Authors: Douglas Edwards
"The founders were okay with a loose shag bag of marketing folks who were at the ready to execute on their whims," Cindy told me, "but a real marketing department with a VP, proper organization, funding, and a strategy was not a priority." As a result, our world was without form and confusion was on the faces of those who dwelled within it.
"Who's working on our letterhead?" I asked Cindy. "Who handles sponsorship requests?" Were these areas that fell into my domain? I was seeking more than organizational clarity. I wanted to be sure that there was some substance to my job, something I could cling to when people asked, as they inevitably would, "What exactly do you do here?"
"No structure, foundation, or control," is how Heather Cairns, Google's HR lead at the time, remembers the company's early days. "Even if someone had a manager, that manager was inexperienced and provided no leadership. People weren't used to authority and wouldn't adhere to it—it was a completely unmanaged workforce that was bouncing off the walls like a tornado. I didn't pretend to have any control over it ... I just went home at night to drink, thinking, 'We're gonna crash and burn.'"
"Our site is kind of a mess," Cindy said to Karen and me my second week on the job. "Can you work up some guidelines to clean it up?"
We had no rules governing what went on
Google.com
. Something new launched, it got mentioned on the homepage. We won an award, that went up too. Our other pages were equally devoid of planning and design. There were job listings, some help content, contact information, and brief profiles of the executive team. As with everything else at the company, our user interface (UI: the look and feel of our website) operated on the principle that we should minimize the time it took for users to find what they wanted.
Unlike Yahoo.
Yahoo's homepage had links to apparel, computers, DVDs, travel, TV listings, weather, games, yellow pages, stock quotes, and chat. It got busier with every passing day. The most prominent feature on the page was Yahoo's hand-built directory with its fourteen major categories from Arts & Humanities to Society & Culture, beneath which were links to all known points in the Dewey decimal system. Buried in the middle of all the text links was a search box powered by our nemesis Inktomi.
Inktomi hadn't always owned that space. AltaVista had provided search to Yahoo until 1998, but they made the fatal mistake of building their own portal site and stealing users from their customer (competing with your own distributor is known as "channel conflict"). Inktomi had no "consumer-facing" search site,
*
so they weren't Yahoo's competitors, which also gave them a clear shot at Microsoft's MSN network and America Online (AOL). Inktomi locked those customers up as well, completing their trifecta of high-traffic Internet sites and ensuring that the state of search across the web was commoditized. You could get any flavor of search you wanted, as long as it was Inktomi. They owned the search market and sat on it as fat and happy as the enormous customers they served.
Other portals wanted a piece of Yahoo's traffic: Excite, Lycos, and Disney's
Go.com
. And other search companies, like AlltheWeb, Teoma, and HotBot, fought alongside Google for the crumbs falling from Inktomi's table. While Wall Street focused on the portal wars, the struggle for search domination wasn't of much interest to anyone but a handful of analysts. There was no money in it. Well, not much money.
In February 1998, a small Pasadena company named GoTo started auctioning placement in search results they bought from other providers. Six months later, they claimed to have more than a thousand paying customers. According to GoTo, you didn't need fancy algorithms to determine relevance, just the invisible hand of the free market. Any company bidding for placement at the top of the results must be a good match for the term being searched. At Google, we found that concept ridiculous. Bidding-based ranking was clearly inferior to results based on an algorithm. Bidding was driven by imprecise humans. Humans bad. Math good. We knew about GoTo, but we discounted their "non-technological" approach. That proved to be unwise. We gave them a head start, and for the next four years we would fight them for supremacy in the online advertising market.
Codifying some UI guidelines
†
would be a good beginning project, I thought. How tough could it be to come up with some design rules for a page containing nothing but a search box, a hundred or so text characters, and some corporate shovelware behind it?
*
Besides, working with Karen was like drawing the right lab partner at school. Even if I screwed up, Karen wouldn't let us fail. We knocked a proposal together in less than a week.
Google was fast, accurate, and easy to use—that's what our users told us. Sergey wanted our site to be "fun" as well. Yeah, great, it's fine to have fun occasionally, but Karen and I agreed that whimsical elements shouldn't get in the way of users getting things done. We explicitly stated the obvious: "The personality of the site should under no circumstances interfere with the speed of results delivery, the accuracy of the results, or the ease in using the search functionality." An axiom we would unintentionally prove soon enough.
The rest of the proposal involved other obvious points—tweaks to what existed rather than a major overhaul—like adding decorative graphics to our corporate section. That didn't fly.
"Yahoo doesn't use images beyond the homepage," Larry reminded us, "and they have millions of users. Images take time to transfer across the Internet and slow things down." Larry and Sergey rolled on the floor rapturously speaking in tongues when we shaved a nanosecond off the time it took a page to load. Or to read. "I want all the content of the About Google section on one page," Larry said. "It would be faster to scroll up and down one page than to click from page to page and wait for it to load."
"But no one's going to scroll down a hundred pages," I said, not sure if he was joking. He wasn't, but we managed to argue him out of it. Other suggestions fell by the wayside, like a help link, a tagline, and an embarrassingly naive idea Sergey had to change the homepage logo every day to build user interest. Professional branding people were in the house now, and we would never abide such amateur antics. Overall, Larry and Sergey gave a thumbs-up, proclaiming our guidelines "sensible." High praise indeed.
I let out a sigh of relief. Now I got it.
This
was what I'd been hired to do. If I hadn't knocked my first project out of the park, at least I'd hit a solid double. Everyone seemed reasonable and receptive to new ideas, and the feedback made sense. I hadn't done anything terribly unconventional, yet my ideas had been accepted.
"Yep," I thought, "it's all going to work out just fine."
"As of last night, Google's result font has become sans-serif," engineer Marissa Mayer announced to the company at large. "We tested the change and Larry and I reviewed it with some other engineers who were here and offered opinions about it."
I had seen Marissa's name on a note Sergey forwarded to the new marketing group a couple of days earlier. She had suggested we replace our temporary slogan—"Best Search Engine in the World. Promise"—with one Urs had come up with: "The Little Engine That Could." I didn't particularly like either line, though Marissa had constructed a detailed rationale for associating Google with the "scrappy," "determined," and ultimately "triumphant" children's book character. Besides, she pointed out, look at the importance of Ask Jeeves's tagline to their valuation.
Shari thanked Marissa and explained that we didn't have a slogan, just a phrase that was printed on some cards until we could properly research our brand character. Marketing had it under control.
Marissa, like Susan, was an old-timer who had come over from the Palo Alto office. Before that she had been a Stanford student. When I finally met her, I was struck by the intensity and scope of her interests. If everyone else at Google was a hundred-watt bulb illuminating a single corner of the company, Marissa was a flashing neon sign, casting light and shadow in all colors across the entire Googleplex. Trying to keep up with her could induce seizures. Her primary role was as a software engineer, but she was temporarily working on UI design. She had uncovered research indicating that sans-serif fonts were easier to read, so she and engineer Craig Silverstein had decided to change the results font to Verdana. It didn't happen because Karen and I had suggested a move to sans serif in our guidelines. In fact, Marissa may not have even seen our proposal.
"It would have been nice to know that engineering was already working on UI," I griped to Karen. It undermined what we had done and made me question our internal communication. No one else seemed terribly upset, so I dismissed the faint alarm bells and chalked it up to the newness of the marketing organization. It was great that our engineers made improvements. I just wanted to be sure we wouldn't be caught by surprise when they launched them.
I was equally sanguine about my first glimpse of the way the two departments approached problems. We in marketing wrote proposals, made suggestions, and looked for broad formal approval before moving ahead one step at a time. Our engineers made quick data-based decisions and implemented them. If data supported a particular option, they rationalized, it was the right choice to make. Data didn't lie. If the numbers said changing
A
to
B
would improve product
X,
why not do it now? This mindset drove much of the urgency at Google. Engineers knew how to make things better, and every hour, every minute, every second we delayed improvements, users had to endure sub-optimal interactions with our site. That was unacceptable.
I would discover, however, that data does lie. Sometimes the method of collecting it is flawed, sometimes it's misinterpreted, and sometimes it provides only part of the answer. Take the change to Verdana. While the new font looked great on certain PCs and certain browsers, it rendered horribly on others, most notably those used by America Online customers—essentially making Google unusable for millions of people. Marissa and the engineers hadn't checked it on that platform. That problem was quickly corrected, but even though I supported the move to a sans-serif font, I was left with a healthy skepticism about the ability of numbers to tell the whole story, a loss of faith that I alone appeared to experience within Google's kabala of data-based divinity.
TGIF was the four-thirty Friday afternoon meeting at which new Googlers (or "Nooglers," as I would dub them) introduced themselves and the founders shared the important news of the week. When I arrived for my debut, the entire company was gathered in the hallway outside Larry and Sergey's office, leaning against walls, sitting on rubber balls, and sprawling on the floor. Urs's dog Yoshka sniffed at Susan's newborn son, who was asleep in his car seat in front of a wall of Sony monitors stacked in their shipping cases.
Larry, dressed in black slacks and a long-sleeved, dark blue shirt buttoned to the collar, stepped to the front of the crowd. He carried himself as if his armature ran on a woefully underpowered processor—stiff, awkward moves and self-conscious grins—as if he reminded himself to smile and then manually executed the command set to make it happen. Lift the edges of your mouth. Stretch your lips back. Crinkle your eyes. It made him seem painfully self-conscious. I found myself rooting for him to compile successfully, though sometimes he got stuck halfway through a sequence and stood leaning to one side with a half-grin frozen on his face. Sergey was more fluid, athletic, acrobatic. Bouncy, even. He laughed easily and seemed to always have an eye out for a railing he could vault or a rafter beam he could pull himself up on.
When my turn came to talk about what made me interesting, I hesitated. "Well ..., I lived in Siberia for six months and, uh, I like to doodle in meetings."
Everyone politely applauded, but I wondered again what exactly I had gotten myself into. I had made a career commitment, and my future success rested in large part on the gray matter behind these happy, shiny faces. I prayed that they had a better idea of what they were doing than I did.
Larry ran through a presentation he and Sergey had made to the board of directors and rattled off a few announcements. Then we broke for beer and cake in honor of those with birthdays that week. For better or worse, I was officially a Googler. Gabba gabba hey.
TGIF was miles from the formality of meetings at my former places of employment, where the emphasis lay on keeping key financial information out of the hands of staffers who might wield it in future contract negotiations. Here we had seen the exact same slides the top executives had presented to our board. I couldn't believe that all this information was thrown out to every employee, as if at Google we occupied an alternate universe of kibbutz-like communalism. Despite the transparency of our accounting, I was troubled by my lack of insight into our business model and our competitive environment. I knew exactly how we were doing financially, but I didn't have a clue about what was going on.
I
WOULD NEVER UNDERSTAND
this new world until I was fully grounded in search technology. Craig Silverstein volunteered to school me and a handful of other nontechnical staff with a lunchtime talk he called "Google 7A: Search for non-majors." Craig, a former Stanford grad student with an impish grin and a sense of humor dry to the point of sere, liked to say he was employee 1.5, officially added to the payroll between Larry and Sergey when the company incorporated. Craig also provided our daily bread, wandering the halls on random afternoons wearing a beatific smile and announcing in an interrogatory falsetto, "Breaaddd?" Googlers would scramble from their cubicles to partake of the air-filled, nutrition-free loaf he had just taken from Google's breadmaking machine.