This book is interactive, supplemented by a daily video about Google and its management practices. www.googlemanagement.tv
This is the first book worldwide on the subject by a management professional.
Table of Contents
I. AN UNORTHODOX CORPORATE SAGA
II. A FORMULA 1 ENGINE
III. “PUT USERS FIRST. THE REST WILL FOLLOW… ”
IV. CHALLENGES AND RISKS
V. A REAL MODEL
Beginning shortly after the First World War, General Motors and Ford created the large modern corporation, with its financial and statistical controls, mass production, standardization, assembly lines organized according to scientific principles', and autonomous divisions. By the 1960s, mass distribution had created the consumer society, with its system of credit sales, self-service stores, media networks, mass media ad campaigns, brands and international products. During the 1980s, Toyota became the archetype for an industrial company focused on the quality of its products. Today, Google is the company that is reinventing management methods: how we work, how organizations are controlled, how people are supervised. Google does this within the specialized context of the Internet, an economy of distributed intelligence born at the beginning of the 1990s in Silicon Valley. These circumstances give the company an aura of carefree dynamism unlike any of its predecessors. But the massages, swimming pools, volleyball courts and hosted lunches with their workforce don’t make Larry Page and Sergey Brin, the Google co-founders, any less formidable than Henry Ford or Taiichi Ohno, creator of Toyota’s production system.
If Google can be considered an enterprise archetype like Ford or Toyota, it is because its management made simultaneous innovations in several fields: human resources, production, customer relations, and most of all control of its production operations. Google did this through its own innovation, meanwhile borrowing from other technology companies; but also -- which is more unusual -- by collaborating with the founders’ university. I have addressed these affiliations wherever possible, and come to the conclusion that Google is the first example where these strategies were identified and utilized in a systematic way. The rapid growth of the company, the personalities of its founders, their vision, their scientific culture, their obsessions, and the expertise surrounding them all contributed to the construction of this unique business model.
The primary purpose of this book is to provide keys to understanding how and why it worked. I mostly discuss Google, but sometimes also Amazon and other companies that have adopted this progressive style of management.
After an historical overview of the company, Part I deciphers the economic model built by Google’s leaders.
Part II, the largest section of the book, discusses in detail the management methods they invented, which are far removed from the best practices taught at top business schools. Each area of innovation -- human resources management, organization, innovation management and production -- is compared with the norm and the differences discussed.
Part III continues with an analysis of the environment in which the company developed. Google’s success in attracting users was largely due to better understanding of its user communities than the competition. The reader will see how the automation of commerce profoundly changed relationships with users by giving them a role they had never had before. Google’s obsession with putting users first has greatly contributed to the success of the company and its growth, so its breakthroughs in management are also breakthroughs in customer relations.
The last two chapters, Part IV, speculate about the limits of this model, and discuss the new challenges it has created.
The last part of the book summarizes these analyses and proposes a series of recommendations for anyone who wants to use this model as a starting point.
The management innovations at Google are all the more interesting as the company explores and creates new economic relationships. This book is addressed to anyone who wants to understand more about this new paradigm and adapt it to his own professional environment.
A glossary follows for anyone who is unfamiliar with the vocabulary and issues involved. .
The history of Google reads like a fairy tale about young people who dream of becoming Masters of the World.
It began at Stanford, one of the most prestigious American universities, during a spring 1995 tour of the university for prospective new students. This included a day trip to San Francisco that was led by Sergey Brin, a student of Russian heritage who was considered a star mathematician. One member of his group was another top student, Larry Page. Both quickly discovered that they had mutual interests. They spend the afternoon in animated discussion, including some heated arguments. Such is the legend of the first meeting held by founders of Google.
That fall, Larry transferred to Stanford. Almost immediately, the two young men (with less than forty years of age between them) began to work together. At the time, it was almost inevitable that technology students would think about starting a company. The Internet was in full boom. Netscape had just gone public and made headlines in all the big news media. Several of their fellow students had already quit school to go into business, but not these two. . Both wanted to continue their studies and get doctorate degrees. Page was interested in the mathematical challenges posed by the Web, especially its graphical organization. He wrote a paper with Terry Winograd, one of the fathers of the artificial intelligence, well known for his work on natural language programming for robots.
Inventing a way to classify pages
As the subject of their theses, our two youthful heroes chose the classification of results obtained by using an Internet search engine. The subject might sound dry and esoteric on the surface, but it was the central concern of everybody interested in these issues. Searching for information stored on electronic media is nothing new. Since the beginning of the 1970s, engineers had collaborated with archivists to develop programs designed to provide quick access to scientific information stored in databases containing tens of thousands of article and book titles. They followed two parallel tracks:
• Some automated the work of archivists and librarians, who index documents, describe them with key words and then compile them in a large dictionary called a thesaurus*. With these programs, like Questel in France [other example needed], the user (usually an expert) can perform complex research with Boolean operators* (and, or, not and so on).
The elegant simplicity of the latter approach interested data processing specialists: there was no need for an archivist to query databases manually. Anyone can type in keywords; so the steps of preparing and indexing documents were eliminated. Just digitize them the way they are and store them in a database.
Language being what it is, however, this method had its own disadvantages. If you tried to introduce synonyms or contextual meanings, you couldn’t avoid creating more junk*.As long as the databases remained specialized within a limited field for use by professionals (like the legal documents for attorneys stored in Lexis), the risk was limited. But using these programs on the Web was another matter. A user could find plenty of documents that contained the words he searched for, but with to many worthless nonsense results. The larger the Web grew, the more pages that were assembled and indexed, the more the relevance of search results was degraded. Brin and Page said in one of their first papers, “‘Junk results’* often wash out any results that a user is interested in. In fact, as of November 1997, only one of the top four commercial search engines finds itself (returns its own search page in response to its name in the top ten results).”
To counter this defect, which threatened their very existence of the Web, early search engines hesitated between two solutions. Some limited the sizes of their databases. Why keep adding pages, since more quantity means worse results? Others, like Yahoo!, began by compiling an index. They created a major system to classify sites by topic. A Webmaster who wanted to register a site was asked to provide key words to indicate the subject matter. Then experts called “ontologists”* would check the relevance of these descriptions. This second method was certainly preferable to the first. If you ask the search engine to find the word “horse” in the zoology category, you’ll get sites about animals; if you use the sports category, you’ll find pages about horsemanship. The art category will lead you to sites about the artist Cheval [in French] or equestrian paintings; in the cooking category, you’ll get recipes for horse meat, and so on. But this approach was also far more expensive. Yahoo! had to hire hundreds of people to analyze web pages. Neither of these solutions satisfied Brin and Page. They were looking for an automatic way to classify pages according to their relevance. They were not alone in trying to solve this problem.
• Some of Yahoo!’s competitors that also employed ontologists, as well as a group of IBM developers, were trying to automate the construction of branching categories. This work still continues today.
The latter technique is obviously a great improvement over former methods, but it still has disadvantages. For one thing, it is not completely reliable. (With browsers that can open several pages at once, a surfer might stay on a site for a long time without reading it, so the data sent to the server is irrelevant.) This method also makes it easy to cheat the system. If I want to push up my site ranking, all I have to do is install a small robot program that clicks on the site, stays there a few minutes, leaves and returns.
Like the engineers at DirectHit, Brin and Page assumed that the best recommendation factors for a site were guaranteed quality and relevance. But instead of using the number of hits as the measure, they took into account how things work in real world research. To evaluate the quality of an author, an idea or a concept, researchers look for the number of references to an article in respected publications. So a scientific article will be judged by how many times it is quoted or cited in other articles. On Internet pages, links are somewhat equivalent to citations. If I put a link in my text leading to another page, there’s a good chance I consider it important, or at least relevant. By counting the number of links between various pages, you can classify them to get more reliable results.
Not all quotations are equally valuable, nor do links all have the same importance. A quotation of a Nobel Prize winner published in a prestigious journal has more value than a quote from a student published in the school paper at a Podunk junior college. Similarly, links coming from another page that is frequently “quoted” by other sties (with many incoming links) have more weight than those from pages with fewer links. More subtleties were added, such as the distance between the different words specified in the search string, and a weighting system that gives higher value to links from sites with many incoming links, but few outgoing ones. This system made it possible to improve the quality of searches without human intervention. As obvious as it may appear to us today, the method uses highly complex mathematics and involves the integration of several classes of problems. This is why initial support came mainly from the scientific community. In fact, Google’s initial success was in the areas of programming theory and scientific data processing. The method is useful in programming to shorten development time, and also to sociology theorists who wanted to model social phenomena such as reputation and cultural influence. So this method of classifying Internet pages on the Net is more than just a technically easy way to limit junk* in a database. Because of its novelty, it qualifies as an invention. That is why it interested scientific researchers, especially mathematicians.
This is no minor detail. As we will see it throughout the company’s history, one of Google’s main strengths is its ability to maintain relationships with the academic community. The quality of these relationships stems from the personalities of the company’s founders and their contacts with high level researchers like Terry Winograd, now a Google consultant after having been their college professor. But it is also due to Google’s work in areas of inquiry that interest researchers. This enables the company to transform questions asked by its engineers into problems that mathematicians will be eager to solve.
Brin and Page’s program for ranking pages was initially called BackRub, then Pagerank (referring both to web pages and Larry’s surname). Made available to students and professors on the Stanford network in 1997, it quickly becomes the search engine of choice, at least partly because it was the one that popped up when a browser was opened.
Soon, Larry Page’s dorm room at Stanford was full of computers. Since money was tight, the two accomplices scavenged components, memory chips and surplus computers wherever they could. This taught them a lesson that would prove very useful later on: You can build a powerful machine by combining separate, weaker machines. Their activity caught the attention of a of a university safety inspector, so they had to find a new location. Only then, with reservations, did they start their company in a garage -- actually an apartment with garage space included -- that they rented from a friend of Brin’s girlfriend.
No doubt they would have preferred to sell their technology. They approached AltaVista; a pioneer search engine founded by Louis Monier, which had just been bought by Digital, the minicomputer company. The engineers of AltaVista were interested, but their managers were suspicious of anything that wasn’t developed in-house. They fared no better with other search engine companies they approached. It was obvious that their technology had a major defect in the eyes of advertising salespeople: by producing relevant results so quickly, it reduced the length of time a user would stay on the search page. The longer an average visitor stayed there, the more they could charge advertisers -- a lesson they had learned from the major media.
Without partners, Brin and Page needed to find funding. They began making contacts with “business angels.” These tend to be engineers who made fortune in technology, and advance seed money to young startup companies with interesting projects. Then they met with venture capitalists, financiers who invest in companies hoping to make them grow big enough to capture a significant share of a market. At this point, Brin and Page demonstrated both their taste for independence and their determination. Venture capitalists are known for predatory behavior, which earned them the nickname, “vulture capitalists.” Knowing this, the two put up a hard fight to maintain their independence. Only after several months of negotiation did two VC firms put up equal amounts to capitalize the company for $25 million dollars. The level of control Brin and Page were able to maintain was a first.
This financing enabled them to develop software and purchase the computers they needed (300 by June 1999 which has increased to several hundred million dollars’ worth today). But with the funding came two requirements. They would need to go public one day, meaning the company had to generate income. Google had no way to make money, since it gave its results away free to users. They also needed to find a manager, someone more seasoned and “rational,” who knew the languages of finance and marketing. It took some time to satisfy both requirements. Meanwhile, Google’s reputation grew rapidly, thanks to highly favorable press. By1998, stories had appeared in 16 publications, including journals in Russia and France. The same year, PC magazine ranked Google among its 100 noteworthy sites, making it known it thousands of surfers who quickly numbered in the tens of thousands as word-of-mouth spread. However, giving away free search results didn’t generate income. Another important level was reached in 2002, when America Online (AOL), one of the very first service providers, chose Google as its search engine for more than30 million users. All of this gave the young company a high level of visibility, but nothing to put in the bank.
How to make money
Brit and Page initially tried to sell services to companies that wanted to find information buried in the mountains of data stored in their computers. With “data mining,” every day (late at night), the computers of Google would index a company’s data so they could search through it the next morning. The two students cum contractors managed to convince several firms to trust them with corporate data. But this market, narrow to begin with, was shrinking as companies became more reluctant to entrust data to outsiders. They had to find another way, and that way had to be advertising. All search engines were selling space. Page and Brin didn’t like the idea, and resisted for a long time. They expressed their disdain for advertising in one their first articles in 1998, and the arguments are no less relevant today. But they had to find a way to make money. As a starting point, they chose the model originated in 1997 by GoTo.com (later to become Overture, and today owned by Yahoo!), a competitor that was generating profits from its search engine.
The solution rests on three concepts that they would refine and modify.
AdWords, Google’s current source of revenue, was born -- and with it, a formidable moneymaking machine. Before long, the most optimistic forecasts were exceeded, Google was making money. Big money.
Since Brin and Page had taken venture capital, a public stock offering was inevitable. They had, said the analysts, waited a long time, and now had a find a solution to the constraints and pressure they faced. Certainly, from their point of view, an IPO had many disadvantages. First, it would limit their freedom. Shareholders make demands that can be expressed very simply: by voting with their feet and selling their stock. A public offering would also force Google to reveal financial information they had withheld until now for a good reason: as long as nobody knew how profitable the business was, they were insulated from competition. Further, a public offering would draw the attention of fearsome predators like Bill Gates, who had shown a few years ago how he could break his biggest competitors. Just recall the mishaps of Netscape when stockholders walked and destroyed more than two-thirds of its value within in a few months. Going public would also require them to structure a more conventional company. These are some important reasons they hesitated so long before making a commitment. But when they finally did, in 2004, they took an unusual route. When a company goes public in the US, it usually turns over the job to investment bankers who have all the rules down pat and know how to avoid troubles with the SEC regulatory cops. They also know how to enrich themselves and their cronies from these transactions.
The mechanism is relatively simple: estimate a weak opening price for the stock by taking a survey of “selected” potential investors. That way, those who buy shares at the beginning will be able to sell their stock at a higher price if the price rises. In order for this mechanism to be fully effective, shares need to be reserved at the deflated “opening price” for friends, who then bid up the stock by trading during the first few days it is on the market. It’s mandatory that the majority of potential investors have no access to the stock beforehand, so they will be anxious to buy. The few investment bankers who specialize in these manipulations have became masters at the art of anticipation, touting stocks during advisory meetings held for likely investors and their financial advisors. These meetings are part of the services investment banks sell to their customers. Of course, the fees are very expensive. .
Larry Page, Sergey Brin and Eric Schmidt, the manager they had recruited at the behest of their investors, wanted nothing to do with that. They researched and discovered a system to avoid the artificial margins: a “Dutch Auction” with sealed bids to set the price of the stock. Bids were accepted over the Internet, as well as by fax and phone.
This unusual, elaborate system was not a first-time innovation. The principles were formulated by William Vickrey, an economist who received the Nobel Prize for the concept in 1996. For some months, the method had been promoted by William Hambrecht, a Silicon s well-known Silicon Valley financier. In 1999, he sold the investment bank he had founded for Chase Manhattan, which had contributed to the financing of several major companies including Apple, Genentech and Sybase. At the end of the 1990s, he developed a method of marketing companies called OpenIPO, a transparent allusion to Open Source* software. His first client was a vineyard, Ravenswood Winery. Since then, he had helped take several small companies public using the system, the best known being Salon, the Internet magazine. These were wonderful companies, but modest in size -- nowhere near comparable to Google.
The remarkable part of this bidding system is that it encourages prudent buyers. . The higher you bid, the better the chance that you will get the number of shares you want. You will probably pay less than you offered to pay, since the ultimate price precipitates all the share sales. In this way, it contradicts standard auction practices, since it is advantageous for the buyer to announce the price he is willing to pay upfront. He is confident, since he knows he won’t pay more than necessary. Bypassing the inner circle inspired an uproar in the investment community, which complained bitterly about the arrogance of Google’s two young founders. The trade press was skeptical, and then lambasted them when Playboy magazine decided to publish an interview with the Google leaders that had been recorded well before they decided to enter the stock market. Companies that make public stock offerings are required by regulation in the US to maintain a “quiet period” during which they cannot make public statements. This publication by a third party amounted to an involuntary violation of their obligation to remain silent, but they were blamed in many articles.
Besides these problems, Google was sued by Overture, alleging patent infringement; and by Geico, an insurance company, alleging that competitors were infringing on its trademarks used as key words to draw customers to their sites. The stock sale by auction was not the only thing that irritated the financial community. Larry Page and Sergey Brin had also found a way for top management to retain a majority of votes on most issues. The two-tier voting system is common in Europe, but far more rare in the United States, where only media companies use it to ensure their independence. (It is based on the assumption that the founders of a brand have a long-term stake in its reputation that outweighs the interests of financiers or transitory stockholders.)
All this dissent forced the Google managers to limit their interests. But, despite professional skepticism and hostility from the investment community, the shares sold well and prices rose. The crazy geniuses, along with some arrogant academics, had built one of the major companies of the 21st century.
Just more entrepreneurs?
This young company has a gilded past, a brief history that has already been told by two notable journalists: David Wise of the Washington Post, and John Battelle, cofounder of Wired, the Bible of technology enthusiasts. Both books recount the story of two highly brilliant entrepreneurs, a tale that resembles many others -- including Steve Jobs and Bill Gates, but also Henry Ford, Alfred Sloan, Dupont de Nemours, Marcel Dassault and Louis Renault. [Other examples needed] Like all their predecessors, they gained from a particularly favorable environment and circumstances. A university (actually two competing schools, Stanford and Berkeley) supplied trained engineers who were well versed in Web technology. Stanford is where two other large search engines, Yahoo! and Excite, were developed. Engineers were looking for jobs after the Internet bubble burst. The financial climate gave them access to startup capital. The legal environment supported the free flow of ideas. And mature technologies were available, enabling them to build a data-processing factory quickly, at a very low cost.
Nonetheless, they have the same entrepreneurial qualities described by Joseph Schumpeter and, before him at the very beginning of the industrial revolution, Jean-baptiste Say. Charisma: They can convince their entourage to follow them where they want to go. Arrogance: they are intelligent; they know it and don’t hide it. Ambition: They are ambitious, not just to make a fortune but also to change the world. Passion: They are driven by what they do. If we consider them today as entrepreneurs who succeeded, it was not the promise of getting rich that led them to begin the adventure, but the will to improve the search systems of the Web. They developed Pagerank on their own, without market research, which they couldn’t afford anyway.
They are iconoclasts who can make decisions that go against the grain. From looking at their history, it’s obvious that they are preoccupied with their independence, have benefited from good luck-- and have a flexible sense of ethics: When necessary, they didn’t hesitate to borrow what didn’t belong to them. Not only were they sued by Overture for patent infringement; they were later taken to task by publishers who accused them of copyright infringement. Finally, they have a strong sense of loyalty to their friends.
The role of friendship in the creation of companies has not, to my knowledge, been much discussed or analyzed. Yet it is often an important factor. Bill Gates and Paul Allen, the two founders of Microsoft, met in college classes. Dan Bricklin and Bob Franckson of VisiCalc, the original spreadsheet, met at MIT. Steve Wozniak and Steve Jobs met when they were both eighteen at a summer class for young data processing enthusiasts that was sponsored. by Hewlett-Packard (also originally founded by two college buddies, Bill Hewlett and Dave Packard). Then there are the founders of Sun. And the list goes on…a beginning based on friendship enables people to exchange and test ideas, to mutually reinforce the motivations for choosing non-traditional paths, and potentially create opportunities to find the best solutions for problems that arise. Later, when the company is founded and begins to develop, the friendship and resultant confidence make it possible to share efforts, to initiate divisions of labor, and also to correct “flaws” of inexperience. When you see two people meet a challenge, one is usually better at seeing the risks involved, while the other looks further ahead to seek solutions. In an environment as difficult and turbulent as Silicon Valley, youthful friendships and the confidence they inspire can help companies resist the pressures that threaten their independence.
Beyond these significant qualities, Brin and Page had the foresight to found and develop a company with unique methods of management that were adapted to suit the company’s economic model.