Moritz was becoming restive, openly wondering if Schmidt was tough enough. A Google insider with direct knowledge said that in 2002, Moritz pressed for Schmidt to be fired. Another insider said the unhappiness with Schmidt was at first shared by others who also worried that he wasn’t tough enough, that he was moving too slowly to galvanize a management team, to challenge the founders, and most especially, to find a revenue stream.
Schmidt remained calm, at least outwardly. By late 2001, he knew of the effort at Google that Sandberg was now working on to devise the new version of AdWords, the advertising program associated with search. In AdWords as it worked through 2001, advertisers paid Google the old-fashioned way, based on a cost per thousand (CPM) whether the searcher clicked on their ad or not. What Google was quietly exploring was switching to a cost-per-click model, an idea that built on the Overture model. In addition to Sandberg, Omid Kordestani assigned Salar Kamangar, the author of the original Google business plan, to serve as a bridge between the engineering and the sales team as they improvised.
This began a long and intense period of brainstorming. The team liked the idea of charging by the click, thinking it was a way they could farm out not just search, as they had done with Netscape and Yahoo, but also advertising. “We knew we needed a lot of ads, and to have a lot of ads we also had to syndicate,” Kamangar said, performing the search function for sites like Yahoo but also selling ads for them. He knew that Page and Brin would resist allowing advertisers to pay for placement within search results. He also knew advertisers were wary. The CPC model was associated with low-quality ads that were harder to sell and were known as “remnant advertising.” Advertisers like to determine where and when their ads appear, and if they allowed a company like Google to put their ads on other Web sites, and allowed the Web sites to choose the times they would appear, the advertiser would lose control. Was there a system to serve both users and advertisers?
For months they came up empty. Members of the team remember Kamangar walking about with two fingers pressed to his lips, muttering, “I’m thinking, I’m thinking.” One day, he said, a “lightbulb went off.” What if Google combined the cost-per-click model with a measurement of whether users found the ad relevant? Google engineers could come up with an algorithm to measure the quality of the ads, he thought, assuming that more clicks meant users liked the ads. To sell them they could use what economists call a Vickery Auction, an idea suggested by a colleague, Eric Beach. In a Vickery Auction, named after William Vickery, the twentieth-century Canadian economist and Nobel laureate, after Google set a minimum bid price per keyword, the advertiser bids, say, fifteen dollars for a keyword. If the next bid is ten dollars, the winner only pays one cent more than the second highest bidder, saving nearly five dollars; the second-place bidder pays a penny more than the price bid by the third, and so on. The advantages for Google were many. By charging per click, Google could syndicate its ads-sell them on other Web sites as well as on Google search. The more ads it sold on different platforms, the more data Google collected, and thus the more reliant on Google advertisers became. And Google could automate the entire system, minimizing the size of its sales force.
The advantages for advertisers were manifest. They knew they were not being gouged, because they only paid a penny more than the next highest bidder. They benefited by being charged only when the user clicked on the ad. This gave them an incentive to produce a better ad because better ads produced more clicks, which lowered their cost per click. And by charging per click, Google opened online advertising to many small businesses who normally had nowhere else to turn but the Yellow Pages. By allowing Google to syndicate ads, advertisers were achieving the online equivalent of one-stop shopping offered by network television, whereby ads appeared on hundreds of local stations. And because the system was automated, advertisers were spared the expense of a monitoring system. They would simply transmit to Google their keywords, their bid per keyword, their monthly budget, and their billing information. And then using Google Analytics, they could monitor the results online.
Page and Brin made a major amendment to the new AdWords before it was inaugurated in February 2002. At the annual technology, entertainment, design (TED) conference in Monterey, they engaged in conversation with the Israeli entrepreneur Yossi Vardi. Vardi is a bear of a man with a walrus mustache and a friendly, even impish manner. His company started ICQ, the Internet’s first instant messaging system, and sold it to AOL for four hundred million dollars. He and the Google founders discussed search ads-how to make them unobtrusive and yet relevant to users. Vardi suggested that they could use two-thirds of the page for search results and wall off the text ads from the search content the way a newspaper walls off ads. They could do this by placing a thin blue line between the search results and a smaller gray box on the right-hand side of the page containing the text ads and links to the advertiser. Users could either click on the link or not. Vardi’s idea, Brin recalled, was the genesis for the way ads were displayed. Page and Brin decided the ads should be small, a couple of lines long, imposing a limit of ninety-five characters, and insisting that they be informational.
It was unclear when the new AdWords was introduced that it would be what it became: a Google money machine. “The AdWords is brilliant because it allows you to scale the advertising solution to what you need,” said former Microsoft executive Nathan Myhrvold. It democratizes advertising, allowing Google to use it for either small or large advertisers. It was also, Myhrvold believes, pirated from Overture. The rival search engine thought so too, and later that year filed a patent infringement lawsuit against Google.
A year later, a second money gusher-AdSense-would spring from the CPC model. At the time, Paul Bouchet was developing Gmail and working on software to match words sent in an e-mail with keywords selected by advertisers, allowing small text ads to instantly appear. Brin wondered why they couldn’t apply this innovation to a new program that would help bloggers and any Web site make money. This idea would be called AdSense. If a reader was looking at an analysis of computers on a Web site like Engad get, an HP or a Dell ad could appear. Similarly, readers of a story about the law in an online newspaper might see a law firm’s ad, while people looking at a Web site devoted to pancreatic cancer could see ads for pharmaceuticals. Google would serve as the matchmaker, delivering the advertising and sharing the revenues. As with AdWords, the advertiser would pay only when the ad received a click. And as AdWords democratized advertising, luring small advertisers online, so AdSense would become a way for Web sites to generate income. The effort was led and architected by Susan Wojcicki, vice president, product management, who later received the prestigious Google Founders Award-paying about twelve million dollars-to honor her efforts. AdSense, Danny Sullivan told USA Today, “basically turned the Web into a giant Google billboard. It effectively meant that Google could turn everyone’s content into a place for Google ads.”
Eric Schmidt recalled how Brin lobbied him for money to market the program. “He and an engineer developed a system of showing ads on people’s blogs or Web sites. They came to show this to me. It was not an exciting demo. And Tim Armstrong’s sales guy is assigned to help them out. Now we’ve got three people out of control! So Sergey comes in and said, ‘I need to buy inventory to make this happen.’”
“How much?” asked Schmidt.
“I need a million dollars,” said Brin.