How Google Built the Internet's Most Profitable Search Business—and Why Its Dominance Is Being Challenged

In 1997, Larry Page and Sergey Brin were trying to persuade established internet companies to buy their search technology for approximately $1 million. Excite came close. Yahoo had an opportunity, too. Neither acquired it.

Three decades later, Alphabet, Google's parent company, had become one of the world's largest and most profitable technology businesses. Its search engine controlled approximately 90% of the United States search market, and its advertising system generated extraordinary profits from something people did billions of times: ask questions.

The remarkable part is that Google did not invent search, keyword advertising, online auctions, or distributed computing. It combined existing ideas with original engineering, aggressive commercial decisions, and an unusually effective distribution strategy.

In their June 30, 2025 episode, Google: The Origin of Search, Acquired hosts Ben Gilbert and David Rosenthal traced how Page and Brin developed that combination. Their account reveals a business built around three breakthroughs: better search results, cheaper computing, and a more profitable advertising marketplace.

But the story also contains a contradiction. Google initially succeeded by overcoming internet companies that placed their advertising interests ahead of search quality. Decades later, its own distribution agreements became central to federal antitrust judgments.

Understanding Google's rise therefore requires examining not just what it invented, but what competitors rejected, what investors financed, what advertisers paid for, and how the company turned early advantages into a dominant market position.

1. Google Was Built Around a Problem Nobody Had Solved Well

The internet was growing beyond human control

By the mid-1990s, the World Wide Web was becoming too large for people to organize manually.

According to figures cited in the episode from John Battelle's The Search, the number of websites increased from approximately 130 in 1993 to more than 600,000 in 1996. Whatever the exact annualized growth rate, the change was extraordinary.

The early internet already had search engines. AltaVista, Excite, Infoseek, Lycos, HotBot, and others helped users find information. Yahoo, established by Stanford students Jerry Yang and David Filo, approached the problem differently.

Yahoo maintained a directory of websites selected and described by human editors. Users could browse categories to discover useful resources.

This worked when the web was relatively small. But editors could not continuously classify hundreds of thousands of websites, much less the millions of individual pages being created.

The problem was becoming less about finding websites that contained particular words and more about finding the right websites.

Larry Page began investigating that problem at Stanford in 1995. At the time, he was not trying to establish a search company.

Larry Page and Sergey Brin came from unusually technical families

Page was born in Lansing, Michigan, in March 1973. His father, Carl Page Sr., was a computer science professor at Michigan State University, while his mother, Gloria, taught programming there. Both had studied at the University of Michigan.

Computers and programming were part of Page's childhood long before personal computers became commonplace.

His father spent the 1979–1980 academic year on sabbatical at Stanford, taking the family to Palo Alto. Page was six or seven. The experience introduced him to the emerging culture of Silicon Valley.

His older brother, Carl Page Jr., later studied computer science at Michigan and worked at Microsoft and Mentor Graphics.

Page also developed an early interest in entrepreneurship. He later recalled knowing by approximately age 12 that he wanted to establish a company. His reasoning was that inventing something useful was not enough; a business could bring that invention to millions of people and finance further development.

Sergey Brin had a different childhood but a similarly technical upbringing.

Born in Moscow in August 1973, Brin grew up in a Jewish family in the Soviet Union. His father was a mathematician who decided to pursue emigration after attending an international mathematics conference in the late 1970s.

After approximately two years of effort, the family moved to the United States. Brin's father became a mathematics professor at the University of Maryland, while his mother worked at NASA's Goddard Space Flight Center.

Brin graduated from high school at 16, completed undergraduate studies in mathematics and computer science at Maryland in three years, and interned at Wolfram Research, the developer of Mathematica.

He arrived at Stanford before Page.

A partnership that began with arguments

Page visited Stanford in 1995 while considering its doctoral program.

The familiar account of his first encounters with Brin describes two young mathematicians and computer scientists arguing about transportation, urban planning, and technology almost immediately.

Anna Patterson, who organized the prospective-student weekend and later became a senior Google engineering executive, supplied the Acquired hosts with another detail.

She recalled that the pair first met at the British Bankers Club in Menlo Park. Their conversations continued late into the evening, and Charles Schwab, the financial-services entrepreneur and Stanford alumnus, reportedly paid the students' bill.

The two subsequently developed an unusually durable founding partnership.

Unlike several technology companies where one founder eventually became overwhelmingly dominant, Page and Brin continued exercising substantial influence together. They even shared an office at Google.

Their instincts were not identical. Brin became associated with experimental projects such as Google Glass, while Page was often more closely associated with focused product development and business operations. However, their technical and commercial responsibilities overlapped considerably.

The episode also challenges the impression that they were reluctant academics who accidentally built a profitable company. Both had substantial entrepreneurial ambitions well before Google became commercially successful.

PageRank emerged from an unrelated research proposal

Page's original doctoral idea involved website annotations.

Instead of relying on Yahoo's centralized directory, people would leave comments describing websites they had visited. Other users could then discover useful observations about those sites.

There was an immediate difficulty.

A popular website could accumulate thousands of annotations. Without a way to distinguish credible comments from worthless ones, the annotation system would become another unmanageable collection of information.

Stanford professor Terry Winograd encouraged Page to refine the problem. Page and Brin eventually realized that ranking annotations was less interesting than ranking the websites themselves.

Page's recollection, quoted in the episode, captures the unexpected direction of the work:

“It wasn't that we intended to build a search engine. We built a ranking system to deal with annotations.”

The decisive insight came from academic research.

Scientific papers are often evaluated partly through citations. A paper referenced by many important publications may be more influential than one cited only by obscure research.

The same logic could apply to the web.

A hyperlink from one website to another resembled an academic citation. A site receiving links from respected websites could be treated as more authoritative than one receiving similar numbers of links from low-quality sources.

This became the foundation of PageRank, named after Larry Page.

Crucially, PageRank did not merely count links. It evaluated their relative importance.

The words attached to hyperlinks provided another useful signal. When someone linked to a website using the phrase "computer science research," those words offered information about what the destination contained—even if the destination did not describe itself particularly well.

The original 1998 research paper by Sergey Brin and Lawrence Page documents these principles. It describes a prototype indexing at least 24 million web pages and using the web's link structure to improve search results.

That paper is particularly revealing because it establishes what Google had already accomplished before becoming a major commercial company.

Why Google needed to copy much of the web

PageRank had a difficult technical requirement.

A webpage could reveal which other pages it linked to. But there was no universal system that could tell a website which pages were linking back to it.

To identify incoming links, Page and Brin needed to crawl large portions of the internet, collect the pages and their links, and then reverse those relationships.

They called the project BackRub, referring to backlinks.

Page wrote an early Java implementation, but it was unreliable. Their Stanford colleague Scott Hassan helped produce a working Python implementation.

Once they began calculating link-based rankings, the results looked promising. Important websites generally appeared near the top.

The timing mattered enormously.

In 1996, the web was already large enough that automated search was useful, but still small enough for an ambitious university project to index substantial portions of it.

The hosts argue that starting several years later would have been prohibitively expensive for a small research team. That is a counterfactual judgment rather than a demonstrated impossibility, but the underlying constraint is clear: crawling, storing, and analyzing a rapidly expanding internet demanded increasing amounts of computing capacity.

Google had identified a better way to rank information just as the need for ranking was exploding.

2. Why Yahoo, Excite and AltaVista Failed to Capture the Opportunity

AltaVista had impressive technology but difficult economics

Google was not the first search engine capable of processing substantial amounts of information.

AltaVista, developed at Digital Equipment Corporation's Western Research Laboratory, was an important predecessor.

Engineers used parallel computing to build a search system that could crawl and index far more pages than competing services.

AltaVista launched with approximately 16 million indexed pages at a time when some rivals covered only about one million.

But three different capabilities determined search quality: the relevance of results, the amount of the web indexed, and the speed of retrieving answers.

A large index was not automatically a good index. Results could become outdated when websites changed, while ranking systems based heavily on keyword matches were vulnerable to manipulation.

Websites could repeat popular terms, including invisible text, to make themselves appear relevant.

AltaVista also relied on expensive Digital Equipment Corporation hardware. That architecture created substantial costs before the search industry had established a reliable way to recover them.

Google would eventually improve both the quality of results and the economics of processing searches.

Yahoo was becoming a media company

Yahoo had different strengths.

Its directory, brand, editorial organization, and homepage made it one of the internet's most important destinations.

Founded in 1994, Yahoo went public in 1996 at approximately a $1 billion valuation and reached roughly $20 billion in market capitalization by 1998, according to the episode.

Its business increasingly resembled that of a media company.

People visited Yahoo to browse news, entertainment, services, directories, and advertisements. The company wanted to keep them engaged across its properties.

That created a problem for efficient search.

Internet portals commonly sold graphical advertisements according to impressions. Advertisers paid when banners were displayed, meaning additional page views created additional revenue opportunities.

An excellent search engine might direct someone to another website immediately.

From the user's perspective, that was a successful result. From a portal's perspective, it could mean losing an advertising impression.

The reported $1 million mistake

In 1997, Page and Brin approached established internet companies about licensing BackRub.

They hoped to help implement the technology and then return to their doctoral studies.

Excite reportedly came closest to adopting it.

According to the account discussed by the hosts, venture capitalist Vinod Khosla, then associated with Kleiner Perkins and involved with Excite, helped negotiate a transaction valued at approximately $1 million in cash and stock.

Page and Brin would assist with implementation before returning to Stanford.

A demonstration reportedly revealed the central conflict.

BackRub produced relevant results that helped users find information quickly. That meant users could leave Excite rather than continue viewing its pages.

The proposed agreement collapsed.

The famous explanation—that Excite rejected Google because its search worked too well—comes from retrospective accounts and should not be treated as a complete, independently established explanation for the decision. Nevertheless, it illustrates a genuine difference between Google's product priorities and the portal advertising model.

Yahoo also reportedly declined an early opportunity to obtain PageRank technology for approximately $1 million. Page and Brin approached Infoseek, Lycos, and other established companies without reaching an acceptable agreement.

They increasingly concluded that the incumbents were focused on media, advertising, and page views rather than improving search.

The failed licensing attempts pushed them toward creating an independent search engine.

The founders considered calling it WhatBox but abandoned the idea partly because the name resembled that of an adult website.

Instead, they selected Google, derived from googol, the mathematical term for \(10^{100}\). Accounts differ about how the spelling changed.

Brin designed the early colorful logo and sparse homepage. Instead of offering a crowded collection of links and services, Google presented a search box.

Its simplicity reinforced a clear proposition: the product existed to help people find information, not to keep them browsing a portal.

3. The Investors Who Helped Turn BackRub Into Google

A research project that overwhelmed Stanford

By spring 1998, Google was handling approximately 10,000 searches each day.

Demand spread from Stanford to other universities and Silicon Valley, largely through word of mouth.

But the growing service was consuming Stanford's computing and networking resources.

At one point, Google's operation reportedly accounted for approximately half the university's internet bandwidth and contributed to a network outage.

Page and Brin were also improvising their hardware arrangements. According to the episode, they sometimes collected computers intended for other research projects from loading docks and returned the equipment when its intended users needed it.

This could not continue indefinitely.

Google needed independent funding, facilities, and equipment. Stanford would also need to establish ownership arrangements for the intellectual property created through the research.

The university ultimately received an equity stake estimated by the hosts at approximately 1%.

Andy Bechtolsheim wrote a check before Google existed

Stanford professor David Cheriton introduced Page and Brin to Andy Bechtolsheim, a cofounder of Sun Microsystems.

Cheriton and Bechtolsheim had previously cofounded Granite Systems, which Cisco acquired for approximately $220 million.

Following a demonstration at Cheriton's home, Bechtolsheim reportedly offered to invest $100,000.

There was an unusual complication. He wrote the check to Google Inc. before the company had been legally incorporated.

The founders needed to establish the corporation, settle ownership arrangements, and open a bank account before they could deposit it.

Additional investors joined the financing.

Cheriton contributed $100,000. Ram Shriram, who had worked at Netscape and Junglee before Amazon acquired the latter, invested $250,000.

Shriram then introduced Page and Brin to Jeff Bezos. The Amazon founder agreed to invest another $250,000 after learning of Shriram's commitment.

These four investments formed approximately $700,000 of Google's eventual $1 million seed financing, at a valuation the hosts estimated around $10 million after investment.

The long-term value of those early stakes became extraordinary.

The hosts calculated that Bezos's $250,000 investment could have been worth approximately $200 million at Google's IPO, or around $20 billion in 2025 if retained. Those are hypothetical values, not evidence of when Bezos actually sold.

Google's first office was a Menlo Park garage owned by Susan Wojcicki, then an Intel manager. She subsequently joined Google and eventually became YouTube's chief executive.

Scott Hassan and Carl Page Jr. showed what internet companies could become

While Google was emerging, Scott Hassan and Larry's older brother, Carl Page Jr., established eGroups.

The company received backing from Sequoia Capital, including Michael Moritz, and later sold to Yahoo for approximately $400 million. Its technology became associated with Yahoo Groups.

For Page and Brin, the success was another demonstration that internet technology could support companies of substantial value.

Their ambition was not limited to publishing an influential research paper.

They wanted to make their technology widely useful, and establishing a business was becoming the way to accomplish it.

The $25 million round investors later questioned

In early 1999, Google hired Salar Kamangar, a recent Stanford graduate, as employee number nine.

One of his assignments was preparing the company's business plan and presentation to venture investors.

The plan proposed three sources of revenue:

1. Selling enterprise search software to companies searching their internal documents.
2. Selling conventional advertisements.
3. Licensing Google's organic search results to internet portals.

Enterprise search was initially presented as the primary commercial opportunity.

Google received approximately $20,000 from Red Hat for an early enterprise-search agreement, demonstrating that customers would pay for the technology.

But demand was limited.

The advertising proposal was less developed. Page and Brin disliked conventional banner advertisements, particularly because they slowed down webpages and interfered with Google's clean experience.

The third idea—licensing search results—resembled the existing business of Inktomi, which supplied search technology to internet companies that did not want to build their own systems.

None of these opportunities yet explained how Google would become exceptionally profitable.

Nevertheless, in 1999, Sequoia Capital and Kleiner Perkins invested a combined $25 million at a $100 million post-money valuation.

Each firm acquired approximately 12.5% ownership, and Michael Moritz of Sequoia and John Doerr of Kleiner Perkins joined Google's board.

The investment was widely presented as an unusual collaboration between celebrated venture investors, although people interviewed for the episode said their firms frequently worked together.

Google reportedly received interest at valuations as high as $150 million but preferred the investors with whom it already had relationships.

Even at $100 million, the price appeared ambitious for a company with limited revenue.

The founders had developed a compelling product. They had not yet proved that it could become a compelling business.

4. Google's Real Engineering Breakthrough Was Making Computing Cheap

PageRank needed an infrastructure capable of growing with the web

PageRank was only one component of a successful search engine.

Google needed to discover webpages, revisit them as they changed, store their contents, analyze links, calculate rankings, and return results quickly to users.

Every expansion of the internet increased the work required.

Traditional enterprise computing relied heavily on expensive servers designed for reliability. Google pursued a different approach: combining thousands of inexpensive computers into systems that could tolerate individual machine failures.

This was central to its eventual profitability.

Google could not afford to operate a global search engine if every increase in demand required proportionately expensive enterprise hardware.

Urs Hölzle and Jeff Dean helped develop Google's technical advantage

Urs Hölzle joined Google as employee number eight.

A former Stanford doctoral student and University of California, Santa Barbara professor, he had worked on Sun's Java Virtual Machine. His initial title at Google was reportedly "search engine mechanic."

Hölzle helped repair and scale Google's systems and remained a central infrastructure leader for more than two decades, until 2023.

Jeff Dean joined from Digital Equipment Corporation and became another defining engineer.

His work eventually extended across search systems, advertising, distributed storage, MapReduce, Bigtable, TensorFlow, and artificial intelligence.

Their engineering challenge was not simply making computers faster.

It was developing systems where thousands of relatively unreliable computers could operate as one dependable service.

Google designed its software to expect hardware failure

Early Google hardware was inexpensive and frequently improvised.

Engineers assembled machines from components instead of buying complete servers. Some were mounted on corkboard without conventional cases, allowing dense configurations and easier replacement.

Its facilities could resemble workshops more than traditional corporate data centers.

At the Exodus colocation facility in Santa Clara, Google occupied space near Inktomi, whose installation reportedly featured carefully arranged Sun servers.

Google's neighboring operation relied on cheaper, densely packed hardware.

The hosts even recount an unverified story that Google borrowed an electrical circuit from Inktomi's neighboring space. The anecdote remains unconfirmed, but reflects the improvised conditions described by early participants.

Data-center pricing also shaped Google's decisions. Facilities often charged substantially according to occupied floor space, encouraging Google to pack more computing capacity into smaller areas.

The disadvantages of inexpensive hardware were real.

The episode cites annual equipment failure rates exceeding 10%, compared with approximately 3–4% for typical industry servers.

Some inexpensive memory also lacked the error-checking capabilities available in premium equipment.

Google addressed these problems through software.

Information was replicated so that the failure of one machine would not destroy it. Systems checked for corruption, detected failures, and redirected work to functioning computers.

Instead of attempting to prevent every component failure, Google designed an infrastructure that could survive frequent failures.

Google File System, MapReduce and Bigtable

As Google's index expanded, no single machine could store and process everything efficiently.

Google developed systems that divided large amounts of information across many computers.

The Google File System stored data in chunks of approximately 64 megabytes and maintained information about where those chunks were located. Replication supported reliability, while multiple machines could work on different pieces of a dataset.

The original 2003 Google File System research paper, written by Sanjay Ghemawat, Howard Gobioff, and Shun-Tak Leung, independently establishes the architecture's purpose.

It documents a distributed system designed for inexpensive commodity hardware, substantial data volumes, and routine component failures.

Google also developed MapReduce.

This approach allowed engineers to divide large computational jobs into smaller tasks that could run across many machines, then combine their results.

The 2004 MapReduce paper, by Jeffrey Dean and Sanjay Ghemawat, describes processing large datasets across thousands of commodity computers, with the system managing task distribution and machine failures.

Bigtable provided distributed storage for structured information. Its 2006 research paper described systems capable of handling petabytes of information across thousands of servers.

These technologies were not all created at Google's founding. They emerged as the company's computing requirements grew.

Their significance extended beyond Google. Yahoo later developed Hadoop as an open-source approach inspired partly by Google's published work on distributed data processing.

Lower computing costs created higher advertising margins

Google's architecture produced two commercial advantages.

First, it reduced the cost of processing each search as traffic increased.

Second, its expanding network of data centers improved response times. Google could serve users from geographically distributed infrastructure rather than depending on a small collection of centralized machines.

The company emphasized these capabilities publicly, often displaying the number of indexed pages and the fraction of a second required to produce results.

Search quality, index coverage, and speed reinforced one another.

The hosts estimated that Google's mature advertising operation on its own search properties achieved gross margins of approximately 85–87%.

That figure refers to the economics of first-party search advertising, not Alphabet's consolidated corporate gross margin.

The distinction matters because Google would later earn substantial revenue from advertisements placed on other companies' websites, where much of the advertising income had to be shared with publishers.

Google's infrastructure made a crucial difference: it could process searches at a lower cost than many competitors, leaving more money available to develop products, recruit engineers, and acquire users.

5. Netscape and Yahoo Financed Google Before Advertising Took Off

Netscape nearly overwhelmed Google's systems

Omid Kordestani joined Google from Netscape to develop commercial revenue.

He had experience in sales and business development but encountered a company whose original enterprise-search plan was producing limited results.

Rather than waiting for enterprise demand to expand, Kordestani pursued relationships with internet portals.

His Netscape connections proved especially valuable after Netscape became part of AOL.

Google agreed to supply organic search results for Netscape's web directory.

The partnership introduced approximately three million searches daily to Google's systems, according to the episode.

Traffic increased so quickly that the infrastructure approached failure.

Sergey Brin reportedly warned Kordestani that Google's systems could not handle both its own consumer search traffic and the new Netscape demand.

Google temporarily shut down its own service to preserve Netscape's availability.

It was a revealing commercial decision.

Google.com was the product that had made the company popular. Yet management prioritized keeping a contractual customer operational.

The episode describes this as an important step in the company's transition from university research to a dependable technology supplier.

The agreement also created distribution.

Users searching through Netscape encountered the "Powered by Google" name. As Google's reputation improved, that branding became valuable to partners rather than merely promotional language Google had to negotiate into contracts.

Yahoo rejected Google's technology, then became its major customer

On June 26, 2000, Yahoo announced that Google would become its default search-results provider.

Google's results would complement Yahoo's manually curated directory.

The original Google–Yahoo announcement confirms the arrangement. Contemporary reporting also described Yahoo's replacement of Inktomi and its users' interest in Google's relevance-based results.

The hosts estimated that the agreement approximately doubled Google's daily search volume to 14 million.

Yahoo also invested $10 million in Google as part of the broader relationship.

The timing was critical.

The NASDAQ had peaked in March 2000, and the technology industry's financial environment was deteriorating.

Google still needed considerable spending on hardware, engineering, and data centers. Fresh venture capital was becoming harder to obtain.

Yahoo paid Google approximately $7.2 million for organic search results in 2001, according to the episode.

That money, alongside revenue from Netscape and other partners, helped support Google while its advertising system remained immature.

The partnership therefore accomplished two things simultaneously: financing operations and introducing the Google brand to large numbers of internet users.

The partnership eventually became a threat to Yahoo

When Yahoo first adopted Google, it was buying technology that improved its existing service.

But users could increasingly recognize Google's brand and visit Google.com directly.

Yahoo was effectively helping distribute a search engine that could eventually compete against its own portal.

The competitive relationship changed quickly.

By 2001, Google's founders were discovering how valuable search advertising might become.

Yahoo CEO Terry Semel reportedly proposed buying Google for approximately $1 billion.

The reasoning discussed in the episode was that Yahoo already represented Google's largest customer and knew how much it paid for search services. Based on those payments, the startup did not appear to justify an enormous valuation.

Page and Brin rejected the offer.

In 2002, Yahoo reportedly returned with a proposal around $3 billion. Google's founders countered at approximately $5 billion.

The negotiations failed.

A $5 billion acquisition would have been particularly difficult for Yahoo to finance after the dot-com collapse.

By the time Google's advertising business became demonstrably profitable, the opportunity for Yahoo to buy it at a modest valuation had disappeared.

Yahoo subsequently developed its own search capabilities through acquisitions and projects, including Inktomi, Overture, and Project Panama.

Its commercial identity and technical architecture were becoming increasingly difficult to reorganize around the focused search-and-advertising model Google had developed.

6. Eric Schmidt Helped Turn Google Into a Managed Company

The founders resisted an outside chief executive

Google's 1999 venture financing came with an important condition.

John Doerr and the board wanted Page and Brin to recruit an experienced CEO.

The founders agreed but were reluctant to transfer meaningful authority to someone unfamiliar with their technology or working methods.

The search reportedly lasted approximately 16 months.

Investors introduced them to numerous experienced executives, but almost none satisfied the founders.

They reportedly identified Steve Jobs as the person they would most enthusiastically accept.

Eric Schmidt eventually emerged as a practical choice.

Schmidt had worked at Sun Microsystems and served as CEO of Novell. He understood engineering but also had experience managing a large public company.

He joined Google's leadership in March 2001.

The arrangement preserved Page's central role in product development and Brin's influence over technology while giving Schmidt responsibility for much of the organization's management and commercial operations.

The three reportedly coordinated through daily meetings.

Schmidt also adapted to Google's informal engineering culture. He famously shared office space with an engineer for months rather than insisting on conventional executive arrangements.

Google wanted engineers willing to attempt unusual things

Google's culture emphasized solving problems that established companies considered impractical.

The founders favored technically gifted employees who might not have extensive experience with conventional industry limits.

There were drawbacks to that approach, but it helped Google recruit people attracted by unusually difficult engineering problems.

The mission, established early in the company's history, was to:

“organize the world's information and make it universally accessible and useful.”

The founders viewed that ambition and commercial success as compatible.

A better search engine would attract users. Their activity could create commercial opportunities. Revenue could then finance improvements to the product.

Schmidt took a considerable professional risk by joining during the dot-com collapse, when Google's ultimate financial prospects remained uncertain.

His contribution was helping establish the organizational capacity to support the founders' ambitions without fundamentally replacing their engineering-driven approach.

7. Bill Gross Had Already Invented the Advertising Model Google Needed

GoTo made search advertising measurable

The breakthrough that eventually made Google extraordinarily profitable did not originate inside Google.

Bill Gross, founder of the incubator Idealab, had developed a search engine called GoTo.com.

He introduced its advertising concept publicly at the TED conference in February 1998.

Gross recognized that internet searches revealed something especially valuable: a person's immediate interests.

Someone searching for a mortgage, insurance policy, computer, or restaurant was demonstrating potential demand.

Advertisers might pay substantially more to reach that person than to display an unrelated banner to a random visitor.

GoTo allowed advertisers to bid on search keywords.

Instead of paying for a predetermined number of impressions, businesses could pay when users clicked their advertisements.

Advertisers also managed their bids through a self-service system rather than negotiating every campaign with a salesperson.

The model made advertising more directly measurable.

An advertiser could compare the cost of obtaining a visitor with the revenue that visitor subsequently generated.

The market initially disliked paid search

Gross faced considerable resistance.

According to the episode, his TED presentation was met with boos from people concerned that selling search positions would corrupt the internet's information system.

Open Text had experimented with paid search in 1996 and encountered similar objections.

GoTo distinguished its commercial rankings from organic results and initially displayed the prices advertisers were willing to pay.

Gross also believed that bids could serve as a commercial relevance signal. Businesses willing to pay more for a particular keyword might have products or services strongly related to it.

That assumption was useful but incomplete. A high bidder was not necessarily the advertiser most helpful to a user.

Still, GoTo demonstrated that businesses would pay for search-driven customers.

The episode reports that GoTo generated approximately $100 million in its first year and had around 8,000 advertisers by mid-1999, when it was publicly traded.

The system eventually became primarily a supplier of paid advertisements to other websites.

GoTo changed its name to Overture, securing relationships with AOL, MSN, Yahoo, and other portals.

Yahoo became deeply dependent on Overture's advertising technology. The hosts estimate that Overture-powered ads eventually represented approximately three-quarters of Yahoo's revenue.

Yahoo ultimately acquired Overture for approximately $1.6 billion, with Microsoft also having shown interest.

GoTo almost acquired Google

Before Google's 1999 Series A financing, Page and Brin approached Gross about a proposed acquisition valued around $200 million.

Gross reportedly liked the combination.

Google could provide superior organic results, while GoTo supplied the commercial advertising system.

Together, the technologies could have produced a comprehensive search business before Google independently developed AdWords.

But GoTo's board rejected the proposal.

GoTo already had substantial advertising revenue and a valuation around $2 billion. Surrendering approximately 10% of the company to acquire a startup with little revenue was difficult for directors to justify.

Google remained independent.

That decision left the two companies competing to determine how search advertising would develop.

8. Google Improved GoTo's Model With Relevance and Auctions

Early AdWords was not yet the system that changed advertising

Google initially attempted to sell advertising through traditional methods.

Omid Kordestani and Tim Armstrong developed commercial relationships with agencies and advertisers. Text-only advertisements preserved search speed, but the sales process involved contracts, faxed instructions, and manual campaign setup.

Advertisements were originally priced according to impressions rather than clicks.

Google's engineers had already conducted experiments suggesting that a more effective model was possible.

Jeff Dean and Marissa Mayer used Amazon affiliate links to display relevant book offers alongside searches.

The experiment showed that search traffic could produce not just clicks but actual purchases.

Because affiliate commissions depended on completed transactions, Google could evaluate commercial results rather than simply count exposure.

The problem was scaling this approach to thousands of advertisers and millions of queries.

In October 2000, Salar Kamangar began leading improvements to AdWords.

Google adopted self-service advertising and moved toward keyword auctions and cost-per-click billing.

But it added a decisive improvement.

A relevant advertisement could defeat a higher bidder

GoTo's early paid rankings placed enormous importance on the amount advertisers bid.

Google introduced a ranking method combining the bid with the likelihood of someone clicking the advertisement.

In a simplified version:

Expected advertising value = bid per click × predicted click-through rate.

An advertisement bidding 50 cents with a 1% chance of receiving a click would have a lower expected value than one bidding 30 cents with a 3% chance.

In that example, the lower bidder could produce more revenue per advertisement opportunity.

This also improved the user's experience by giving relevant advertisements a better chance of appearing prominently.

The model aligned three interests: users received more useful advertisements, advertisers could compete through relevance rather than financial strength alone, and Google increased the expected revenue generated by each search.

Google initially implemented click-through-based rankings while retaining impression-based billing.

That created an unintended loophole.

Advertisers could click their own advertisements to increase measured click-through rates and improve rankings without paying additional charges directly tied to those clicks.

The company completed its transition toward cost-per-click auctions by early 2002.

Second-price auctions reduced bidding complexity

Google also adopted a variation of the second-price auction.

Instead of making an advertiser pay the full amount it had bid, the system could charge an amount determined by the next-best competing offer and the applicable ranking rules.

In the simple example discussed in the episode, an advertiser bidding 50 cents against a competitor bidding 30 cents could pay approximately 31 cents.

The actual advertising marketplace became more sophisticated than this illustration, particularly when relevance and quality were incorporated into auction calculations.

But the underlying idea was important.

Advertisers could submit bids reflecting what they were willing to pay without continually adjusting prices to remain just ahead of competitors.

Google accepted potentially lower immediate revenue from individual clicks in exchange for a more dependable and widely usable marketplace.

Sheryl Sandberg, who became important to Google's advertising operations, sought advice from former Treasury Secretary Larry Summers about explaining the auction.

Summers connected it to established Vickrey-style auction theory.

That gave Google a recognized economic framework for a system advertisers initially found unfamiliar.

Project Sunset was a difficult commercial transition

Changing an advertising marketplace is more complicated than changing a piece of software.

Google already had paying customers using its original advertising arrangements.

Moving them into the new system required replacing existing pricing structures while preserving relationships and revenue.

The project was called Project Sunset.

Overture had also experimented with incorporating click-through rates into paid rankings. According to the episode, it retreated when advertisers became confused by a system in which visible bid prices did not directly determine placement.

Google persisted.

Its infrastructure also supported real-time advertising decisions at a scale that would have been difficult for smaller competitors.

The company had combined GoTo's commercial model with a relevance-ranking system and inexpensive distributed computing.

The result was not simply an improved advertising product. It was a marketplace whose economics improved as participation increased.

The Overture patent settlement was more complicated than a single payment

GoTo and Overture also pursued patents related to search advertising.

The episode discusses Gross's difficulty obtaining broad protection after publicly disclosing his concepts, along with narrower patents his company later secured.

Overture sued Google for alleged infringement in April 2002. Google disputed both infringement and the validity of the patent.

The case ended in August 2004, after Yahoo had acquired Overture.

Retrospective accounts sometimes describe the settlement as a payment worth approximately $360 million, but Google's SEC disclosures establish a more precise financial picture.

Google issued Yahoo 2.7 million shares, valued at $229.5 million using Google's $85 IPO share price.

The settlement resolved two separate disputes: Overture's patent-infringement lawsuit and a disagreement concerning stock warrants issued through Google's earlier Yahoo partnership.

Google allocated $201 million to a nonrecurring charge associated with the settlement and $28.5 million to intangible assets.

Overture dismissed its lawsuit and granted Google a perpetual license to the relevant patent and related applications.

This matters because the transaction was not simply a cash payment for stealing an advertising invention. It was a negotiated stock settlement covering both intellectual property and an earlier investment-related disagreement.

9. Google Bet Its Survival on AOL

The company guaranteed money it did not have

One of Google's most consequential commercial decisions came in May 2002.

AOL had approximately 34 million users.

Inktomi supplied its organic search technology, while Overture handled paid search advertisements.

Google wanted both businesses.

It proposed an integrated system in which Google supplied search results and AOL's advertisers entered Google's advertising marketplace.

Winning AOL would produce a major increase in search activity. More importantly, it would expand the inventory and demand flowing through Google's auctions.

AOL negotiated demanding terms.

Google reportedly agreed to provide 85% of advertising revenue generated through the partnership.

AOL also received warrants to buy approximately 7.4 million Google shares at $3 each, representing a potential investment of roughly $22 million.

The most dangerous provision was a $100 million minimum revenue guarantee.

Google did not have $100 million in cash when negotiating the agreement.

If advertising performed poorly, the company would still owe the guaranteed amount.

Omid Kordestani warned that the founders were effectively gambling with Google's existence.

Page's response, quoted in Ken Auletta's account and repeated in the episode, was:

“We should be able to monetize the pages. If not, we deserve to go out of business.”

This was a genuine financial risk rather than an ordinary optimistic sales forecast.

A fixed guarantee could convert disappointing advertising performance into a crisis requiring money Google did not possess.

Why AOL's advertisers mattered more than the contract's immediate margin

Google's reasoning went beyond AOL's search traffic.

An advertising marketplace becomes more useful when it contains a wider range of participating businesses.

Imagine a search for a highly specialized industrial component.

If only a few large advertisers participate, Google may have no commercially relevant advertisement to display.

With thousands of specialized businesses in the marketplace, there is a greater chance that a suitable supplier is willing to pay for that particular search.

The same effect appears in more competitive categories.

Additional advertisers provide more candidates from which Google's ranking system can select the advertisement likely to produce the most value.

The AOL contract therefore strengthened Google's ability to monetize searches across its entire network.

This was why the founders were prepared to accept an unusually small share of immediate advertising revenue.

The guarantee succeeded, and Google's business changed

The results exceeded AOL's minimum guarantee.

According to the episode, AOL received approximately $35 million during the remaining months of 2002 and approximately $200 million in 2003.

Google had acquired major distribution while substantially weakening Overture's position.

The later 2004 IPO prospectus provides independent evidence of AOL's importance.

It reports that advertising and other revenue generated through AOL accounted for approximately 15% of Google's total revenue in 2002 and 16% in 2003.

That is a substantial concentration in a single commercial relationship.

The contract also created operational demands. Google needed to support thousands of advertisers rather than a relatively small collection of agencies and major customers.

Sandberg helped build the sales and advertiser-support organization, recruiting staff and working with engineers to automate tasks that otherwise would have required rapidly growing teams.

She later carried substantial experience in scaling online advertising businesses to Facebook, where she became chief operating officer.

Google's AOL bet demonstrated something unusual about its business model.

The company could accept low immediate margins on a distribution partner because that partner simultaneously increased the value of the wider advertising marketplace.

10. Google's Financial Advantage Grew With Every New Search

More searches lowered costs and increased advertising opportunities

Most businesses benefit from scale because producing additional units becomes cheaper.

Google developed an additional advantage: the revenue generated by each unit could increase at the same time its cost decreased.

As search traffic expanded, Google spread infrastructure expenses across more queries.

Meanwhile, a larger advertiser base improved the chance of matching each query with a commercially relevant advertisement.

More competition for popular keywords could also increase the price advertisers were willing to pay.

This gave Google two reinforcing financial benefits.

Its cost per search could fall, while its expected advertising revenue per search could rise.

The hosts described this combination as central to Google's exceptional economics.

It also changed the competitive significance of distribution.

A search query that generated limited revenue for a smaller rival could be substantially more valuable to Google because its advertising marketplace offered more potential matches.

Consequently, Google could afford to pay more than competitors for access to the same user.

Google's advertising marketplace developed network effects

A network effect occurs when a product becomes more valuable as additional people or businesses participate.

Google's marketplace developed a particularly interesting version.

Additional search users attracted advertisers because they represented commercial opportunities.

Additional advertisers made the system more effective at finding relevant advertisements for those users.

But advertisers were also competitors.

More advertisers could increase auction prices for businesses bidding on the same keywords.

That made Google's network effects different from a conventional social network, where additional participants might directly increase the value of communicating with one another.

Google's marketplace became more commercially useful as it grew, even though advertisers could face greater competition.

A smaller search engine attempting to reproduce the same advertising system encountered a difficult starting problem.

Without sufficient traffic, advertisers were reluctant to participate. Without enough advertisers, the company could not generate comparable revenue or serve similarly relevant commercial results.

Google had already developed both sides of the marketplace through its search popularity and large distribution partnerships.

Product quality alone could not preserve the advantage

The hosts identified scale economies and advertising-marketplace effects as Google's most important durable competitive advantages.

They placed less emphasis on traditional switching costs.

In Google's early years, users generally did not need an account and could try another search engine whenever they wanted.

The company developed a powerful consumer brand and became highly attractive to engineers, but that did not necessarily allow it to charge advertisers a conventional brand premium.

Advertisers generally evaluated search campaigns according to performance.

The hosts also found less evidence that unique physical resources or organizational practices alone explained Google's economic position.

The stronger explanation was the interaction of technology, marketplace size, and commercial distribution.

Google's search quality attracted users, but its superior monetization helped prevent competitors from acquiring those users away from it.

That became increasingly important as Microsoft and Yahoo developed their own competitive responses.

11. Google Toolbar Turned Distribution Into a Competitive Weapon

Microsoft controlled the browser through which people searched

In the early 2000s, Internet Explorer occupied a powerful position.

Microsoft controlled the browser used by a large proportion of internet users.

If Microsoft made its own search service the easiest option to access, Google could lose traffic regardless of whether its search results were better.

Google therefore had to develop methods of placing its search engine directly into the software people already used.

In December 2000, it introduced Google Toolbar for Internet Explorer.

The toolbar allowed people to search Google without visiting its homepage first.

This reduced friction and encouraged more frequent searches.

According to the episode, toolbar users performed approximately seven times as many searches as ordinary Google users.

That made them far more commercially valuable.

The hosts estimated that ordinary users initially generated approximately $2 annually, while toolbar users could produce more than $10.

These figures helped explain why Google was willing to spend money acquiring toolbar installations.

Google paid other companies to distribute its search box

The company established distribution arrangements with software developers whose applications already reached large numbers of computers.

Adobe, RealNetworks, and WinZip were among the partners associated with toolbar distribution.

Their installers could introduce Google Toolbar to users who had downloaded entirely different software.

Some of these arrangements resembled the aggressive bundled-software practices commonly criticized as adware. Users did not always deliberately seek out an additional toolbar.

Google nevertheless made the product useful enough to encourage adoption.

It added pop-up blocking, a valuable feature when intrusive web advertising was widespread.

The strategy extended to computer manufacturers.

Dell agreed to preinstall Google Toolbar on new Windows computers, giving users immediate access to Google search through Internet Explorer.

Google also became Firefox's default search provider as Mozilla's browser gained popularity.

Payments from Google subsequently became an important part of Mozilla's revenue.

Google Earth and desktop search also supported distribution

Google's software strategy extended beyond the toolbar itself.

Google Desktop allowed people to search files and information stored on their own computers, applying ideas related to the company's original enterprise-search ambitions.

Google Earth offered an entirely different experience: exploring geographic information through an interactive globe.

Rather than immediately forcing advertising into every new consumer product, Google could use applications to strengthen its brand, expand distribution, or encourage use of its profitable search services.

Toolbar installations bundled with other software, including Google Earth, created additional routes to search traffic.

Sundar Pichai joined Google in 2004 and later took responsibility for a client-applications group that included Google Toolbar.

His work placed him near one of Google's most important commercial priorities long before he became its CEO.

Search defaults eventually became an enormous business

Google's willingness to pay for distribution increased as its revenue per search improved.

The company sometimes offered partners nearly all advertising revenue associated with their traffic. The hosts also describe instances where Google reportedly accepted arrangements that could exceed the immediate revenue generated.

The underlying expectation was that a user acquired today would become more valuable over time.

More searches meant more advertising opportunities, greater revenue, and stronger user habits.

The strategy eventually expanded into substantial payments to browser developers, device manufacturers, and platform operators.

Apple became one of the most important recipients.

Evidence presented in the later United States search antitrust case showed that Google's default-search payments to Apple had reached approximately $20 billion annually.

That development revealed the full commercial significance of the distribution model Google had begun building with Toolbar.

It also introduced a regulatory problem: agreements that helped Google defend its market share could simultaneously make it difficult for other search companies to obtain users.

12. AdSense Took Google's Advertising Marketplace Beyond Search

Contextual advertising emerged from an email experiment

Google's search advertisements matched commercial offers to explicit user queries.

But people revealed useful interests in other ways.

Someone reading an article about gardening might be interested in gardening equipment. A webpage discussing mortgages might attract visitors considering financial services.

Google realized it could analyze the subject of a webpage and display advertisements related to that subject, even when the visitor had not entered a search query.

This idea developed partly from Paul Buchheit's experiments while building Gmail.

Buchheit had been working on an email system that used Google's search technology to make messages easy to find.

He tested whether the company's advertising system could identify relevant advertisements from email content.

Page and Brin recognized that similar technology might work on ordinary websites.

Google already had advertisers and the systems needed to select and deliver advertisements.

It could extend those capabilities beyond its own search pages.

Jeff Dean developed an early system in six weeks

According to the episode, Jeff Dean built an initial contextual-advertising implementation in approximately six weeks during early 2003.

Google experimented with advertisements on Google Groups and third-party websites.

It initially bought conventional advertising placements and filled them with groups of relevant text advertisements drawn from its existing advertiser base.

HowStuffWorks provided a particularly useful testing environment.

Its articles explained subjects that frequently had commercial connections, creating opportunities to match specialized content with relevant advertisements.

Susan Wojcicki became involved in developing the product.

Google called the advertising network AdSense.

The name eventually covered two related activities: advertisements displayed alongside search results on partner websites and advertisements displayed according to the content of publishers' pages.

These are important distinctions because not all AdSense revenue came from contextual advertising.

Publishers gained access to an advertising business they could not build themselves

Before automated advertising networks became widespread, a small website faced substantial difficulties selling advertisements.

It needed to find advertisers, negotiate prices, place advertisements, track activity, and collect payments.

Even publishers with sales teams struggled to monetize specialized content.

AdSense simplified this process.

A publisher could register, place Google's code on its website, and receive advertisements selected from Google's existing marketplace.

Google managed advertiser acquisition, targeting, payments, and much of the associated technology.

Publishers received a share of advertising revenue.

The hosts describe historical publisher shares generally ranging from approximately 67% to 80%, depending on arrangements.

This made AdSense substantially different from advertising on Google's own search pages.

Google retained most revenue from its first-party search advertisements, but partner websites received much of the income generated on their own properties.

Google's IPO filing shows how important the network became

AdSense developed into a substantial business quickly.

The episode reports that by late 2003 it was generating more than $1 million daily.

The hosts also estimate that AdSense contributed approximately half a billion dollars of Google's 2003 revenue.

Google's 2004 SEC filing provides more detailed evidence.

Advertising delivered through Google Network members accounted for approximately 43% of total company revenue in 2003 and approximately 50% in the first half of 2004.

Those figures include partner search advertising as well as contextual advertising, so they should not be attributed entirely to the newly launched content product.

The filing also confirms that AdSense for search launched in the first quarter of 2002, while AdSense for content followed in the first quarter of 2003.

Google acknowledged the financial trade-off.

Network advertising expanded revenue and advertiser reach, but sharing fees with publishers reduced operating margins compared with advertising on Google's own properties.

Its filing identified this difference as a significant consideration for future profitability.

Google had created a way for independent publishers to participate in a large advertising marketplace without developing their own advertising infrastructure.

That arrangement later influenced monetization models associated with services such as YouTube, where creators supplied content and the platform supplied advertisers, distribution, and payments.

AdSense's early benefits to publishers also contrast with Google's later disputes over its influence in digital advertising markets.

Both developments belong to the same business history.

13. Google Expanded Internationally Before Its Rivals Could Catch Up

Half the searches were already coming from overseas

Eric Schmidt discovered that approximately half of Google's searches originated outside the United States.

But the company's advertising-sales organization remained heavily concentrated in the American market.

This created a substantial gap between international product usage and international revenue.

Google already possessed users in many countries without having established comparable local commercial operations.

Schmidt instructed Omid Kordestani to expand internationally.

Kordestani traveled extensively, helping establish small teams responsible for developing local advertiser relationships.

The strategy did not require recreating Google's search engine independently in each country.

The underlying technology was already available.

Google needed to build advertiser participation, customer support, and local commercial capabilities around it.

The hosts cite international revenue contributions of approximately 18% during the early expansion period, 22% in 2002, and 29% in 2003, eventually approaching half of the business.

Google's early financial disclosures confirm that international revenue became a growing component of its operations.

China demonstrated the limits of global expansion

Google's international strategy was not equally successful everywhere.

China presented regulatory and political conditions substantially different from those in markets where Google's search service operated more freely.

The company provided mainland Chinese search services for several years but withdrew its mainland search operation in 2010 amid disputes involving censorship and security.

This episode in Google's history illustrates that technical superiority and advertising economics were not sufficient to guarantee worldwide market access.

Government rules, political conditions, and local competitors could constrain the global application of its business model.

14. Google Continued Improving Search After PageRank

A search engine could not remain static

PageRank established Google's early reputation, but the algorithm did not remain unchanged.

The web continued expanding, websites changed, and users developed new expectations.

Google added increasing numbers of signals to evaluate relevance, quality, and the meaning of queries.

By 2007, the company was reportedly considering approximately 200 ranking signals.

The number illustrates how far search had progressed from relying primarily on links and keyword matches.

Google also learned from patterns of user behavior.

When people repeatedly reformulated searches, those changes could help identify relationships between words and concepts.

For example, repeated searches involving "cat" and "kitten" could help a system recognize their relationship.

Patterns involving visits to search results and subsequent returns could also supply information about whether a result appeared helpful, although such behavioral signals require careful interpretation.

Search history and accounts created opportunities for personalization

In the early years, Google's lack of mandatory accounts meant many visitors were effectively anonymous to the search engine.

It had limited ability to customize results based on an individual's interests or previous activity.

That changed as Google introduced more account-based products.

Search history and other user information could provide additional context for personalization.

This increased the potential value of staying within Google's services, although users continued to have alternatives.

Google's engineering approach also remained iterative.

Rather than treating its original ranking system as a completed invention, the company repeatedly invested in crawling, indexing, natural-language understanding, fresh information, and increasingly sophisticated ranking techniques.

Google's current documentation on search ranking systems confirms that PageRank remains part of a much broader collection of technologies.

It also documents later improvements addressing link spam, content quality, freshness, and language understanding.

PageRank started the company, but continual improvements prevented the original invention from becoming its only technical advantage.

15. Gmail and Google News Extended the Company's Information Mission

Gmail began as a search engine for personal messages

Paul Buchheit started developing Gmail around 2001.

Its earliest version was essentially a search system pointed at a user's Unix email directory.

Buchheit wanted to find information buried in messages without having to manage email through cumbersome folders and storage limitations.

Search was therefore central to Gmail from the beginning, not a feature added afterward.

The project developed into a web-based email application that allowed people to access and search their messages across different computers.

Google launched Gmail on April 1, 2004, the same month it filed for its IPO.

Its most attention-grabbing feature was one gigabyte of free storage.

At the time, competing webmail providers commonly offered storage measured in megabytes. The announcement seemed so extravagant that some observers believed it was an April Fools' joke.

Google's own account of Gmail's history confirms its launch date, unusually large storage allowance, searchable messages, and conversation-based interface.

Buchheit was also associated with originating Google's "Don't be evil" phrase, an expression that became closely connected with the company's early culture.

Gmail offered Google another important business advantage.

People who established accounts and regularly accessed their messages had a continuing relationship with the company.

That created opportunities for personalization and greater loyalty across Google services.

The product was independently useful, but it also made Google's broader information ecosystem more attractive.

Google News created a different relationship with publishers

Google News emerged from engineers experimenting with how to organize constantly changing information.

The demand for rapidly updated coverage became especially apparent following the September 11, 2001 attacks.

Rather than maintaining a conventional editorial newsroom, Google developed technology to collect and organize reporting from multiple publishers.

The service made it easier to discover news across different sources.

However, it also changed Google's relationship with professional journalism.

Google could become an important gateway to publishers' work without producing the original reporting.

That relationship later contributed to disagreements concerning traffic, compensation, and the use of news content.

Google's role was becoming broader than indexing websites. It was increasingly influencing how audiences discovered and accessed information produced by others.

16. Google's Revenue Grew Fivefold in a Year

The advertising breakthrough is visible in Google's financial statements

Google's growth between 2001 and 2003 provides unusually clear evidence of what happened when search technology found an effective commercial model.

Its 2004 IPO prospectus filed with the SEC reports the following audited figures:

Revenue increased more than fivefold between 2001 and 2002, followed by another increase of more than threefold in 2003.

These were not merely increases in search traffic.

They reflected Google's ability to convert traffic into advertising revenue through an increasingly efficient marketplace.

The growth allowed Google to finance engineering, infrastructure, international expansion, and distribution without repeatedly raising venture capital.

The company that had struggled to buy computers was becoming capable of funding large technology projects from its own operations.

Advertising had replaced the original business plan

Google's initial plan had emphasized selling enterprise-search software.

By the time it prepared for its IPO, advertising dominated revenue.

Its prospectus reports that advertising represented approximately 77% of total revenue in 2001, 94% in 2002, and 97% in 2003.

By the first half of 2004, that figure had reached approximately 98%.

Enterprise search and technology licensing still existed, but they were no longer the central economic opportunity.

Google had originally expected companies to pay for searching their internal information.

Instead, advertisers paid Google to reach people searching the public internet.

That change also altered Google's relationship with partners.

Initially, Google charged websites such as Yahoo for access to organic search technology.

Later, it could offer search results and advertising together, sharing commercial revenue with partners who delivered traffic.

The advertisements could be worth substantially more than a conventional search-technology licensing fee.

AdSense growth introduced a margin trade-off

Google's financial results also reveal why consolidated margins did not simply rise with revenue.

Advertising on Google's own search pages was extremely profitable.

Advertising through partner websites required payments to those partners.

As AdSense expanded, a larger proportion of revenue came from lower-margin network advertising.

Google explicitly acknowledged this in its IPO prospectus.

The company also reported that growth in paid clicks, rather than simply increasing prices, was an important driver of revenue expansion.

That distinction matters.

Google's business was not growing solely by charging existing advertisers more.

It was processing more commercial activity, attracting more participants, and expanding the amount of internet traffic it could monetize.

17. Google's 2004 IPO Was Designed to Preserve Founder Control

Google did not need the IPO to survive

By 2004, Google was profitable and generating substantial cash.

Its leadership did not have the same financing needs that had motivated the 1999 venture round.

Page and Brin also had reasons to avoid public disclosure.

They feared that revealing the profitability of search could encourage Microsoft and other competitors to invest more aggressively.

But Google was approaching an important shareholder-reporting threshold.

Under the securities rules then applicable, companies with sufficiently large numbers of shareholders could become subject to public reporting requirements even without a conventional IPO.

Google had also distributed equity to numerous employees.

Venture investors and employees had understandable reasons to want opportunities to sell shares.

The company therefore prepared to go public.

According to the episode, the IPO ultimately helped make approximately half of Google's 2,000 employees millionaires.

The founders rejected a conventional share offering

Page and Brin disliked the usual IPO process.

In a traditional offering, investment banks allocate shares to selected investors at an established price.

If the stock trades substantially higher on its first day, investors who received the original allocation benefit from that immediate increase.

The founders believed this could mean the company had sold shares for less than the market was willing to pay.

Google investigated a Dutch auction, an approach championed by investment banker Bill Hambrecht.

Investors would submit bids, helping determine an offering price through market demand rather than relying solely on conventional allocations.

The concept appealed to Google's founders because they had already seen the advantages of auctions through AdWords.

Google's engineers even helped develop software for the offering process.

Schmidt later recalled receiving a letter from an elderly woman who wanted the opportunity to buy Google shares, reinforcing the founders' interest in making participation more accessible.

Dual-class shares protected Page and Brin

The IPO also required decisions about corporate control.

Selling shares to public investors could dilute the founders' influence over major business decisions.

Google adopted a dual-class structure.

Class A shares carried one vote each, while Class B shares carried ten votes each.

This allowed Page and Brin to retain substantial voting power without retaining an equivalent percentage of the company's economic ownership.

The founders drew inspiration partly from media companies whose founding families used different share classes to protect editorial and organizational independence.

The arrangement subsequently became influential across technology.

Companies including Meta, Alibaba, Shopify, Spotify, Airbnb, and Coinbase later adopted forms of enhanced founder voting control.

The structures varied, but the underlying purpose was similar: separating ownership economics from decision-making authority.

Google's founders believed this would allow them to pursue long-term projects without being forced to satisfy every short-term preference of public-market investors.

The auction did not deliver the expected pricing result

Google's IPO occurred in August 2004.

An earlier expected price range of $108–$135 per share was reduced before the offering.

The final price was $85.

The offering sold approximately 19.6 million shares at that price, for total gross proceeds of roughly $1.67 billion.

The SEC's subsequent disclosure clarifies an important distinction: approximately 14.1 million shares were issued by Google itself, while roughly 5.5 million were sold by existing shareholders.

Consequently, Google directly received gross proceeds of approximately $1.2 billion before offering expenses, rather than the entire $1.67 billion.

The IPO valued Google at approximately $23 billion.

Despite the Dutch auction, the stock closed near $100 on its first trading day, approximately 18% above the offering price.

Over roughly the next 16 months, its value increased almost fivefold.

The founders' auction therefore did not eliminate the immediate price increase they had hoped to prevent.

The two distinctive features of the IPO had different legacies.

Dual-class shares became common among major technology companies.

The Dutch auction did not become the standard approach for large technology offerings.

The founders publicly warned investors that Google would remain unconventional

Google's IPO documents included an unusual explanation of the founders' philosophy.

In the shareholders' letter, Page wrote:

“Google is not a conventional company. We do not intend to become one.”

The statement appeared in Google's IPO registration materials.

The letter explained the founders' intention to retain substantial independence, make long-term investments, and avoid managing the company primarily around predictable quarterly earnings.

Google's unusual structure was not merely cultural decoration.

It was an attempt to preserve the decision-making freedom that had allowed management to accept unconventional technical and commercial risks.

That freedom had contributed to Google's success.

It also concentrated power in the hands of its founders.

18. Google's Dominance Created a New Kind of Competitive Problem

The company became difficult to challenge without becoming impossible to imitate

By the time Google was publicly traded, its competitive position involved much more than PageRank.

The company had developed:

• A search engine with substantial relevance, speed, and index-coverage advantages.
• A distributed computing infrastructure capable of reducing processing costs.
• An advertising marketplace that became more valuable as participation increased.
• Relationships that supplied both revenue and access to users.
• A recognizable brand associated with search quality.
• The financial resources to invest continuously in products, acquisitions, and distribution.

Each advantage reinforced others.

Better results attracted users, improving Google's ability to acquire advertisers.

More advertisers increased the revenue Google could generate from search.

Higher revenue financed more computing infrastructure, stronger engineering teams, and additional distribution agreements.

Those agreements brought more users to Google's products.

The resulting system made the company increasingly difficult for competitors to challenge.

The hosts argued that Google's long-term economics were particularly powerful because greater scale could lower costs per search while increasing expected revenue per search.

This allowed Google to outspend competitors on distribution without necessarily sacrificing its longer-term returns.

The company's original philosophy had created a contradiction

Google was founded partly because existing portals had reasons to resist better search.

Their businesses depended on page views, advertising impressions, and keeping people within their own services.

Google rejected that approach by prioritizing fast answers and sending users to relevant destinations.

But as Google became dominant, it developed commercial incentives to defend its own traffic and advertising marketplace.

The toolbar and browser-default arrangements illustrated this shift.

What began as methods of making Google easier to access eventually became important mechanisms for preserving its market position.

The question was no longer simply whether Google had the best search engine.

It was also whether competing services could obtain sufficient distribution to challenge it.

The US courts found unlawful monopolization

The antitrust cases brought against Google provide important independent evidence about its later market power.

In August 2024, the US District Court for the District of Columbia ruled that Google had unlawfully maintained monopolies in general search services and general search text advertising.

The court concluded that Google's distribution agreements contributed to maintaining those monopolies.

The Department of Justice's September 2, 2025 announcement described remedies requiring changes to exclusive distribution arrangements, along with access to certain search data and syndication services for competitors.

A final judgment was entered on December 5, 2025, and became effective on February 3, 2026.

It restricted exclusive distribution agreements and included requirements concerning search index data, user-side data, and access to search and text-advertising services.

Google appealed in January 2026, and government plaintiffs filed cross-appeals in February.

The court's May 7, 2026 order documents that procedural status.

As of October 2026, the Justice Department's case docket also showed continuing work on implementing the judgment, including proposed arrangements for search and advertising syndication.

These developments materially deepen the story told in the 2025 Acquired episode.

The distribution mechanisms that helped Google establish market leadership became part of the legal basis for requiring competitive remedies.

Google faced a separate advertising-technology judgment

A second federal antitrust case concerned Google's position in open-web advertising technology.

This involved a different set of markets from general search.

In April 2025, a federal district court in Virginia found that Google had unlawfully monopolized important digital advertising markets.

The case concerned technology used by publishers and advertisers to sell, buy, and manage advertising inventory.

In September 2026, the court ordered remedies intended to improve competition, including requirements affecting interoperability, access to publisher data, and the operation of Google's advertising tools.

The Justice Department's September 16, 2026 announcement described requirements involving Google's AdX exchange, publisher ad-serving technology, and restrictions on preferential treatment involving AdWords.

These are separate legal proceedings from the search case.

Together, however, they demonstrate how Google's advertising marketplace and distribution infrastructure became central subjects of competition law.

They also complicate a purely positive interpretation of Google's early network effects.

A marketplace can become more efficient as it grows while also creating barriers that make entry difficult for rivals.

The economic mechanism is the same. Whether the resulting commercial conduct violates competition law depends on the specific market, agreements, and evidence.

19. Artificial Intelligence Raises the Question Google Once Asked Yahoo

Google's profitability did not remove investors' concerns

When Gilbert and Rosenthal recorded their episode in 2025, Alphabet's market capitalization was approximately $2.1 trillion.

That was roughly 100 times Google's valuation at its 2004 IPO.

The hosts described Alphabet as generating more profit than other major American companies, including several of the largest technology and financial businesses.

Their specific profitability comparison reflected the financial period available when they recorded the episode and should not be treated as a permanent ranking.

Despite its substantial earnings, Alphabet traded at lower valuation multiples than several major technology peers.

The hosts cited a price-to-earnings ratio of approximately 20 and a price-to-sales ratio around six.

Their comparable earnings multiples were approximately 35 for Amazon, 37 for Microsoft, 46 for Nvidia, 30 for Apple, and 27 for Meta.

These were 2025 comparisons, not current valuations.

They illustrated an important concern among investors.

Google's existing business was highly profitable, but artificial intelligence might change the way people accessed information.

The original search model depended on people entering queries, receiving results, and sometimes clicking advertisements.

AI assistants could potentially answer questions directly or complete tasks without sending users through conventional lists of search results.

The commercial consequences remained uncertain.

Google's researchers helped develop the technology challenging search

The situation was particularly unusual because Google had contributed substantially to modern artificial intelligence research.

In 2017, researchers including members of Google published Attention Is All You Need.

The paper introduced the transformer architecture, which became foundational to modern large language models.

This was not simply an example of a competitor developing technology that Google had ignored.

Google had helped establish an important part of the technical foundation for the new industry.

That created a distinction between technological leadership and commercial leadership.

A company could help invent a technology without necessarily becoming the dominant provider of every product built from it.

Google's earlier history contained similar examples.

Bill Gross developed crucial search-advertising ideas, but Google established the more successful commercial marketplace.

AltaVista pioneered large-scale web crawling, but Google developed a more effective combination of ranking and computing.

The question for artificial intelligence was whether the companies building the strongest new consumer and enterprise products would also develop the distribution and economic advantages needed to lead the industry.

Could AI assistants adopt an advertising marketplace?

The hosts also considered whether the history of paid search offered clues about AI monetization.

In the 1990s, selling search positions was controversial.

People worried that commercial payments would undermine the reliability of search results.

Yet a model combining labeled advertisements, auctions, and relevance signals eventually financed much of the search industry.

AI assistants might encounter their own version of this problem.

An assistant responding to a request for a hotel, restaurant, product, or service could potentially connect users with businesses willing to pay for commercially valuable referrals.

The hosts discussed commercially influenced chatbot responses as a hypothetical possibility, not an established outcome or recommendation.

Such a model would raise substantial questions about disclosure, trust, commercial influence, and whether paid recommendations were distinguishable from independent answers.

Google's history demonstrates that objections to an advertising model do not necessarily prevent its eventual adoption.

It does not establish that advertising will become the dominant business model for AI assistants, or that any particular method of integrating advertisements would be appropriate.

By 2026, Google's search business was still growing

Subsequent financial results provide an important qualification to the fears discussed in the 2025 episode.

Alphabet's second-quarter 2026 earnings announcement, published on July 22, 2026, reported consolidated quarterly revenue of $119.8 billion, up 24% year over year.

Google Services revenue rose 15% to $94.5 billion.

Within that business, Google Search & other revenue increased 17% year over year.

Google Cloud revenue increased 82% to $24.8 billion.

These figures do not establish that Google's search business is permanently protected from AI-driven changes.

But they demonstrate that, as of the second quarter of 2026, growing adoption of AI had not translated into a collapse in Google's reported search revenue.

Alphabet was also expanding its cloud business substantially, providing another source of growth connected to AI infrastructure and enterprise services.

The longer-term competitive question remained open.

20. How Google's Original Advantages Became a Self-Reinforcing Business

Google's history contains several decisions that appeared reasonable to individual companies but ultimately helped create a formidable competitor.

Excite and Yahoo reportedly declined early opportunities to acquire its ranking technology.

GoTo rejected a proposed acquisition that could have combined organic search and paid advertising before Google built its own system.

Yahoo later helped Google survive the dot-com collapse by becoming a major search customer.

AOL accepted a generous commercial agreement that supplied Google with users and advertisers, while Google assumed financial risks that could have threatened its survival.

Google then used the resulting advertising revenue to acquire still more traffic through software bundles, toolbars, browsers, and device manufacturers.

None of those decisions independently explains the company's ultimate size.

Their interaction does.

Google's original search breakthrough attracted users. Its infrastructure made serving those users inexpensive. Its advertising system extracted commercial value from their searches. Its growing marketplace improved advertising relevance and potential revenue. That revenue financed additional distribution and product development.

The company also remained willing to adopt ideas developed elsewhere.

Academic citation analysis informed PageRank. GoTo demonstrated paid-search auctions. Established auction theory informed AdWords. Distributed computing research and Google's own engineering innovations helped make the system scalable.

The resulting business was not simply the product of one original algorithm.

It was the product of a technology company that repeatedly improved how it organized information, reduced costs, attracted commercial demand, and reached users.

That history also explains the questions Google faces in 2026.

The company built its dominance by entering a market where established businesses had strong reasons to protect their existing models. Its own search and advertising businesses later became valuable enough to create similar incentives.

Federal courts have now ordered changes to important aspects of Google's conduct, while artificial intelligence is introducing new ways for people to find and use information.

The same distinction that mattered in Google's early years matters again: inventing a useful technology, building a superior product, and establishing the strongest commercial business are three different achievements.

Google became exceptional by accomplishing all three in search. Whether it can repeat that combination as AI changes the internet remains an unresolved business question.