■
LAUNCHED AT THE TED CONFERENCE in February 1998 by entrepreneur Bill Gross, GoTo.com was conceived of as a completely new type of search engine. Instead of search results generated by spidering the web and returning pages based on an algorithm, GoTo returned results that were almost exclusively provided by sponsors. GoTo served up text ads designed to look like search results, but which were paid for by advertisers who bid for position. It was an eBay-like auction model. For any given keyword, a company could pay whatever it cost to rank first for that search term. If you wanted to show up first on a search for “flowers,” you could bid, say, 10 cents a click. If someone bid 7 cents, they were listed second. Bidding a nickel might get you third place, and so on. If you wanted to go crazy and bid $1,000 a click, you could theoretically rank number one for any search term you wanted.
The idea of a “search” engine that only returned ads was extremely distasteful to most; indeed, Gross was nearly hissed off the TED stage during his presentation. But advertisers loved the idea, and signed up in droves because they quickly intuited that Bill Gross had stumbled upon one of the greatest advertising models in the history of the world. Paid search represents a uniquely powerful nexus point for advertisers to insert themselves into. Users who search are searching for something. You don’t perform a search like “hotels in Marietta Georgia” without having at least some passing interest in booking a hotel in that city in the near future. Advertising around search allowed marketers to reach consumers at the very point of intentionality, at the very moment they were either researching a purchase or actually looking to buy.
An important component of this entire process was the ability to “pay per click” —as opposed to paying based on the number of people who (theoretically) viewed your ad, as every other online advertiser did in the dot-com era. This was the second key innovation: with the GoTo model, an advertiser only “paid for performance.” If no one clicked on your ad, you paid nothing. This was a radical but extremely enticing option at a time when click-through rates on banner ads had dropped to minuscule percentages.
Gross had intended for GoTo to become a shopping destination, thus the active tense of the name. And yet, even though advertisers eagerly signed up to hawk their wares, the consumers didn’t follow. Undeterred, Gross had the brilliant idea of chasing the traffic he needed. GoTo approached nearly all the extant portals and search engines and offered them what was essentially free money: GoTo would “syndicate” its paid search results so that for almost any keyword on a site like AOL Search, the first three or four results would be GoTo’s text links which, though they looked just like the other search results, would actually be ads. When searchers clicked on these paid links, GoTo would share the ad revenue with the portal, thereby instantly monetizing the search traffic.
GoTo succeeded in signing deals with all the major portals, and at a stroke, turned search—which had been a loss leader for portals throughout the nineties—into a cash cow. In 2002, GoTo changed its name to Overture to better reflect its true business model of introducing customers to advertisers. The company was earning more than $78 million dollars a year on $668 million in revenue—all from paid clicks syndicated to the likes of Yahoo, AOL and MSN. Overture saved the portals by fixing a fundamental flaw in their business model. Portals had sprung up in the first place because they needed to be “sticky.” None of the early search sites could make money when they sent users out onto the web, so they attempted to hoard the eyeballs and keep them on-site in order to create impressions for banner ads. But now, clicking itself was finally worth something. As the writer John Battelle has put it, Overture could generate billions of dollars, one click, one nickel, at a time.
GoTo/Overture came along at a very opportune moment for the Internet. As the bubble burst and the advertising market cratered, paid search stepped into the breach to replace the lost revenue from all those bankrupt dot-com advertisers. In the case of Yahoo, by the summer of 2002, the paid links it was getting from Overture accounted for more than 10% of the ailing portal’s revenue, and almost all of its much-diminished profits.13 It’s not an exaggeration to say that Overture and paid search saved the portals and the search industry in general. Lucky for Google, there was now a very lucrative new advertising model it could copy, and what was more, this new form of ad had proven the immense value of search, Google’s crown jewel. But since Larry and Sergey never met an idea they didn’t think they could improve upon, Google was not interested in merely copying. If Google was going to have ads, the ads would have to be better than traditional ads; they would have to be useful.
■
GOOGLE’S FIRST EXPERIMENT with advertising came in January of 2000 when it began showing unobtrusive text (in keeping with its minimalist aesthetic) links above certain keywords. But the ads were still priced like flashy banner ads, on the traditional CPM (cost per impression) model. Page and Brin wanted something more scientific, more automated. They liked how anyone could buy an ad through Overture by simply using an online form. In October 2000, they launched what they called AdWords, which allowed any advertiser, no matter the size of their operation, to purchase search ads online in a matter of minutes, using a simple credit card.
As GoTo/Overture had discovered, advertisers were eager to get in front of Google’s burgeoning search traffic, and the first influx of AdWords advertisers put an end to Google’s immediate money issues by bringing in $85 million in 2001. But since the ads remained CPM-based, advertisers were still paying for impressions, not for actual clicks. Google was missing out on the performance-based advertising revolution, and it showed. Overture’s 2001 revenues were $288 million, and that number was growing at a faster rate than Google’s.14 In February 2002, Google unveiled a new version of AdWords that copied Overture’s cost per click and auction-pricing model. In typical Google fashion, however, its Overture clone had a key innovation that made all the difference in the world.
The new version of AdWords had advertisers bid against competitors’ ads, but Google’s system was not strictly pay-for-placement. Ever enamored with math and the power of algorithms, Google introduced an important new ranking factor for the ads that it called a “Quality Score.” In essence, Google’s system took into account how often that ad was actually clicked on, in addition to how much an advertiser was willing to pay per click. Each time a search was run and AdWords results were generated alongside the search results, the ranking of the eventual ads decided how relevant the ads actually were. This prevented deep-pocketed but ultimately irrelevant advertisers from dominating every keyword. You could no longer guarantee to rank high just by being willing to pay the most. Your ad also had to be clicked on the most in order to rise up the rankings. Successful advertisers paid less per click, but ranked higher. If your ad was of good quality, and tended to get clicked on more often, AdWords trusted that it was more relevant for that search phrase and would therefore rank you higher even if you didn’t increase your bid. Google did this because, almost counterintuitively, it knew that it stood to make more money when the ads were ranked this way. Over time, more money would come in from a 5-cent ad that was clicked on twenty-five times—than from a dollar ad that was only clicked on once.
From a searcher’s perspective, the ads felt less annoying the more relevant they became. To a certain extent, Google’s AdWords began to seem almost as useful as the organic search results for certain keywords, because the quality score kept them germane to the searcher’s original query. And on the advice of early Google advisor Yossi Vardi, the bulk of the AdWords appeared on the right-hand third of the search results page. This had the consequence of increasing the amount of ads delivered per search, all while seeming to make them less intrusive. The original, organic search results still filled the main two-thirds of the page, pristine and untarnished. When Google ran limited control experiments where it showed one group of searchers results without the ads, and another group search results with the ads, users who saw the ads actually searche
d more.15 It became a classic win-win-win: Google started making more money per search than Overture did, advertisers felt like they were paying less per click while reaching more potential customers, and users felt like they were getting supplemental search results, in the form of ads that were often quite useful.
Overnight, Google’s fortunes were transformed. Led by a new hire named Sheryl Sandberg (later, more famous for her leading role at Facebook), AdWords became the blockbuster success that Google had been looking for all along. It helped considerably that Google had what Overture didn’t: its own highly trafficked search destination. Google didn’t have to cut syndication deals with other portals in order to get traffic for its ads, since its own website was already servicing hundreds of millions of searches per day. It didn’t have to cut deals, but it did anyway, especially a partnership with AOL, announced in May of 2002. Google would not only provide organic search results to AOL, but paid search results as well, stealing the business away from Overture, which had previously provided AOL’s paid links. Two thousand two would become Google’s first profitable year, with $440 million in sales and $100 million in profits.16 By 2003, profits were more than $185 million and the AdWords program could boast more than 100,000 advertisers, all without a commensurate rise in Google’s head count, because the AdWords sales system was automated.17
In retrospect, going into advertising played into Google’s deepest strengths. For a company full of data-obsessed nerds, Google looked at advertising as just another problem smart algorithms could solve. Indeed, serving the appropriate ads alongside the organic results, running auctions in real time for billions of searchers, and reranking the ads according to their performance became an even more complicated algorithmic trick than even search had been. But then, Google’s entire infrastructure was devoted to crunching numbers and organizing vast amounts of data, so it was uniquely positioned to get this sort of thing exactly right. Just as with web search, when Google turned on its new advertising algorithms, it found that the ads got better over time; so much so that Google’s computers could eventually predict with stone-cold accuracy which ads would work and which wouldn’t.
Google can be thought of as a company born from two miracle inventions, one of which it came up with itself, and the other of which was cribbed from Overture. Definitively solving the problem of web search is obviously the miracle that has made the largest impact on our society. The web and the Internet itself are now so big that without decent search, it’s easy to imagine that the whole edifice would have collapsed under its own complexity by now. But by improving on Overture’s pioneering work with paid links, Google was able to achieve something just as amazing: it made the Internet profitable at scale and for the first time. Paid search would prove to be the greatest advertising engine yet devised by man. Furthermore, algorithmically served ads would support nearly every product Google would release subsequently: Image Search, Google News, Gmail, Google Maps, Google Books. In a few short years, search ads would surpass traditional banner or “display” ads, and within a decade, Google would be generating more than $50 billion in revenue,18 having captured nearly 50 cents of every dollar spent advertising online. Today, most advertising is automated in ways similar to what Google pioneered, and even now the largest market for online advertising remains tied to search. It turned out that the gold mine on the Internet was search all along, as Yahoo and others had first intuited, but had subsequently forgotten.
■
BY 2003, GOOGLE WAS OBSESSED with one thing: keeping all this a secret. As ever, Google feared tipping Microsoft off to the value inherent in search. Sure, Microsoft was ailing from the antitrust trial and was already entering its lost decade, but it was still the only technology company that had the resources, talent and size to do to Google what Google had done to Overture.
Helping to keep Bill Gates and company in the dark was Google’s new “grown up” CEO, Eric Schmidt. Schmidt had been a longtime Microsoft adversary going back to the 1980s, when he was an early manager at Sun Microsystems, and then briefly in the 1990s as CEO of Novell. Years of experience managing a relationship with Microsoft no doubt played a role in Schmidt’s eventual selection as CEO, but a willingness to swallow his ego was probably what put Schmidt’s candidacy over the top. Becoming the Google CEO meant having to share the limelight—as well as some degree of the decision-making process—with Google’s founders. Indeed, the working relationship Schmidt would go on to form with Page and Brin evolved into a sort of triumvirate where all three had meaningful say. Though, if push came to shove, the founders could outvote the CEO. Page and Brin’s dream candidate for the job had been Steve Jobs, but it’s hard to imagine the Apple founder being willing to take a back seat to two twenty-seven-year-olds, as Schmidt eventually agreed to do.
Capable management was crucial as competitors circled. Thanks to its investment in Google, Yahoo had the best inkling as to what was really going on behind the scenes at the Googleplex. In the summer of 2002, only a few months after the new version of AdWords debuted, Yahoo made a $3 billion bid to buy Google outright. Google, with Schmidt newly at the helm, turned down the offer. Too late, Yahoo realized that search was the motherlode of business models, so it canceled its organic search partnership with Google, purchased what was widely considered to be the company with the second-best search technology, Inktomi, for $257 million, and in 2003, paid $1.4 billion dollars to acquire Overture. The idea was to combine the two properties under the Yahoo umbrella and replicate Google’s algorithms-and-advertising juggernaut, complete with a quality score and bidding systems that mimicked AdWords in efficiency and effectiveness. Called Project Panama, this next-generation system was not released widely until February of 2007, by which point Google had run away not just with search market share generally, but virtually the entire search advertising market.
By then, the whole world knew what Yahoo had intuited: Google was printing money. On April 29, 2004, Google filed for an initial public offering of stock. It would be the highest-profile technology IPO since the dot-com bubble burst. When Google released a snapshot of its financials so that potential investors could evaluate the company’s prospects, both the technology and financial worlds were amazed. Venture capitalist Mitchell Kertzman told the Wall Street Journal that Google’s numbers were “stunning.”19 Google’s PR head David Krane remembered the general response being “ ‘Holy shit!’ ”20 Google had generated more than half a billion dollars in cash flow in 2003 and its operating margins stood at an astounding 60%. These were Microsoft-level numbers.21 The online market for search ads had reached $2.5 billion in 2003 (nearly tripling the size of the market from the $927 million spent a year before), and Google had captured approximately $1 billion of that.22 A lot of this success was thanks to the fact that 35% of all web searches were now being done through Google, surpassing Yahoo’s 30% market share for the first time.23
Brin and Page had not actually wanted Google to go public, having filed only because financial rules put into place after the dot-com bubble burst would soon compel them to do so. In the letter the founders wrote to prospective investors, which they called “ ‘An Owner’s Manual’ for Google’s Shareholders” (and which the New York Times declared to be “part financial document, part populist manifesto”)24 Google’s founders began with a simple statement: “Google is not a conventional company. We do not intend to become one.”25 Brin and Page went on to state their intention to continue to operate Google in the service of their own lofty ideals, to “develop services that improve the lives of as many people as possible—to do things that matter” rather than bow to the quarterly whims of Wall Street’s expectations. Throughout the coming months, as the ramp-up to the IPO began, the Google guys were accused of “thumbing their nose” at Wall Street and its traditions.26 Larry and Sergey demanded that the underwriters of the IPO receive a fee of only 2.8% for their services, about half the rate bankers usually expect.27 During the “road show” when the founders crisscrossed the country,
ostensibly to sell the company to investors, Larry and Sergey drew fire for flat-out refusing to answer specific questions about Google’s operations or future plans.28 Even the amount of shares Google was offering to the public was a bit of a prank. Google wanted to sell exactly $2,718,281,828 worth of equity. Math geeks (like the Google founders) knew that this number represented the first 9 decimal places in the mathematical number e, which is, of course, an irrational number.29
On August 19, 2004, Google went public at $85 a share, and rose 18% on its first day of trading, to close at $100.34. The 38 million shares that Larry and Sergey each held in the company were worth approximately $3.8 billion at the close.30 Google was valued at $27 billion,31 more than a bit behind Yahoo’s $38.7 billion market cap. But that disparity wouldn’t last long. By the time Google’s first quarterly report as a public company revealed that sales had doubled from the previous year, Google stock passed $200.32
It is impossible to overstate how important Google’s IPO was to the Internet, Silicon Valley and the stock market overall. As the New York Times said on the day after the company filed to go public, it was “as if the dot-com glory days never ended.”33 Google’s success was validation that the Internet as a social, cultural, and (most important) a financial phenomenon was not dead. The revolution had merely been regrouping. Google was also proof that not only were some of the original ideas from the dot-com era still valid; some new ideas might also be out there ready to build on the dot-com era’s faded promise. Within Google itself, there were whispers of exciting new projects, like, some sort of a Google “phone” so that searchers could get answers to queries at any moment no matter where they were.34 More than anything, Google’s success provided the template to make these new ideas profitable. Just as with the Netscape IPO nearly a full decade before, a new generation took notice: there was fire in Silicon Valley again.
How the Internet Happened Page 26