I'm Feeling Lucky (31 page)

Susan countered that "AdsDirect" reminded users they didn't need a middleman to create or place their ads and thus suggested the benefit of a discounted price.

Cindy expressed a slight preference for BuyWords, because she viewed it as a way to engage the press on our refusal to sell placement in results. But she could live with DirectAds too.

Once again we were at an impasse. I went home that night and spent an hour generating more names while my boys did their homework and Kristen gave the baby a bath. In the morning I sent out a new list of possibilities.

"promote Control."

"Ad-O-Mat."

"Ad Commander."

"Impulse Ads."

I didn't love any of them, but I did feel pretty good about the last name on my list: "AdWords."

"It's new! And improved!" I shilled. "It's like BuyWords without the Buy! It also sounds like Edwards, which is a big plus in my book."

Salar liked AdWords better than BuyWords. Omid liked it. Bart liked it. Larry liked it.

Sergey cast the final vote. He informed the engineering team that our new system would be called "AdWords." He neglected to tell me—or anyone else in marketing—about his decision. It's hard to imagine such an oversight at a brand-driven company like Procter & Gamble. From an engineer's perspective, however, the name question only needed to be resolved so the proper string of characters could be entered into the program before the "code freeze," at which point no major changes were allowed. The marketing stuff that went along with establishing a brand was secondary and could be dealt with once the program had been locked down. It was Jen McGrath of the front-end team who let me know that my family's name would live forever, enshrined in Google's revenue stream.

AdWords had a long way to go before it could go out to the public. The engineers wrestled with the software that would actually place the ads and charge for them. The UI team worked with Salar to finalize the interface advertisers would navigate to create and activate their campaigns. PR would introduce AdWords to the press. I owned the language we used to explain it directly to our users. There was a great deal to explain. We needed an FAQ, error messages, terms and conditions, customer service emails, and sales materials. And engineering kept changing things in the weeks leading up to the launch. Many long nights ensued.

I didn't mind at all, because AdWords added another dimension to my role. I was not just building Google's brand but also our bottom line. From that point on, I wrote almost all of the words that appeared on our website aimed at either consumers or clients. I began envisioning myself as the human interface between the company and the people who used our products. It may have gone to my head a bit. If I was going to be the voice of Google, Google's voice would sound like me. I terrorized colleagues with imperious screeds about tone, style, and the overuse of capitalization and exclamation marks. Engineers viewed letter size and punctuation as knobs they could twist to raise excitement in otherwise uninspired writing—LOOK! THIS MUST BE IMPORTANT! I emphatically told them that most exclamation marks were completely superfluous! Unless they were writing about Dick and Jane! And Spot!

It all got done.

AdWords launched quietly in a beta test version late on September 27, 2000. Two hours later, Lively Lobsters in Kingston, Rhode Island, signed on as the first customer (keyword: lobster) with an eighty-three-dollar ad buy. Ryan Bartholomew, Lively Lobster's owner, ad manager, and sole employee, quickly figured out that AdWords would be an easy way to make money from other sites' affiliate programs. He placed AdWords ads for books on
Amazon.com
, as Jeff Dean had done with his first Google ad system, and earned a commission that more than covered his costs. He found adult-site affiliate programs to be especially lucrative. According to Bartholomew, he grew his affiliate optimization empire to a sixteen-person business that placed more than twelve million dollars in ads on Google over the next decade.

Lobsters and porn. From its very first customer, AdWords proved its versatility and potential as a moneymaking tool. Twenty other customers signed on in the next two days, and with them came new issues we hadn't considered. Someone wanted to use "Google" as a keyword trigger to display an ad. Was that okay? Jane Manning, the project manager, decided it set a bad precedent, and we disallowed it. Should we treat ads targeting the trademarks of other companies the same way? An engineer suggested we forbid any trademark targeting, but how, then, I wondered, would you handle words like "staples," which is both a company name and a generic term? Larry and Sergey decided not to restrict others' trademarks, but the issue would eventually go to the courts, who at the time of this writing have sided with Google. That was part of the excitement of working for a company unrestrained by precedent. We improvised as we went along and watched as the rest of the world caught up.

AdWords formally launched on October 23, 2000, and accelerated past the outer moons of Jupiter on its way to some distant galaxy made entirely of money. No PR crisis ensued. If inappropriate ads did run, no one noticed because they were merely wet spots within an ocean of searches. Still, we instituted a limit to ensure ads were reviewed before we delivered too many impressions, and we hired a temp to check each new ad by hand.

Would Google never tire of succeeding with big ideas that I found patently ludicrous? It was starting to make me feel like a crotchety geezer yelling at kids to get off his lawn.

I had christened AdWords with my own name. Next I worked on giving it my own voice. Emboldened by the feedback on MentalPlex and my growing confidence as Google's word guy, I started gently introducing bits of humor where they weren't specifically called for.

"Okay, we've got a problem here," I wrote for an error message. "Not only did something not work right with our software, it went so spectacularly wrong that our automatic report generator can't begin to describe it for our engineering group. You can help." I went on to suggest that users email us details including their horoscope signs.

"Sorry, something didn't work correctly," another message read. "If we knew exactly what the problem was, we would tell you instead of giving you this useless error message. Actually, if we knew, we would most likely have fixed it already. Rest assured. A report will soon be in the hands or our engineering team detailing the bad thing that happened here."

When the AdWords team not only raised no objections but thanked me profusely, I rejoiced. I no longer worked at a company where everyone wanted a cut at neutering my language. My colleagues didn't know "professional" marketers would assume user ignorance, target the lowest common denominator, and eschew polysyllabic words. I wasn't about to tell them. Besides, they were focused on the code swimming across their screens. Words not preceded by a command prompt were insignificant.

I began writing copy for the site as if the person reading it were a friend. I added Simpsons references to our FAQs, made puns in our newsletter, and, after engineer Amit Patel confessed a love of prosimians and their googley eyes, started including lemurs in all my examples. ("I don't want anyone to know I'm into lemur racing. Is my information private?") It made my job a lot more fun, but also made it clear that an actual human being had touched the page the user was reading.

"Sometimes we have to shut the service down to implement improvements," I wrote about an unstable product launch. "Sometimes it decides on its own to break for a nice pot of Earl Grey and some fresh silicon wafers."

I found out humor wasn't so funny when it came time to translate our FAQs into languages in which the jokes didn't work. Or when a joke wore out from overexposure. The error message I wrote for orkut, our experiment in social networking, quickly grew tired as the service crashed time and again.
^*

"Bad, bad server," it said. "No donut for you. We're sorry, the
orkut.com
server has acted out in an unexpected way. We apologize for the inconvenience and our server's lack of consideration for others."

I thought I had written an amusing tribute to an obscure half-remembered cartoon, but the page became the focus of intense user frustration. A Google search for "bad, bad server" still brings up close to ten thousand results, most of them rants about connectivity problems.

Despite the occasional misfire, I knew I had found my stride. I was increasingly aware that Google was developing a voice and increasingly confident in my ability to speak with it, not just to users, but to those working inside the Googleplex as well.

In November 2000, almost a year to the day after I started at Google, I wrote two lines that defined for me exactly what Google should be as a new kind of corporate entity—two lines that distilled the essence of the company I wanted to be part of and believed I had joined. It was the click of a tumbler falling into place, securing my role and locking in the sound of Google's voice once and for all. It was good timing, because Google was adding new people every day. They brought fresh energy and ideas, but with new blood the complexion of the company couldn't help changing.

I
WAS AT LUNCH
," said Allegra Tudisco, the new marketing coordinator I'd hired in September, "and I introduced myself to everyone at the table. They all gave me their names except this one guy, who seemed kind of shy. I asked him who he was and he said his name was Larry, so I asked him what he did. And he said he was Larry Page, the CEO. I had no idea."

Google had reached a cultural milestone. The company had grown so large that Larry and Sergey no longer had time to interview each hire personally. That proved a problem for Larry a week later when he visited our new nine-person office in New York. They refused to let him in because they didn't believe he was who he claimed to be.

Mostly, however, the growth meant we had more going on. More work on infrastructure, more deals in negotiation, more salespeople calling on clients, and more products in development. One of those products was a toolbar that tucked a search box right into users' browsers, enabling them to conduct Google searches without going to
Google.com
. The product had been worked on by Joel Spolsky, a contract developer, based on prototypes developed by my UI team colleague Bay Chang. Googler David Watson created the first working version, and Eric Fredricksen finalized the software we actually launched. Larry was very keen to get it out the door.

The Google toolbar came with more than just a search box. It had "advanced features." One displayed a green bar with a relative length approximating the PageRank of the web page the user was visiting. PageRank was Google's assessment of the importance of a page, determined by looking at the importance of the sites that linked to it. So, knowing a page's PageRank could give you a feel for whether or not Google viewed a site as reliable. It was just the sort of geeky feature engineers loved, because it provided an objective data point from which to form an opinion. All happiness and joy. Except that when the "advanced features" were activated, they also gave Google a look at every page a user viewed.

To tell you the PageRank of a site, Google needed to know what site you were visiting. The Toolbar sent that data back to Google if you let it, and Google would show you the green bar. The key was "if you let it," because you could also download a version of the toolbar that would not send any data back to Google. The user could make the choice, though Larry and the engineering team believed—and hoped—that most people wouldn't pass up the advanced features just because Google might learn their surfing habits. We're talking free extra data here. While knowing the PageRank of a page might have only nominal value to users, knowing the sites users visited would be tremendously valuable to Google. The PageRank indicator provided a justification for gathering it.

I thought it was an enormous privacy tradeoff. I knew we planned to anonymize the data and wouldn't match the list of visited sites to a user's identity. Still, it felt creepy to me and I figured I wouldn't be the only one. So how to inform users without scaring them off?

If we said that turning on advanced features showed Google every web page a user visited, many people would never download any version of the software. That would be a disaster. The Toolbar was a secret weapon in our war against Microsoft. By embedding the Toolbar in the browser, Google opened another front in the battle for unfiltered access to users. Bill Gates wanted complete control over the PC experience, and rumors abounded that the next version of Windows would incorporate a search box right on the desktop. We needed to make sure Google's search box didn't become an obsolete relic. To do that, we would turn Microsoft's strength to our advantage and pit one group within the Redmond-based company against another.

If the Google toolbar became so popular that people downloaded Internet Explorer just to run it, then the browser gang within Microsoft might defend us when the Windows mafia inevitably tried to snuff us out. In the meantime, we would piggyback on the world's most popular web browser to gain millions of users.

The wording users saw when downloading the Google toolbar had to be subtle and assuage their concerns while downplaying the risks. We could just bury it in the EULA (end user licensing agreement) in tiny type and no one would be the wiser. Almost nobody reads the legalese terms of use before installing software. Knowing that, I considered long and hard and came up with an appropriately nuanced response—the two lines I referred to earlier.