MICROSOFT AND THE INTERNET: HOW TO RESPOND TO THE RISE OF GOOGLE?
Microsoft started investing in the market for internet search and internet advertising in 2003. Despite some interesting technological and marketing moves, however, the company was still a distant third to Google 5 years later and was losing money. In the early 2009, Microsoft’s executives were wondering which bold strategic moves could be made to change this situation. How the Microsoft adventure began At the end of 2008, Microsoft, founded by Bill Gates in 1975, was a giant with revenues of $60 billion and operating cash flow of $22 billion. A large part of Microsoft’s success had come from operating systems for personal computers. Operating systems were the core software that coordinated a computer’s activities. In the early 1980s, Microsoft was chosen by IBM to develop the MS-DOS and to tailor it to IBM needs. Bill Gates, who believed that many other companies might also adopt the same system, sold IBM the rights to use the software for $80,000 but insisted that Microsoft kept the copyrights on MS-DOS. This worked beyond expectations as the many competitors -for instance, Compaq or Dell- who imitated the IBM PC adopted MS-DOS as well, paying Microsoft licence fees of $15 per machine. By the early 1990s, Microsoft had over 90% of the operating system business, Apple gathering most of the rest. Armies of developers were also developing hundreds of software applications that ran on MS-DOS. Microsoft’s dominance over the computer industry was under way. Microsoft upgraded its operating software many times, especially through the different versions of Windows, the first operating system with a graphical interface. For the popular Windows 3.0, Microsoft was charging PC makers $30 per machine, $55 per copy for Windows XP and up to $80 for Windows Vista. Development costs also rose from roughly $500 million to over $5 billion for the different versions of Windows. How Microsoft responded to opportunities and threats over the years The road to market dominance was not linear, however. Microsoft had to face many potential threats and also seized many opportunities over the years. First, Microsoft entered step by step into the development of application software, starting with Word and Excel. In the beginning, neither worked very well on the IBM/MS-DOS standard for which WorkPerfect and Lotus 1-2-3 were dominant. However, Word and Excel became standards on the Apple system, and Microsoft used its experience with Apple to develop much better graphical interfaces for its Windows-based software. In the 1990s, Microsoft also started bundling several of its key applications (Word, Excel and Powerpoint) into a popular suite: Microsoft Office. Many upgrades followed, including a much improved compatibility across the applications. As a result, by 1995, Microsoft was the leader in the market for word processors, spreadsheets and presentation software. Second, the rise of the internet generated demand for a new type of application: Internet browsers, through which users could retrieve information on any connected server around the world. The first browser was Mosaic, soon followed by a newcomer called Netscape. Netscape Navigator was launched in December 1994 and priced below $50. Netscape’s clearly asserted objective was to gain as much market share as possible early on, expecting that these will transform into revenues later. Netscape expected browsers to take over the key role in computers, making the domination of Microsoft’s operating system obsolete. By May 1995, Netscape had 70% of the browser market. This is at this point that Bill Gates fully understood the importance of the Internet and made the development of a competitor to Netscape’s browser a priority for Microsoft. The first version of Internet Explorer was released a few months later. In August 1996, Explorer 3.0 was introduced and fully matched the latest version of Netscape Navigator’s capabilities. Versions of Explorer 3.0 were also released for the Apple and Unix operating systems. Microsoft also convinced several large internet service providers (such as AOL) to make Explorer their default browser. Netscape tried to

1

survive by quickening its path of innovation and by fully opening its source code so that others could contribute to the development of Navigator. However, Nestcape’s market shares started going down fairly fast. In November 1998, Netscape was sold to AOL for $4.3 billion in stocks. By late 2002, Microsoft Explorer’s market share was over 95%. This prompted the US department of Justice to bring an antitrust case against Microsoft, starting from 1999. Prosecutors argued that bundling Internet Explorer with Windows was an abuse of a dominant position. An initial judgment was made asking for breakup of Microsoft into two, with the operating system being split for the rest of the company. Microsoft appealed, however, and was finally successful in 2001 as a settlement was reached with the US government. Under the terms of the settlement, Microsoft had to disclose some parts of its source code to other firms. Microsoft was also barred from bundling other software with Windows. In 1998, as the browser war raged, another threat appeared for Microsoft: the development of an opensource operating system called Linux. Open-source software was developed not by a company, but by a community of independent online programmers. As it was free and received a lot of press attention, Linux started getting some momentum. Microsoft’s response to Linux was first to issue stripped down versions of Windows at lower costs, targeting especially price sensitive consumers from developing countries. In 2006, a second response surprised most observers when Microsoft struck a deal with Novell, one of the companies commercializing Linux, with the purpose of making Windows compatible with Novell’s version of Linux. From 2000 to 2007, Linux’s market share of the server operating system market rose from 10% to 20%, as Microsoft’s market share went from 50% to 70%. However, Linux’ market share of the individual computer’s operating system market was below 1% in 2008. Internet, search engines and the rise of Google Another interesting development in the computer industry came from the development of search engines. A search engine consisted of three parts: webcrawler software that systematically sifted through the internet, downloading and decomposing pages; an index that stored information from the webcrawler in a way that could be searched quickly, much like an index in a book; and a user interface that took keywords from users, retrieved relevant document from the index, ranked results according to relevance and presented them in a way that users could easily understand. Most early search engines ranked pages according to how many times it contained the keyword. This led, however, to very uneven results that were easy to game and spam websites would massively embed popular search terms (such as ‘cars’) simply to top the list of queries. By the late 1990s, most of the companies that started a search engine, such as Exite or Yahoo, lost interest in pure internet search and transformed into portals that were offering many integrated services. The vision for these firms became to keep customers as long as possible, which was somewhat incompatible with pure internet searches which were pushing visitors away. Most, at the time, thought that there was no real profit potential in the search business. Two PhD students at Stanford, Sergey Brin and Larry Page, thought differently and created Google in 1998. Page’s PhD thesis focused on the mathematical structure of the internet, and suggested that a website importance could be evaluated by how many times it was referenced by other websites. Capturing all the links between websites was therefore a critical but also very challenging technical task, in which the two young men embarked recklessly. The algorithm that came from this work was at the origin of a much more effective search engine than what was offered by existing portals. By 1999, Google had a nice technology, some venture capital funding, but still no business plan. As Brin and Page disliked advertising, they first tried to licence the technology but met very little interest. The advertising idea came back when they realized that any search by an internet was also saying something about what this user was really looking for. Searches could therefore be a powerful basis for advertising. They first implemented the idea by creating a system in which advertisers had to bid to come on top of internet searches based on a keyword. To attract companies, they took a ‘cost per click

2

approach’ in which advertisers would pay only when their ad was clicked on. This was a totally different approach than what was used before, in which advertisers would pay when their ad was being displayed. Ads with Google were sold in an ongoing online auction with advertisers bidding on keywords and the highest bidder getting the best spot on the page that showed the search results. By 2002, Google had moved fully to this new approach. Advertising on Google’s search pages differed in two ways from what could be found on other portals. First, Google allowed only text-based ads and no images as they were slowing down searches. Second, instead of giving the top advertising spot to the highest bidder per click, Google ranked an ad according to the combination of the bid per click and the frequency with which users clicked on the ad. The idea there was to keep the most relevant ads for users, in order to make sure that they would come back to Google. In other words, Google tried to balance revenues from advertisers’ bids and the highest relevance of the searches for consumers. Google also kept the sign-up fees ($5) and minimum bid (1 cent) very low for advertisers. The system quickly gained momentum and revenues started flowing. Quickly, Google became a financial success (see Exhibit 2). In 2003, Google introduced AdSense: any website publisher could let Google place ads on its pages in exchange for a share in the revenue. Google would scan the publishers’ webpages to identify keywords, select relevant ads and place the ads on the pages. This provided an easy source of income that dramatically increased Google’s revenues. Many other innovations followed, such as Image search, Google News, Google product search, Google maps and Google scholar. These were characterized as ‘vertical search’ and produced more refined and precise results for searchers. Finally, Google also tried to personalize the search experience as much as possible, using information from a user’s earlier search to improve search results. Google also increased its reach by providing searches in more and more languages, by creating Google Toolbar, which allowed users to search directly from their browser without opening additional pages, or by making distribution agreements. One of these agreements was with Firefox, an open-source internet browser competing with Microsoft’s Explorer; Firefox made Google its default search engine. By 2008, Firefox had market shares of 20% in the US, 30% in Europe and over 50% in some other countries. Google toolbar was quickly imitated by Yahoo and Microsoft, as it was not only more convenient for users, but also allowed search engines to track users’ online behaviors, even when they visited unrelated sites. Google also started reaching out to other fields, introducing Gmail (a web-based email program), Google checkout (a payment system), Picasa (a software to manage photos), Google Docs (an online application providing a word processor and a spreadsheet), Google talk (an instant messaging and voice call tool), Google Android (a mobile operating system), or Google Chrome (a browser). The Google Chrome browser was meant to provide a more stable environment for running web-based software. This array of innovations didn’t seem to be particularly planned by Google, but was rather the result of internal management practices that facilitated and encouraged innovation by the company’s engineers. Google, as a company, was in fact relatively silent about its vision and future ambitions. The internet search business By 2007, internet search had grown into a $20 billion business globally, with 60-70% gross margins and was expected to double by 2013. At the same time, search had become complex and sophisticated. Google’s index had grown to 40 billion pages in 2008 (compared to 1 billion in 2000). Handling that much data was a real challenge, and required many specific capabilities. For instance, managing thousands of servers required know-how in power and heat engineering, load balancing, remote monitoring, etc. Google designed its own servers, paid a third-party to produce them, and simply by doing so became the world’s third largest maker of enterprise servers with a single customer: itself.

3

Managing fast growing indices was also something that required continuous improvements, and Google allocated many engineers to do so day in day out. Improvements were coming not only from the work of the engineers per se, but also from a process of continuous learning from customers and from their search habits. The search process was not only complex for Google, but was also quite opaque for advertisers. In addition to the historical click rate of an ad, its relevance or the loading time of the landing page, many other factors seem to be taken into account but were not clearly revealed by Google. This created some frustration among some of the advertisers. Retailers, for instance, would typically show up at the bottom of most rankings when a user searched for a product. In fact, the process of bidding on keywords was often time-consuming for advertisers. Some advertisers retained hundreds of employees to bid on thousands of keywords continuously. To alleviate associated costs, search engines helped advertisers manage the process of planning, executing and evaluating online advertising. Google Analytics, for instance, helped advertisers track traffic on their site to identify the effectiveness of certain keywords, while a large staff at Google provided personal assistance to current and potential advertisers. In spite of its complexity, many observers thought that internet search was still in its early days. For instance, users complained that 50% of their searches were unanswered, and that a complex search required 30 to 50 minutes. Interest in internet search had also increased by 2009 as some thought that the software industry would move to cloud computing. Under this model, PCs would become only devices to reach software, files or processing power hosted by some large central servers. No more software would be sold directly for PCs, as users would just subscribe to centrally maintained software available online. It was unclear however if this is how the future would unfold, and what role advertising would play in that model. Microsoft in the Internet search business Microsoft first got involved in the search business as a consequence of other business ventures, especially the conversion of some of its CD applications –such as encyclopedia- into online sites. This pushed Microsoft into the creation of the Microsoft Network: MSN. First conceived as a site open only to subscribers (in the same spirit as other portals such as AOL), MSN quickly became open to all under the competitive pressure of other free portals such as Yahoo!. Like other portals, MSN provided search services. Behind the scenes, these search services were however outsourced to other firms, especially Inktomi for the search function itself and Overture for the advertising functions. By 2003, however, the rise of Google pushed Microsoft to develop search capabilities internally. The search engine was therefore built internally and pretty much from scratch, as few Microsoft engineers were aware of the algorithm challenges arising from navigating through the web. In 2005, as over 5 billion pages were referred, Microsoft totally stopped using Inktomi and the search function for MSN became operated internally in full. Similarly, Microsoft developed internally its own advertising network, called AdCenter, which also became fully operational in 2005. Microsoft tracked several performance measures. Searches are generally split into active and passive searches. Active searches meant that the user targeted a specific search service and navigated through it. Passive searches meant that the user would use a portal for other uses (say to look at weather forecasts) and would then use the search afterwards. 40% of users used multiple search services as a result of a passive search. Google’s share of active search was much greater than its share in all searches. Active searches typically involved more words and generated more page views. On the other hand, Microsoft’s market share was mostly based on passive searches. Internet Explorer made MSN the default home page for users, and large revenues were generated for the websites by many services and advertising that had little to do with internet search per se. In fact, Microsoft would typically increase its revenues by increasing the number of MSN features, which made the website much busier than others (such as Google).

4

The performance of a search engine was also measured in terms of quality of results, i.e. how relevant these results were for the user. Through a lot of efforts, Microsoft started closing the relevance gap with Google. Experts estimated, nonetheless, that Google employed twice as many search engineers as Microsoft did in its search operations. Fast delivery of the results was also an important performance aspect of internet searches. An important aspect there was the physical proximity of the search from the data center. As a result, both Google and Microsoft built massive data centers around the world each costing roughly about half a billion dollar. Google operated 15 data centers and Microsoft 5. On average, Google delivered search results within 100 milliseconds, while Microsoft did so within 200 milliseconds. The price per click for the same keyword varied considerably and systematically across various search sites, even though this price was set by auction. A search engine that matched consumers to ads well and induced many advertisers to bid against one another would see a higher price per click. In addition, the revenue per search that a site realized depended on the firm’s ability to attract enough advertisers (so that it could fill as many advertising slots as possible with relevant ads on) and on its ability to target the right ads to the right users. Microsoft managers estimated that their price per click was comparable to Yahoo’s and 80% of Google’s (see Exhibit 5). The search companies also differed in the reach of their advertising networks. In 2008, Google catered to roughly 400,000 advertisers, while Yahoo served 175,000 and Microsoft about 60,000. Over time, the search engines introduced new features to their websites (such as corrections for misspelled queries, for instance). In fact, a company that introduced a new feature could expect its competitors to imitate it, at least crudely, within a couple of months. Since its launch in 2005, Microsoft had tried to differentiate MSN’s search process in several ways. For instance, Microsoft launched in 2007 the ‘Live search club’ in which users would win prizes in online games. For a short period of time, this increased MSN’s search volume by one third. For advertisers, the Live search club allowed to target by user demographics and not only by search query. Microsoft did not persist in all its efforts to match Google’s initiatives. For example, Google started in 2003 a process of scanning books in university libraries and make them searchable through Google Books. Microsoft started replicating this in 2006, but stopped in 2008. For years, rumors had also circulated that Microsoft was willing to buy Internet portal Yahoo in order to strengthen its competitive position. Microsoft made these rumors concrete in January 2008 with a $45 billion bid on Yahoo (a 62% premium on Yahoo’s market capitalization). The bid, however, was unsolicited and Yahoo’s CEO convinced its board to reject it. Clashes of culture and technology were alleged to undermine the deal. The deal was withdrawn by Microsoft in May. What was the right course of action for Microsoft now?

5

Exhibit 1 – Microsoft revenues per business segment
Segment Client Server and tools Product line Windows operating system for personal PCs Licensing products, applications, tools, contents for IT professionals including licences for Windows server, product support services and consulting services Online advertising platform, emails, instant messaging, online information offering, MSN portal Microsoft Office (including Word, Excel and Powerpoint) and business solutions for financial management, customer relationships or supply chain management Xbox 360 platform, digital music and entertainment platform, PC software games, online software games Sales, marketing, product support devices, human resources, legal, finance, IT, R&D, legal expenses 2008 Revenues (billion $) 16.9 13.2 2008 Operating income (billion $) 13.1 4.6

Online services

3.2

(1.2)

Microsoft Business Division

18.9

12.4

Entertainment and devices Division

8.1

0.4

Corporate level activity

-

(6.7)

Source: Microsoft

Exhibit 2 – Google’s results
2002 Income statement (in billion $) Revenues Google websites Good AdSense Licensing and other revenues Total Cost and expenses Traffic acquisition costs Other costs of revenues R&D Sales and marketing General & Administrative Total Profitability Income from operations Net income Other information Capital expenditures Cash flow from operations International revenues as % total revenues Number of permanent employees (in 000) Source: Google 2003 2004 2005 2006 2007

0.1 0.3 0 0.4

0.8 0.6 0 1.4

1.6 1.6 0 3.2

3.4 2.7 0.1 6.2

6.3 4.2 0.1 10.6

10.6 5.8 0.2 16.6

0.1 0.04 0.04 0.05 0.03 0.3 0.2 0.1 0.04 0.2 22% 0.6

0.5 0.1 0.2 0.2 0.1 1.1 0.3 0.1 0.2 0.4 29% 1.6

1.2 0.2 0.4 0.3 0.2 2.5 0.6 0.4 0.3 1.0 34% 3

2.1 0.5 0.6 0.5 0.4 0.1 2.0 1.5 0.8 2.5 39% 5.7

3.3 0.9 1.2 0.8 0.8 7.1 3.5 3.1 1.9 3.6 43% 10.7

4.9 1.7 2.1 1.5 1.3 11.5 5.1 4.2 2.4 5.8 48% 16.8

6

Exhibit 3 – Search engine market shares
Market shares July 2007 Google Yahoo MSN Others 55.2 23.5 12.3 9 Market shares November 2007 58.6 22.4 9.8 9.2 Market shares May 2008 61.8 20.6 8.5 9.1 Market shares Nov. 2008 63.5 20.4 8.3 7.8 Searches (million) Nov. 2008 7,784 2,506 1,015 959

7

Exhibit 4 – US and Global online advertising market (actual and estimated) (billion $)
2007 Global Search marketing Total online advertising Percent of total advertising US Search marketing Online display marketing Email marketing Emerging channel marketing Online video marketing Total online advertising Percent of total advertising 2009 2013

20 45 8%

25 67 11%

40 98 13%

8.1 6.1 2.7 1.0 0.5 18.4 8%

13.3 9.4 3.6 3.5 1.9 31.6 12%

25.3 14 4.3 10.6 7.2 61.3 18%

Exhibit 5 – Search engine: Prices and revenues comparison
Google Average price per click (2008) Share of US search queries (2007) Share of US search revenues (2007) Average price per click relative to Google (2008, in %) Revenue per search relative to Google (2007, in %) 0.71 58.2 66.2 100 100 Yahoo 0.51 24.1 17.1 72 62 MSN 0.59 10.8 7.0 83 57

8