The Google Guys (11 page)

But those ideas did not take the form of an actual formal business plan. In fact, there was no business plan, which would normally lay out in detail the proposed revenue stream, with projections on how fast revenues would grow in five years. “We worked on a business plan for a little bit, but we were basically never even asked for it,” Sergey told Lemm. He then added, “Recently we got an e-mail from one of our investors saying, ‘Oh, do you guys have a business plan? I don't think I ever saw one.' ”

Although they didn't yet have any idea how the ads would work, they already knew one thing: the ads had to be useful rather than an annoyance. Said Sergey: “The key there is to put up advertising that will be really useful to our users and not slow down our site. That way we won't push people away from our site, but we'll still take in some revenue.”

It took them nearly three years to figure out how to fill the requirements Sergey had stipulated. Silverstein says that Larry and Sergey felt that no advertiser on the Internet had solved that problem. It's likely none of them was trying. Just as Larry and Sergey demonstrated that there was a huge market for a search engine that gave better results, they set out to show that there was a market for Internet advertising that gave better results, with the needs of the user—not the advertisers—given the highest priority. “We had this idea that if we could get a lot of users, we could make money,” says Silverstein. “That said, we did not have advertising for a long time because we couldn't think of a way to do it that we thought was good for our users. Which I think gets to what Sergey was warning about. There are a lot of ways to do advertising. It took us quite a while to find a way that was actually beneficial to users and have an appropriate separation between editorial and advertising. We noodled it over, we talked about it.”

Larry and Sergey have maintained this attitude that advertising should be done only when it helps the user in some way. In 2006, engineers met with Larry and Sergey with a simple proposal: to include ads with image-search results. They argued that this would add $80 million a year to revenues. Larry's response was to ask: “We're not making enough money already?” Sergey was equally skeptical. “I don't see how it enhances the experience of our users,” he said. They rejected the proposal.
⁶

In 2002, most advertising at the time, including ads on Microsoft's site, MSN, came in the form of banner or display advertising, flashing billboards that appeared at the top of the page. Some search engine/portals were already “selling” search terms to advertisers, charging them for making the ads appear when people searched using certain words. But most contracts set a predetermined price negotiated by ad reps, rather than using an auction. As CEO Schmidt puts it, the philosophy was, “Give us half a million dollars and we'll show the ad whenever it's appropriate.”

A company called
GoTo.com
(later renamed Overture) had come up with the idea of an online Yellow Pages system, where users would type in search words and be taken to advertisers who bid to have their ads appear when people searched with those words. (In 2002, after Google started showing the way, Microsoft and Yahoo both considered buying Overture. Yahoo won the bid.)

Larry and Sergey started placing ads next to search results almost immediately, in 2000, but also sold them for a set fee. But they were not satisfied with that; they had to be sure the ads were as relevant to their users' interests as were the search results.

Google executives today insist that their approach was unique, and continues to evolve. “The idea came out of GoTo,” says Google chief economist Hal Varian. “But [GoTo] didn't really improve it. We took their model and refined it. When the guys at Google looked at it, they figured out how to advance it. This strong emphasis on quality came from Sergey and Larry.”

In order to ensure that quality, Google essentially created a separate search engine dedicated to determining the relevance of ads, allowing them to select from millions of ads before displaying them. “Larry and Sergey came up with the targeted ad model,” says CEO Schmidt. “That model coupled with search is a gold mine.”

The focus in the early days was small advertisers—a market that would put Google in competition with eBay online and classified advertisers in newspapers. It fit well with Larry's and Sergey's desire to focus on small players and individuals. The two founders believe in the idea of the “long tail,” all the millions of people on the Internet who wanted services that were not being offered by others. In Google's 2006 annual report to shareholders, Sergey discussed the importance of small advertisers: “Our goal is to create a single and complete advertising system. Diversity in our advertising and publisher base continues to be central to our business and is important to our long-term success. Advertisers large and small use Google to reach their target audiences easily and get measurable ROI [return on investment].... As more and more users look for local information online, we must continue to improve our ability to attract local advertisers. This year we partnered with companies . . . to help us bring more business information online and convert more small businesses into happy Google customers. Small business is big business.”

Because prices were to be low, sometimes just pennies per click, volume had to be extremely high. This required a fully automated system with very little human intervention.

What they came up with was a system that would let advertisers bid online to set prices, with those ads automatically matched to search terms without advertisers' ever talking to an ad rep. “Getting to that vision of what [advertising] could be was the big bottleneck,” says Silverstein. “We could run a system ourselves that could support having a million advertisements from one advertising company [i.e., Google] and just show the right ad for the right kind of search.”

After some testing, it seemed to work. So in January 2002, Larry and Sergey gave the go-ahead to convert Google's premium ad system—in which fixed-fee ads were placed in a box at the top of the search results—with the auction-based AdWords program. Larry and Sergey were solidly behind the switch, but Schmidt was worried that the auction system would not set ad prices as high as those the Google ad reps were getting. “I said, ‘Promise me that revenue won't fall.' I was terrified.”

As a precaution, Schmidt instituted a period of restricted spending, also known within Google as the “crap period.” The rule was wickedly simple: people could spend money only one day a week. Every Friday morning at 10:00 A.M., anyone who wanted to buy something had to head to Schmidt's office to justify the expense. But this lasted about three weeks. By then Schmidt had discovered that the ads priced through bids were collecting twice as much revenue as the ones that had been sold by the ad reps. It turned out that the reps were pricing the ads too low for the market. One of the beauties of Google's ad system is that it automatically reaches exactly the price the market will bear.

It was the support Larry and Sergey gave that made the difference. Other companies did not have the push from top management to take that leap. In November 1998, Microsoft bought a company called LinkExchange, which was in the business of distributing ads to other Web sites. Along with that acquisition came a man named Scott Banister, a young college dropout who had come up with the idea for something he called Keywords, a business he had sold to LinkExchange. His concept was to create a software system that would auction off search terms to advertisers, placing the ads next to the search results.

In early 2000, Microsoft's online group ran an experimental system to match search queries with ads. But some managers were worried that the system would eat into display ad revenue, so those ads were shoved to the bottom of the page, and minimum bids were placed at fifteen dollars per ad. In May the service was shut down. Despite the fact that Banister's boss, Ali Partovi, pitched the concept to Microsoft as “the next big thing,” executives at Microsoft rejected the idea.
⁷
They were too tied to banner ads. Microsoft executive Satya Nadella later admitted that shutting off the service was, in retrospect, “a terrible decision. But in all honesty, none of us saw the paid-search model in all its glory.”
⁸

Microsoft wasn't the only company to reject the idea of socalled “search advertising” as a revenue model. Partovi, frustrated with his lack of progress at Microsoft, started shopping the idea to others. Yahoo also turned him down. So he tried Google. But Larry and Sergey had already started pursuing the idea on their own.

CEO Eric Schmidt has also made significant contributions to the development of Google's advertising system. The biggest may be in luring economist Hal Varian to the company.

Varian, a sandy-haired professorial type with Bill Gates glasses, was dean of the business school at UC Berkeley in 2002. He had written
Information Rules: A Strategic Guide to the Network Economy
, a book that discusses how to market and distribute goods in the network economy, including how to price them. He had also been involved with Inktomi, the search engine that spun out of the university. When he ran into his old friend Eric Schmidt at a Super Bowl party in January 2002, Schmidt, who had read Varian's book, said, “ ‘Why don't you take a look at this ad auction?' ” Varian recalls. “ ‘I think it might make us a little money.' ”

Varian wasn't sure he was impressed enough with Google to make the leap from academia. That spring, Schmidt and Larry met up with him at a conference for big thinkers at the Aspen Institute. Varian wondered who the young man was with Schmidt. “I thought, gee, why did Eric bring this kid along? He could have been in high school, as far as I was concerned.”

But the two persuaded him to meet with some people at Google the following April, and in May, Varian took an academic leave from Berkeley to consult for Google.

In 2007 he became the company's chief economist. “It's a lot more fun here than at Berkeley,” he says.

One of Varian's areas of expertise was predictive pricing. He worked on a system that could predict the rate at which people would click on ads and compare it to the bids people placed. It's part of a discipline called economic mechanism design, and it includes elements of game theory. The system allows advertisers to decide how much to bid based on the number of predicted clicks on an ad. This is an important feature, since advertisers pay their bid price to Google only when somebody clicks on their ads. In order to make the system more efficient for advertisers, they are allowed to see where they stand in the rankings and the bids made by other advertisers.

Larry and Sergey did not scrimp on investing in the technology to handle the ad placement system. “I've got to say, they thought big,” Varian says. “They built a very large and complex system. It was significantly more ambitious than the competition's.”

And this was at a time when they spent little money on the normal accoutrements of a growing company. “Most of their money went into the servers,” Varian recalls. “There were still five people to an office, with desks made from doors and steel legs. That was mostly Larry. He plays his cards close to his chest. When [Larry and Sergey] started to understand the potential of auctions, the way growth was going on, they didn't move into bigger offices. The goal was to fly beneath the radar as long as possible. People were very surprised when numbers were released about how profitable Google was.”

Larry's famous secrecy, however, was something an academic researcher was not accustomed to. The system Varian first worked out in the summer of 2002 was used internally, but was not publicized, or even written up in academic papers. But he was invited to give a talk at Stanford, and was allowed to say a little about the work, since it was a small class. A visiting professor there, Mark Schwartz, had been working on a similar idea, and talked to Varian about it. “It became pretty clear that he was going down the same path,” Varian says. So after his talk, he went back to Google and reported that other researchers were working on the same ideas, and asked if he could publish a research paper on the work. “It went all the way to the top, to Larry and Sergey,” he says, and they finally gave their approval.

Still, other companies were slow to catch on. A couple of years later Varian presented the ideas to a Yahoo executive. “He said, ‘What?' ” Varian recalls. “He had never thought about it. That amazed me, because we knew about this in 2002.” After that, Yahoo hired its own chief economist, as did Microsoft.

The advertising system designed at Google has another clever quirk. Instead of just giving the highest ad placement to the highest bidder, Google created a feedback loop to give preference to the most effective ad, not the priciest. Each ad builds up a reputation. If people don't click on it, it drops in the rankings, while lowercost ads rise. Google executives have said the reason for this is to ensure that the ads are as relevant to the users as possible. But, once again, it turns out to be the most profitable approach. It's better to get many clicks on a low-paying ad than none on a highpaying one. It's obvious in hindsight: keep the most-clicked ads at the top of the page, and Google collects more revenue.

Competitors, however, questioned the originality of Google's ad system from the beginning. In 2002, after Google introduced its AdWords program, Overture sued Google for violating its patents. At the time, Overture spokesman William Tell took a few well-targeted shots at Google. “We've spent millions on attorneys' fees and lawyers' fees,” he told TechUser.Net. “We're not going to let some company with a bunch of hot shot programmers come in and steal our best ideas.” The suit was settled in August 2004, after Yahoo had bought Overture. Google was granted a perpetual license to the patents in return for payment in the form of 2.7 million shares of Google stock, about 1 percent of the company at the time.