with. He's also talked to potential suitors interested in buying Marchex.

This time, though, Horowitz says cashing out is not his aim. He already has the jet and the big house on Lake Washington. What drives him, he says, is the prospect of redefining the way people and businesses use the Web. "We're trying to empower people and local advertisers in underserved markets," he says. "We have the chance to do something profound." 

1 2