Authors: Douglas Edwards
"Found a problem," I thought. "Now fix it."
I set up an interdepartmental meeting to address the need for project prioritization and resource allocation. The marketing and business development departments, the two mighty legs upon which the company's future stood, would collaborate to construct a stable platform for growth. My office mate Aydin wanted in. So did Shari and Susan and a couple of others on the business side.
At our first meeting, we brainstormed process structures. We analyzed the competition, our overall market position, and where the industry as a whole was headed. It was a good meeting, a productive meeting, and at its end we were prepared to offer the strategic leadership Google lacked. How much market share could our process-fed Google gain? The world will never know.
Once Larry and Sergey heard that we intended to dictate product plans to engineering, they threw a monkey wrench at us. That wrench was Salar. He didn't kill our committee directly, but it died just the same. Salar said that Larry was fine with answering a list of questions to clarify our corporate strategy, but that product development would remain ad hoc until further notice.
Something must have gotten through, though, because the next day Sergey issued a company-wide manifesto of sorts, listing our top three priorities: "product excellence, user acquisition, and revenue." It wasn't much of an action plan. Nor did it answer broader questions like "How will we ensure user loyalty?" and "What is our market-expansion strategy?"
I stopped asking those questions eventually, as I became convinced our founders intended to pick a path to the future based on gut instinct, then haul ass through a fire swamp of competition with the entire company riding on their backs. Such a strategy required them to hold a degree of certainty in their own abilities.
"They had so much self-confidence that Sergey was convinced he personally could find a cure for AIDS," engineer Chad Lester marveled. And why not? With twenty-five million dollars in the bank, the founders' wild hubris was free to roam the plains. Life gives you few chances to make decisions as if no one else's opinion matters. Google offered just such an opportunity. Venture capitalists John Doerr of Kleiner Perkins and Mike Moritz of Sequoia Capital sat on Google's board,
*
but the board didn't have control—that belonged to Larry and Sergey. All the board could do was try to guide them.
Larry and Sergey had anointed Cindy VP of corporate marketing and put her in charge of public relations and promotion, but not the development of products. The board wanted a different leader to build that organization—someone with technical savvy, but not an engineer. When engineers drive the gravy train, sometimes they focus on how fast they can go rather than where they're headed.
Larry and Sergey reluctantly agreed to take a look around, though no traditional consumer-marketing person could possibly impress them. The right candidate would have to be able to communicate with coders, execute quickly, and be very, very smart. And smell nice. Sergey once rejected an applicant in part because "I thought he had kind of a bad body odor."
Jonathan Rosenberg, an executive at Excite@Home, was very, very smart, and the board strongly hinted that he would be the best fit for Google's needs. Jonathan had other ideas and decided to stay at Excite for two more years before claiming the role of Google's vice president of product management. He left behind more than a résumé, though. He convinced Larry and Sergey that there might be a role for a product-management group if, and only if, it didn't usurp the divinely ordained primacy of engineering. God forbid that Google become a marketing-driven company. In marketing-driven companies, researchers identified customer needs and then product managers (PMs)
*
directed engineers to create products to fill those gaps. I had been taught that was a good thing to do.
"Let's do a gap analysis," I used to say at the
Merc.
"What's the unmet need? Where's the market opportunity? How much share can we gain?" Engineers hate that kind of thinking.
If you're an engineer with a brilliant idea, seeing it dumbed down or abandoned because it doesn't test well is like watching a bully pull the wings off a butterfly. The right thing to do is build it regardless, to prove that you can and because building cool things is—well, you end up with cool things.
Pure-hearted geeks flee the hellish realm of product-driven companies, where soul-sucking suits shuffle after profits instead of perfection and the boss doesn't understand any technology more complicated than a binder clip. There, product management rules, pandering to a public that hasn't a clue about what it really wants, while enslaved engineers make tiny improvements to cruft—crappy products wrapped in promotion and paraded before consumers like lipsticked pigs.
Larry and Sergey would not allow that at Google. Their company would appreciate the inherent beauty of inspired design in breakthrough technologies like a spell checker that recognized a hundred variations of the name "Britney Spears" or a calculator that could convert the speed of light into furlongs per fortnight. For that to happen, a Google PM would need to be someone smart enough to understand engineers but not intimidate or subordinate them. Someone the founders would feel comfortable having as a direct report. Someone who would stand up to them when need be, but who wouldn't waste their time with trivial concerns.
"I was the first person who wasn't an engineer working with engineering to define the product," Salar recalls. "We decided to call that product management."
Salar's transformation into a one-person department complemented the other process piece that fell into place in January 2000.
"The Internet is under-hyped," John Doerr had declared early in the dot-com boom. Silicon Valley listened, because Doerr's early investments through Kleiner Perkins in winners like Sun, Netscape, Compaq, and Amazon proved his uncanny acumen. Now, as one of our board members, he prescribed the best practices from his portfolio companies to beef up Google's anemic process management and tighten our flabby decision-making.
"We had a marvelous meeting this morning with John Doerr," Cindy said one day. "I was terrifically impressed with the thinking he's trying to instill in our guys."
Doerr's corporate growth regimen comprised a system for setting goals and measuring progress that he called Objectives and Key Results or OKRs. It was far more methodical than the ad hoc list of nebulous goals Sergey had unleashed on us just a week before.
"Objectives," Doerr instructed Larry and Sergey, "should be significant and communicate action. They state what you want to accomplish, while key results detail how you will accomplish those goals." Key results, therefore, should be aggressive, measurable, and time-specific. Doerr warned the founders not to overdo it: five objectives with four key results each should be sufficient.
Larry and Sergey's initial objectives included "move toward market leadership," "best search user experience," "meet or exceed revenue plan," and "improve internal organization." Those broad categories remained for many quarters, but the key results—the steps we would take to achieve them—kept changing, from "distribution deals adding half a million searches per day" to the "launch of Google in ten languages" and "CEO candidate selected."
If I wanted to see webmaster Karen White's priorities, or Cindy's or Larry's, I had only to call up the company phone list on our intranet (nicknamed "MOMA" by Craig Silverstein after the Museum of Modern Art) and click their names. It helped, because in the final days of the quarter everyone rushed to check things off their lists. If someone else's OKRs were contingent on me, I wanted to be forewarned they'd be hunting me down, and I wanted to know that the people whose help I would need to complete my own OKRs were aware of them.
In a company filled with overachievers, I assumed everyone would accomplish all their OKRs. No. It turned out that that would indicate failure. The ideal success rate was seventy percent, which showed we were stretching ourselves. Larry and Sergey assured us that missing OKRs wouldn't factor into performance reviews, because if they did we would take too few risks.
OKRs forced us to reevaluate priorities at least four times a year. If our industry changed or our corporate goals did, we had an obligation to bring out our near-dead initiatives and load them in the tumbrel. At the
Merc,
it had been nearly impossible to stop projects once they started. I had sat on a committee that decided our neighborhood news sections bled cash and should be put down, but for months no one found the will to pull the trigger. Not so at Google. Larry and Sergey expected instantaneous results, and products lived or died based on data, not sentiment.
I had mixed feelings about the efficacy of OKRs as a gauge of progress and individual accomplishment. I could support product launches on my own, because that just entailed writing copy for the website or drafting promotion plans, but I struggled to complete projects requiring engineering resources, like revamping our online store, or refitting a van as a portable wireless-access point (a Larry idea), or launching an affiliate program to pay webmasters to send traffic to our site (also a Larry idea).
As I sat in Cindy's mid-quarter update meetings, I studied my colleagues' reports and the traffic-light-colored circles next to each of their OKRs. Green meant completed or soon to be. Red meant major obstacles or a dead end. PRs' slides always looked like a well-watered fairway. My tech-dependent goals resembled a jaundice victim working on a sunburn. Yet the end result did feel like progress. After the OKRs had been in place a while, I emailed a friend I'd left behind at the
Merc:
"We're suddenly a much more focused company, which makes me think we could actually be on to something here."
"There are contentious issues," I emailed another friend, "but there's no animosity built up around them. Nobody blind-copies anyone and there's not a culture of blame fixation. Corporate politics will undoubtedly come with scale, but for now, folks are too focused on getting things done to cast aspersions."
Egos, yes, we had those in abundance, but we lacked Napoleons building personal empires. Skirmishes sprang from convictions, not power lust, and I saw nothing like the bloody trench warfare I had witnessed at other companies, where heads on pikes decorated fortified domains. Engineering was engineering. Marketing was marketing. It was clear on which side of that line you stood. Unfortunately, my prescience about politics developing with scale didn't take into account the rate of Google's growth. Instead of the years I envisioned, it was just a matter of weeks before the first border guards and checkpoints appeared.
Cindy took the occasion of the new millennium to formalize areas of responsibility within our department, dividing marketing into two groups—one under Shari's direction and one under mine. But it was Salar who erected the first fences along interdepartmental lines.
My effort to instill organizational clarity had been smothered in its cradle, but Salar recognized that uncoordinated decision-making could lead to dissonance, especially with user interface (UI) issues. So he split the baby. Marissa Mayer, who was filling in part-time as a human-computer interactions (HCI) engineer, would develop proposals for the look and feel of the results pages, while Karen and I would oversee content and design of all the other pages on
Google.com
. Marissa and I could both present homepage modifications, and Larry would make final design decisions on the basis of testing data and his own Larry worldview. The allocation made sense in theory, but as the Catholic Church discovered in 1378, two popes don't make you twice as infallible.
A few days later, Marissa set up a "UI-team" email list for the group that would manicure Google's appearance and shape debate on usability issues. On the list were Karen, Marissa and me, Salar, Shari, and engineers Bay Chang, Krishna Bharat, and Jen McGrath. We would meet weekly to hash out the text describing new services or the color of visited links or the size of the font for the link to our help page. That way Larry could maintain a thirty-thousand-foot view and not get lost in a sandstorm of granular detail.
UI team meetings dragged on for days—or seemed to—in the sterile, airless conference room in which we met. The lights would go down, the projector would go on, and the drone of voices and page design variations would commence. Of course I had opinions about design (everyone did), but after ten minutes debating the exact shade of blue for a three-pixel line on the fourth page of results, I'd be doodling bored skeletons sitting at keyboards in endless cube farms. Human-computer interaction, indeed.
"The table of contents has a hardwired width of six fifty," Bay might say to kick things off. "How about creating another column in the middle, and making it a fixed width?"
"The headings seem to be set farther right on the page than the text," Marissa might jovially rejoin. "Can we fix this with margins or reducing cell padding or cell spacing?"
"You missed a comma after the second item in the series," I'd toss out, just to keep the rollicking good times rolling and oxygen flowing to my frontal lobe.
Mockup wars broke out constantly and quickly escalated, with design ideas expressed as HTML sketches. The projector linked to our laptops fired dozens or even hundreds of layouts at the wall as we fought our way toward the most user-intuitive implementation.
I was handicapped by my degree in English—a burden I alone struggled to bear. It wasn't that I didn't understand basic HTML tags (as a concept, anyway), but my limited knowledge left me unilaterally disarmed. I had brought a cap gun to an Uzi drive-by. As my colleagues furiously revised code on the fly and sent iterations ricocheting, I resorted to paper and color markers to render my ideas. That patronizing chuckle you bestow on toddlers for stick-figure portraits? It became the soundtrack of my presentations.