First, Google adopted advertising auctions, an idea proposed and implemented by GoTo, an early Google competitor run by the entrepreneur Bill Gross.9 Buyers of advertising would bid against one another for the rights to show their message when a user searched for a term like “insurance,” with the winner paying a penny more than the second-highest bidder (a “Vickery second-bid auction” in economics parlance).10 This maximized the amount paid for the advertising inventory available with each search result.
Second, Google incorporated an algorithm that assessed the quality of an ad by looking at its various elements and predicting whether users were likely to click on it. This metric would then influence the value of the bids made by various advertisers, sometimes leading a bidder offering a lower price to win the auction based on the quality of the ad. This system was intended to align incentives such that advertisers would create ads that were relevant and useful to Google users.
The combination of auctions and adjustment via a quality metric was a huge success. AdWords transformed Google from a speculative venture to one of the most profitable enterprises of the last century. From then on, there was, in the words of the journalist Steven Levy, “nothing but glory in the bottom line.”11 Other products followed, built on the infrastructure developed for AdWords. Google AdSense, which allows publishers to earn money by offering space on their websites that the platform sells to advertisers, launched soon afterward and was similarly successful.
The success of AdWords and AdSense depended on a major technical feat, which the engineering-centered culture of Google was well positioned to achieve. Coordinating and resolving a colossal number of simultaneous auctions required a high capacity to deploy reliable and robust computer systems. The development of this infrastructure was led not by an MBA but by Eric Veach, a Stanford computer science graduate whose previous work involved developing the software that rendered animations at Pixar.12 In fact, the debut of AdWords was viewed with consternation by the existing sales team at Google. The replacement of their more traditional approach with an algorithmically driven, automated auction was seen as confirmation of their long-held suspicion that Google’s cofounder, Larry Page, “wanted to do away with [the sales team] entirely.”13
Products like AdWords originated within a profoundly technical culture and were made possible by software engineers. But quite a different set of experts would take those innovations and mold them into the present-day programmatic advertising infrastructure. Google began enlisting economists to manage these vast marketplaces; the new team members viewed the algorithmic exchange for packaging and selling attention as deeply akin to the financial markets.
Hal Varian, who joined Google in 2002 and who would eventually play a role as the company’s chief economist, became the “godfather to the advertising effort.”14 He would oversee a project to create a “search-word advertising equivalent of the stock market” that would divide keywords into high-cap, mid-cap, and low-cap segments and generate its own version of the Dow Jones Industrial Average.15 AdWords and AdSense changed the paradigm for distributing advertising inventory, moving it from a world of qualitative judgment and persuasive salespeople to one of quantitative analysis and automated algorithms. As Steven Levy would recount in his 2011 history of Google, “Varian realized that [Google] was the embodiment of the Silicon Valley ethic … Most Internet companies were still selling ads the way Madison Avenue had always done it. Google saw the entire exchange differently.”16
This adoption of the financial markets as a metaphor for the online advertising markets was not a phenomenon limited to Google, or even to the domain of search advertising. As the early templates of programmatic advertising pioneered at Google became more popular as a model for buying and selling attention, many continued to draw inspiration from the financial markets. One Wall Street Journal feature reporting on the nascent platforms that would evolve into modern programmatic ad exchanges remarked that these tools allowed “selling web advertising space like pork bellies.”17 The New York Times observed that “quants have become a force in the advertising industry, much as they became a force on Wall Street starting in the 1970s and 80s.”18 As the hype grew around these platforms, the stock exchange was a frequent point of reference in explaining the distinctiveness of open marketplaces for ads.19
DoubleClick, founded in 1996 by Kevin O’Connor and Dwight Merriman, was a major player in the nascent business, helping advertisers place ads across various publishers online and measuring their performance. DoubleClick survived the 2001 dot-com crash and in 2007 launched one of the first exchanges allowing publishers and advertisers to buy and sell inventory. The New York Times described the platform as a “Nasdaq-like exchange.” DoubleClick representatives explained it as a “mix of eBay and Sabre, the airline reservations system that travel agents use. The service … let advertisers see information about what competitors bid for particular ads … [and] let publishers try to ensure that they sell their ad spots at the highest possible price.”20 DoubleClick was acquired by Google in 2007 for $3.1 billion.21
Other players in this primordial phase of the online advertising market would adopt similar framings, sometimes even boasting previous careers in finance. Right Media, another early ad exchange pioneer acquired by Yahoo in 2007 for $680 million, was described by The Wall Street Journal as a virtual markets “where buyers and sellers of Internet ads can efficiently find one another … not unlike the role stock exchanges play for shares.”22 The founder and CEO, Michael Walrath, had worked at the brokerage firm D. H. Blair and the hedge fund Sands Brothers Asset Management.23 Walrath cautioned against assuming a strict similarity, but he recognized that his platform was “loosely modeled on the stock market.”24 AdECN, another early competitor in this space, explicitly adopted the stock market model in much
