Google Radio Automation and the broadcast plights behind platformization

Abstract

In 2006, Google entered the broadcast radio industry in an attempt to prove it could diversify its ad sales technology beyond the web. With a product called Google Radio Automation, the company simultaneously installed its ad tech and ingratiated itself with radio producers. Histories of automation and programmatic advertising in American radio help to contextualize Google’s project and trace the roots of its failure. Publicity materials, industry reporting, and interviews with engineers and managers illuminate Google’s efforts toward cultural and industrial integration with radio. Problems related more to consolidation than to technological change thwarted those efforts and saw Google abruptly shift its attention to streaming audio in 2009. Revisiting a time when platform companies sought to infiltrate rather than replace older media, this article proposes that the turn toward a new media paradigm had more to do with old media plights than with the insight of technology firms.

Between 2006 and 2009, Google was a major player in the American radio industry. In the same year that the ascendant web search and advertising giant acquired YouTube, it also bought a company called dMarc Broadcasting. dMarc had recently purchased Scott Studios, the United States’ leading vendor of software and hardware for automating broadcast radio programming. The Scott Studios products became Google Radio Automation, a system that merged Google’s ad sales technology into the core infrastructure of American commercial radio through installations at AM and FM stations around the United States and in Mexico. But, as Google soon experienced, technological integration opens a conduit for existing cultural and industrial problems that may escape even a powerful actor’s control. In this article, I show how Google’s failure to anticipate technocultural frictions, including distinct troubles for radio in the wake of the 1996 Telecommunications Act but also new ones arising at the seam between Silicon Valley and radio’s production cultures, doomed its efforts to enter radio and took part in a significant change of strategy for the company. Along with parallel ventures in print and TV advertising, Google pulled the plug on its radio project by the end of 2009 amid the worsening global recession. It quickly redirected its attention from radio to streaming music, reproducing within an Internet platform the medium it had failed to infiltrate.

Google leadership, in particular then-CEO Eric Schmidt, pinned the retreat on technological problems inherent to audio media. Jessica Vascellaro (2009) reported for the Wall Street Journal that, as Schmidt saw it, “the radio effort failed because Google never came up with a good way to measure listener response.” Schmidt was quoted, “With an enormous data corpus, our computers can do the math really well . . . But in the audio case, there wasn’t a good signal back to us about which ads performed.” Schmidt’s explanation is curious in that it would imply a surprising degree of technical naivety on Google’s part. That AM/FM transmission is a one-way signal with no direct way to track its reception is not just a fundamental difference between older broadcast media and the Internet; it has also motivated the existence of an audience measurement industry that has been an economic and media-shaping force in its own right since the 1920s (Lotz, 2014; Razlogova, 2011). But Schmidt’s narrative contributed to a larger frame that platform firms were constructing: one where problems are chiefly technical in nature and can best be addressed by technological, especially data-driven, interventions. An admission of failure in this one case could still bolster the company’s effort to cast itself as a de facto leader in media by insisting on media’s inevitable datafication (Van Dijck, 2014).

Vascellaro’s reporting pointed to a likelier explanation for the failure, one in which a number of factors had contributed to a lack of buy-in from station owners and from advertisers, resulting in a feedback loop of low pricing in Google’s ad auctioning system and declining interest from the stations who might use it to sell ad time. Those factors included primarily industrial rather than technical differences between the media Google tried to straddle:

[M]edia-buying agencies, fearing Google’s technology would put them out of business, were a tough audience. Google refused to create bundles of spots and negotiate prices ahead of time, which was how radio was generally sold, say people familiar with the discussions. Agencies also wanted more control over what stations their ads would play on. Google initially argued that it was enough to let advertisers target, among other things, station types and specific geographic regions delineated by radio-ratings service Arbitron Inc. Google ultimately agreed to allow advertisers to exclude certain stations by emailing a representative at Google (Vascellaro, 2009).

The problems that sank the first version of Google’s Audio Ads project, under this account, called up anxieties about agency, targeting, and automatic control that had in fact been brewing in radio for most of a century. Google Radio Automation was the newest and most sophisticated entrant in a long succession of automation products for radio broadcasting, one that began in the 1950s with simple techniques for cuing tape reel players to start and stop (Stuhl, 2022). As Lee McGuigan (2019) has shown, programmatic advertising—the technique that Google and similar firms claimed to invent afresh under a big data paradigm—had likewise taken hold in American broadcast media by 1957. Despite steadily displacing and consolidating the labor of broadcast technicians and DJs across that evolution, radio automation had become a widely accepted presence in commercial radio, as had the practices central to ad tech.

Although Google had acquired Scott Studios’ radio automation products somewhat incidentally to its pursuit of dMarc’s radio ad tech, the automation system’s settled centrality to station operations made it a promising “boundary infrastructure” (Bowker and Star, 1999) for the juncture Google hoped to bring about between radio producers and online ad-buyers. Google Radio Automation became face, for the radio community, of a project whose aim was to automate the roles of radio ad salespeople rather than radio producers (the latter’s roles having already been automated about as far as they possibly could). Yet Google failed to convince already automation-friendly station owners that this new ambit for automation would prove valuable to them. Understanding this failure requires a look back at the history of automation in American commercial radio as well as the unique political-economic conditions that surrounded Google’s media ventures.

The brief and thwarted existence of Google Radio Automation presents an opportunity to revisit the particular stakes that attended media change and media divergence at a moment that analysts now recognize as the dawn of the platform era. According to Nick Srnicek (2016: 41):

In the early years of the century it was hardly clear . . . that data would become the raw material to jumpstart a major shift in capitalism. The incipient efforts by Google simply used data to draw advertising revenues away from traditional media outlets like newspapers and television. Google was performing a valuable service in organising the internet, but this was hardly a revolutionary change at an economic level. However, as the internet expanded and firms became dependent on digital communications for all aspects of their business, data became increasingly relevant.

The drive toward datafication was an active rhetorical project for executives like Schmidt. And in contrast to the notion that the Internet expanded steadily and inevitably, which crops up in Srnicek’s description, the migration of media forms into Internet platforms was also a reaction by firms like Google when they failed in an earlier imperative to install datafication within existing media. The New York Times noted in 2007 that Google’s “lofty valuation is partly based on investors’ expectations that it will be able to expand beyond its core Internet advertising business to media like radio, newspapers and even television” (Helft, 2007).

Just 2 years later, with consolidated and debt-burdened media companies flailing under the recession’s impact, Google’s calculus had shifted decisively. The cost and difficulty of proving itself in “traditional” media had overcome the imperative to diversify beyond the Internet. For its audio division, the shift meant the company “decided to exit the broadcast radio business and focus our efforts in online streaming audio,” as Susan Wojcicki (2009) announced in a 12 February 2009 blog post.

This article treats the Google Radio Automation saga as a parable about media infrastructures in the emergent years of a platformized media regime. Jean-Christofe Plantin et al. (2018) have situated infrastructure studies and platform studies amid a period of infrastructural decline in the United States, aptly noting that “[p]latforms rise when infrastructures splinter” (p. 299). Google’s radio venture and its subsequent pivot to the music streaming sector neatly retrace this characterization as well as a more historically narrow tilt from convergence to deconvergence that the researchers in the political economy of communication have noted (Jin, 2012). They also, though, spotlight the powerful yet constrained role that a platform company played in a specific moment where platformization broke away from existing media systems. This article draws on infrastructure studies, media industry studies, and radio history to help explain the technical and cultural factors that first propelled and then thwarted an attempt to integrate radio and Internet economies through their working interiors. It foregrounds automation as a sociotechnically complex presence in media that, in this case, served both as the vehicle for integration and as an avatar of the political-economic forces that would ultimately annul the desire for that integration.

Radio automation, from cue tones to ad tech

Radio automation had roots in the Californian wing of the post-WWII American technology boom, but it would take a long arc to bring broadcast radio and Silicon Valley back to their sudden reconvergence in 2006. Radio automation refers to a set of technologies that help broadcasters sequence sound recordings ahead of time, allowing stations to “stay on the air for hours with virtually no human intervention” (Douglas, 2004: 280). More than a technological breakthrough in its own right, early automation in radio was a side effect of magnetic tape recording, which had seen a major uptick in adoption by the radio industry at the end of the 1940s. Ampex, a young electronics company in San Mateo County, had imported German tape technology to the United States and achieved a significant leap forward in audio quality for applications like pre-recording radio shows (Clark, 1992). And it was Ampex that first, starting in 1953, marketed a device for “automatic programming” whereby inaudible control tones overdubbed onto a recording would trigger tape machines to start and stop (Tinkham, 1953). Within a few years, other magnetic tape applications including computer data storage and videotape had eclipsed the radio market so thoroughly that Ampex abandoned this product. But the technique it had established of using 25-Hz “cue tones” to automatically execute a programmed audio sequence continued on as the basis for radio automation, which a set of vendors with longer ties to radio and entertainment industries (including Gates, Schafer Electronics, and a side venture of the Muzak Corporation) developed further.

As microcomputers and personal computers became available in the 1980s, automation’s established presence in radio—according to Susan Douglas (2004: 280), at least one in seven American FM stations were automated by the end of the 1970s—provided an ideal vector for their adoption in radio; and computerization, in turn, dramatically increased the range of operations that radio automation could provide (Leupold, 1984). Software products retained as design metaphors many of radio automation’s earlier materials like paper program logs and looping tape cartridges, but piece by piece these elements moved from external media into data representations. Likewise, processes that had previously run through separate systems, like music scheduling and traffic (the broadcast industry term for selling, scheduling, and tracking advertisement time), migrated into shared software suites.

By the early 2000s, a company called Scott Studios had emerged as a leading vendor in the US market for radio automation software. Its products included SS32 and Maestro, flagship systems for prerecorded and live in-studio broadcasting, as well as supplementary tools like Voice Tracker that supported remote pre-recording operations. A promotional brochure for SS32 touted: “More stations pick Scott Studios’ air studio systems than the second and third ranked vendors combined. Of the 25 largest broadcasters, 24 use Scott digital audio systems” (Scott Studios, 2004). In keeping with the radio automation sector for most of its history to that point, Scott Studios was a radio company first and a software company second. Founder and president Dave Scott (2021) had worked as a disk jockey, owned stations, and helped develop syndicated programming libraries for automated stations as the CEO of TM/Century prior to helming his own automation software effort.

In October 2004, Scott sold the three companies he now owned—Scott Studios, Computer Concepts, and Scott Concepts—to Chad and Ryan Steelberg, the owners of dMarc Networks. The Steelberg brothers brought these assets together under a new firm called dMarc Broadcasting (Radio World, 2004). A press release from dMarc announced:

The integrated company will boast the largest installed customer base for radio automation and digital systems, with more than 4,600 radio station clients and over 1,800 stations in Arbitron-rated US markets. This represents over 40% of the stations in the top 50 radio groups (dMarc Broadcasting, 2004).

The former dMarc Networks had focused on integrating advertising into a new facet of terrestrial radio: the Radio Data System (RDS), or “radio text,” that enables stations to encode short digital messages into their analog broadcast signals (typically the station name and current song title as displayed on a car stereo console). Pitching radio stations and advertisers on a capability to sell advertisement space within this text layer, dMarc had developed an infrastructure for radio ad sales. Acquiring Scott Studios gave dMarc an inroad, in terms of both technology and product recognition, to the heart of the medium. Analysts explained at the time that dMarc’s purchase of the Scott systems “enabled them to get closer to radio stations at the broadcast point” (Stine, 2006). The automation products additionally served as a financial incentive for stations to give dMarc access to their advertising inventory: under a barter program announced alongside the acquisition, stations could upgrade or obtain new SS32 and Maestro systems in exchange for ad space (Radio World, 2005). While advertising revenue drove dMarc’s efforts forward, the automation products facilitated access and integration crucial to their project.

Google audio ads

In January 2006, Google announced that it would acquire dMarc Broadcasting. The firm’s eagerness to fold radio advertising into their existing AdWords and AdSense services reflected what industry commentators at the time saw as a pressure on Google to prove they could expand their growth beyond online revenue streams into older media industries. Google’s move into radio coincided with a similar venture into print publishing. According to the Wall Street Journal, the firm had been eyeing these other media sectors for some time, with the idea of eventually offering advertisers a centralized dashboard for purchasing ad space across media—and dMarc emerged as an appealing avenue through which to pursue the radio leg of that envisioned platform (Vascellaro, 2009). dMarc had combined industry credibility and an ad sales mechanism through a product (SS32/Maestro) already centrally positioned in many stations, and Google saw an opportunity to harness this combination for the entrance they hoped to make.

To Jon M. Garon (2010), the dMarc purchase fit clearly into Google’s “strategy of acquiring companies to expand beyond ad placement into the role of ad broker and broadcaster”—a strategy specifically to “purchase technology companies that provide toeholds into advertising delivery in each media, allowing it to expand both horizontally and vertically.” In contrast, another purchase later that year seemed riskier and more strategically murky even from a 2010 vantage point: that of YouTube. Noting YouTube’s continued failure to generate a net profit, Garon pointed out that “traditional media” were shrinking but still profitable; and, furthermore, that with Hulu, a web distribution channel that worked within the broadcast TV ecosystem instead of competing with it, Fox and NBC had “already earned greater revenue than YouTube on a fraction of its audience.” YouTube’s apparent value to Google lay in its function as a “beachhead against other participants entering the business” of digital media distribution and in its provision of synergy with the broadcast TV industry—where Google had not quite yet, at the point of Garon’s writing, given up on its advertising efforts. Despite the setback with radio (and print), it still remained apparent that Google “has been motivated by a strategy to control ad placement across all media” (Garon, 2010: 435–7).

Other analysts responded to the dMarc purchase as a promising experiment in media convergence: a “merging of Internet with legacy traditional media” (Stine, 2006). A successful outcome in this experiment, they felt, might trigger a shift in strategy from other tech giants as well as changes within the radio world. Google product manager Josh McFarland, quoted in RadioWorld, said that the firm viewed “radio [as] a very complimentary medium to online” for the purposes of ad sales integration, and that “dMarc is a very good business fit on both the technological level and the team level” (Stine, 2006). With Google bringing on the Steelberg brothers and retaining most of the Scott employees at their existing office in Dallas, the Scott radio automation systems provided continuity for broadcasters as the advertising infrastructure attached to these products underwent an ambitious expansion.

A press release from Google touted the tools dMarc had developed for automatically scheduling advertisements and placing ad audio in the broadcast programs of participating stations; the company planned “to integrate dMarc technology into the Google AdWords platform” and create “a new radio ad distribution channel for Google advertisers” (News from Google, 2006). Automation in broadcast ad scheduling was nothing new. Only a decade or so after American radio had moved from sponsorships to the spot broadcasting model (Russo, 2010), and less than a decade after the term “automation” had been coined, broadcast advertising agencies had endeavored to automate as much of the process of selling and scheduling spot advertisements as they could, following logics of optimization and quantitative analysis similar to those Silicon Valley would tout a half-century later (McGuigan, 2019).

What dMarc had done, beyond the relatively gimmicky exploration of RDS text as advertisement space, was to begin reinventing programmatic radio advertising in a significant time and place for American technology industries: dMarc was, in the words of William “Dub” Irvin (2021)—who worked for Scott and followed the automation products to dMarc, Google, and finally WideOrbit—a “professional company in California . . . a completely different kind of company” from the Dallas-based “mom and pop company” that had produced the radio industry’s leading automation tools. In other words, dMarc initiated a transformation and integration for broadcast radio technology that was as much geographic and organizational as it was technological. For Irvin, who happily accepted dMarc’s invitation to relocate to California, this integration was a welcome development. To other employees and for Google itself, the merger of cultures would soon become more problematic.

By the close of 2006, Google had made their first announcement that a radio option would join services for advertisers. In a 7 December post to the AdWords blog, a member of the new Audio Ads team (quoted as “Josh M.,” likely McFarland) revealed that the group had completed their effort to integrate dMarc’s advertising system into the AdWords platform. The post explained that, while the new capability to purchase radio ads was not yet enabled for general users of the platform, a “U.S. beta test of Google Audio Ads with a small group of AdWords advertisers” was underway. It stressed the ease and efficiency of the integrated system, as well as the provisions for targeting and tracking that Google’s data-focused approach brought to the table:

Google Audio Ads brings efficiency, accountability, and enhanced ROI to radio advertising by providing advertisers with an online interface for creating and launching radio campaigns. You’ll be able to target your customers by location, station type, day of the week, and time of day. After the radio ads are run, you will be able to view online reports that tell you exactly when your ad played (Inside AdWords, 2006).

While posted to the AdWords blog, the message seemed to address advertisers already using radio campaigns, emphasizing the appeal of AdWords in terms of an increased return on investment by way of integration with a maximally datafied system.

Next, Google pitched the broadcasters on Audio Ads. Referring this time to “Google AdSense™ for Audio,” a press release in April 2007 would announce that the system for ad sales and delivery was “now supported by the leading radio station systems”—not only “Google’s own automation systems, SS32 and Maestro,” but also broadcast systems owned by competitors who had agreed to integrate support for Google’s ad services (News from Google, 2007). The wording this time addressed station managers, explaining how the automated capacities of the new integration could provide a non-disruptive benefit to commercial stations:

AdSense for Audio is an automated way for radio stations to supplement their existing revenue streams by making their inventory available to Google advertisers, most new to radio, that aren’t otherwise easily accessible today. By integrating directly with stations’ systems, AdSense for Audio allows ads purchased by Google’s Audio Ads advertisers to be placed directly into the stations’ broadcasts, while at the same time making sure stations retain total control over their inventory through the tools and systems they’re already accustomed to. Radio station system compatibility is built on the AdSense for Audio API (News from Google, 2007).

The automation systems again served as an anchor or assurance for broadcasters, playing down Google’s efforts to destabilize the attached technological and financial arrangements of advertising.

At the same time that these entreaties from Google were nearing a public stage, its project showed what some analysts interpreted as a significant sign of trouble: in February 2007, the Steelberg brothers left Google. Responding to news of the abrupt departure, Google affirmed that they remained “committed to the audio business” and were still pursuing their plans for radio (Ali, 2007). But the news made outside onlookers doubt that the criteria for pay-out targets Google had set in its acquisition of dMarc—an additional US$1.136 billion beyond the US$102 million cash purchase, contingent upon “certain product integration, net revenue and advertising inventory targets” (News from Google, 2006)—had been met. A New York Times story following the Steelbergs’ exit identified frustrations and skepticism toward the Audio Ads project on the part of radio station owners and advertisers. Integration with AdWords meant offering an automated, auction-style purchasing system that would let advertisers purchase time and deliver audio for their ads up until shortly before broadcast. Google insisted that an ability to reserve ad slots further in advance distinguished their offerings from what radio insiders derided as “remnant inventory” (Inside AdWords, 2007a), but the Times piece noted that the head of one partnering station network had in fact used this term to describe the ad space his stations had opened through Google (Helft, 2007).

The box and the boundary

The threat of devalued ad time was not the only thing that made some radio professionals bristle at Google. Rumors had already begun to circulate by late 2006 that a deal between Google and a newly gargantuan radio station conglomerate, Clear Channel Communications, was in the works. Clear Channel’s impact on radio had been so profound as to surpass economic effects on the radio industry and extend to effects on the character of the radio medium, transformations for which the name Clear Channel served as a metonym and in which automation figured prominently. Before digital competitors for listenership emerged as “external threats” to radio, Eric Klinenberg (2007) wrote, American radio had experienced an internal erosion:

Clear Channel had become the industry’s own worst enemy, flooding the airwaves with standardized formats, automated programs, rip-and-read journalism, endless commercials, and a uniform diet of politically partisan, parochial talk shows that dulled local radio and pushed large segments of the audience off the dial (p. 63).

The existing terrain in which Google attempted its project of media convergence had already been warped by ownership convergence within those individual media. This warping had, among other things, turned the settled fact of radio automation into a newly salient point of resentment among many radio professionals. Ironically, the commercial broadcasting lobby had helped produce this situation by using the threat of incursion from digital media to justify their own deregulation. As Patricia Aufderheide (1999: 48–49) explained in her account of the 1996 Telecommunications Act, broadcasters “wanted to abolish concentration of ownership and cross-ownership restrictions, especially for radio” and “argued that radio advertising was underpriced, financially crippling the medium, because of limits to concentration of ownership.” The result of success in this lobbying effort set off a consolidation race that Dal Yong Jin (2012) has described as a cross-media convergence wave rooted in corporate strategy and with implications reaching across technological and economic sectors. By the early 2000s, Clear Channel had developed a substantial lead in this race over other radio conglomerates. In 2002 Clear Channel, which had owned just 39 radio stations in 1995, owned 1205 stations in the United States, gaining a 27% share of national radio listenership and dominance in most major metropolitan broadcast markets (DiCola, 2006).

Under these conditions, it was hardly surprising that insiders would assume (correctly) that Google would court Clear Channel as a partner in its radio venture. But even before Clear Channel’s size in the industry led to uneven terms in their contract with Google versus smaller station owners alienating the latter (Vascellaro, 2009), the prospect of their involvement combined with the material trappings of Google’s project to provide fodder for detractors within the radio industry. A November 2006 blog post quoted the owner of a competing radio advertising company: “Google requires a ‘box’ in the station to deliver commercials. If you were a station manager, would you allow Clear Channel/Google to place equipment into your station?” (Bogatin, 2006). This “box”—the Mk-14 rack-mounted server that hosted Google Radio Automation, updated from its Scott Studios and dMarc incarnations to sport a distinctive green color and a multicolor Google logo—could either embody the company’s success in integrating into radio stations or become the foreign barb that provoked a kind of immune response from the radio industry’s working interior.

According to Amelia Arsenault (2012), the kind of cross-media synergies that Google hoped to achieve depend on “the ability of a corporation to successfully merge cultural customs, machine code, methods of operation, and external network associations across multiple holdings” (p. 111). Google was well-equipped to handle machine code and to scale its modes of operation. It was the first and last items in Arsenault’s list that tested the limits of the firm’s ability and control. External associations would significantly mark the project for broadcasters who resented Clear Channel, with whom Google confirmed an “important strategic relationship” (Inside AdWords, 2007b) on 16 April 2007, the same day it announced the new compatibility arrangements between AdSense for Audio and automation systems. Under the long-sought deal, Google would become the broker for a portion of advertisement time in the schedules of 675 stations—more than half of Clear Channel’s radio holdings—around the country. Having incurred negative but perhaps unavoidable attachments in radio’s existing network of power, Google’s fortunes now rested largely on cultural integration between Silicon Valley and the US radio industry.

The timing of multiple announcements on 16 April was hardly a coincidence. That evening, Eric Schmidt appeared as a keynote speaker for the National Association of Broadcasters’ (NAB) convention. The NAB has been the main trade association and lobbying group for commercial broadcasters in the United States since the 1920s, and in 2007 and 2008 its annual convention would host the most visible efforts by Google toward cultural integration with the radio industry. Speaking immediately on the heels of the Clear Channel news, Schmidt could not avoid addressing this development, which was the subject of his interviewer’s (news anchor John Seigenthaler) first question. Schmidt embraced the deal as a critical success, even calling it “the defining deal for our radio business” (Eric Schmidt at NAB 2007, 2007). But despite this triumphant tone at the outset, Schmidt spent much of his time assuaging the sentiments that the time and place of his appearance presumably provoked—namely, fear and resentment that a rising technology giant had, just a year after deciding to enter the radio industry, inked a deal with its dominant player and been ushered into the brightest spotlight at its annual gathering, from where it might try to dictate how things were going to be from now on in a broadcast industry remade in the Internet’s image. To the contrary, Schmidt insisted (after Seigenthaler helped him transition into the defensive mode by noting that “you’re kind of walking into the lion’s den . . . with broadcasters here today”) that “Google is a new phenomenon. It doesn’t replace radio or television” (Eric Schmidt at NAB 2007, 2007). As the hour-long keynote progressed into questions about Google’s role as a content aggregator in its relationship to broadcast industries (a US$1 billion copyright infringement suit by Viacom against YouTube and Google was also fresh news), Schmidt depicted Google both as a revolutionary user experience and as a narrowly concerned advertising business that would bring non-disruptive benefit to existing media concerns.

Despite the strains that had become apparent with the Steelberg brothers’ exit, Google pushed ahead with the radio advertising project and with the refinement of the automation software that provided its point of contact for radio broadcasters. Behind the scenes, work had already begun under dMarc’s tenure to build a new code base that would underpin the software successor to the SS32 and Maestro automation systems. This process involved collaboration between California-based engineers that dMarc had hired and Scott employees, with Irvin (2021) acting as product manager, and it accelerated under Google’s ownership. By May 2007 the web address google.com/radioautomation displayed information on SS32 and Maestro under the heading “Google provides complete digital audio solutions to radio broadcasters” (Google, 2007). A minimum viable product for the updated system was ready by January 2008, giving Irvin (2021) and his team time to perform a first installation at Nucléo Radio in Monterrey, Mexico ahead of the software’s official launch as Google Radio Automation at the 2008 NAB convention.

Google’s presence at NAB 2008 differed markedly from the previous year. The firm’s entrance in 2007, focusing on Audio Ads, had combined the most and least visible forms of engagement between Schmidt’s keynote and a series of “private meetings in suites with our customers, laying out our vision and getting buy-in”; in 2008, it joined other industry vendors on the trade show floor with an exhibit that was “primarily automation focused” (Irvin, 2021) and that another Google employee who attended described as a “Scott Studios-as-Google booth” (A. Widdowson, 2019, personal communication). In a demonstration for a journalist on the show floor, Irvin focused foremost on the system’s utility from an in-studio broadcaster perspective, only noting toward the end that a major appeal of automation features was to arrange content remotely and in advance and thereby to reduce personnel needs (Aving USA, 2008). The presentation indicated Google’s sensitivity toward the established roles in broadcasting, steering away from a consideration of whether their automated system might replace more of these roles.

The automation software was the operational object through which the union of radio broadcasting and platformized programmatic advertising would or would not play out. It was, in Google’s aspirations, what Susan Leigh Star and Geoffrey Bowker have termed a “boundary infrastructure”:

What we gain with the concept of boundary infrastructure over the more traditional unitary vision of infrastructures is the explicit recognition of the differing constitution of information objects within the diverse communities of practice that share a given infrastructure (Bowker and Star, 1999).

Google Radio Automation acted as a familiar but newly high-tech production tool for a broadcast community that had already accepted automation. The technical integration of Audio Ads and Google Radio Automation depended on a degree of integration between this community and the world of online ad-buyers for whom the same infrastructure would act as a frictionless inlet to the radio medium.

Google’s efforts at NAB and in promotional materials to play up continuity with Scott Studios (an updated web page in 2008 touted first the industry mainstay status of SS32 and Maestro and, second, the cutting-edge technological supports and modular architecture that Google had contributed (Google, 2008)) belied a high degree of turnover in the two acquisitions. Dave Scott had left the radio automation industry soon after the sale to dMarc (PR Newswire, 2005), and a first wave of his employees departed as the Steelberg brothers closed down some projects and pressured developers to work from the California or Dallas offices (Freeman, 2023). Even within California, following Google’s acquisition of dMarc, cultural tensions reportedly flared around the “brash salesman” persona that the Steelbergs cultivated—Chad Steelberg reportedly told stories about “wrestling a shark on the beach in front of his home” that resonated poorly with Silicon Valley engineers (Vascellaro, 2009). Conversely, Google’s preference to hire only graduates from elite colleges foreclosed the usefulness of the Steelbergs’ tactical connections in a radio ad sales world where power was far less correlated to college degrees (Vascellaro, 2009).

Ultimately, the question of whether Google Radio Automation could succeed as a boundary infrastructure was settled in material fact: many stations that had upgraded to the new system, including that first installation in Monterrey, had not activated the integration with Audio Ads by the time Google shuttered the project in early 2009 (Irvin, 2021). Google had turned the automation software asset they acquired in their multi-media ad tech acquisitions into an object that could signal their seriousness about joining the radio industry while acting as a convenient on-ramp to stations who might join in their larger ad sales pursuits. A 2008 iteration of the web page for Google Radio Automation showed off a sleek new interface with Google-style blocks of primary color and touted “the industry’s only third-generation Radio Automation System,” yet still placed greater emphasis on reliability and continuity (Google, 2008). But this object was tainted by radio’s internal politics before its installation began, and only in a minority of instances did both halves of this boundary object actually get to function in tandem (Clear Channel stations used an in-house automation system, so they participated only in Audio Ads (Irvin, 2021)).

Following the Clear Channel deal, station owners found new frustration in the preferential treatment Google had, by necessity of their partner’s outsize market share, offered the conglomerate in terms of inventory control and minimum pricing for auctioned ad spots; and adopters reportedly saw their ad time prices driven further down (Vascellaro, 2009). Google had demonstrated that it could incorporate and upgrade radio technology with the Google Radio Automation project but not that it could produce profits for smaller actors in the industry through Audio Ads. Instead of a high-technology lifeline that could rescue and refresh older media industries, the view of Google from within radio that won out was one where it simply helped accelerate the sinking trajectory the medium had already put itself on.

Platformization as escape hatch

Google’s Clear Channel deal, almost immediately, spelled more top-down troubles in addition to the qualms of broadcast professionals on the ground. Just a week after the two companies announced their partnership, Clear Channel sold off 161 radio stations to Providence Equity as a streamlining measure it had initiated in 2006 after announcing plans to pursue a buy-out from another set of private equity firms (McBride, 2007). That initial buy-out offer had valued Clear Channel at US$26.7 billion, and Providence had agreed to take on the firm’s US$8 billion of debt (Ahrens, 2006). Clear Channel, once “the epitome of the mega-media company made possible by the 1996 Telecommunications Act” (Jin, 2012) began to signal a wave of deconvergence as conglomerates divested many of the assets they had raced to acquire at the end of the 1990s.

When the 2008 recession and global financial crisis hit the economy, it seemed to some analysts that the already troubled outlooks for Clear Channel and for Google’s Audio Ads initiative were doomed (Stine, 2009). In January 2009, Clear Channel laid off around 1500 employees—roughly 7% of its workforce. The layoffs fell primarily on ad sales (McBride, 2009), suggesting an alignment with Google’s aims to automate these roles; or perhaps a desperate optimism that Google’s efforts would succeed.

A blog post on 12 February 2009 to Google’s “Traditional Media Blog” (Pacher, 2008) announced that the company had “decided to exit the broadcast radio business and focus our efforts in online streaming audio” (Wojcicki, 2009). Citing a level of impact that had fallen short of the firm’s expectations for the experiment, the post explained that the exit from radio would entail shutting down the Audio Ads program and seeking a buyer for the automation software. The post did note that Google would continue on with a newer project in ad technology for television; this experiment, though, would come to a similar end in 2012 (Google TV Ads Blog, 2012). In August 2009, the media management software firm WideOrbit purchased the Google Radio Automation assets (WideOrbit, 2009). With a remolding as WideOrbit Automation for Radio (Perro, 2014), this software product completed its journey through a world of new media integration, ending Google’s brief tenure as a radio company.

Radio, according to Andrew Bottomley (2020), had allowed the emerging Internet to host “the first major instance of digital convergence in the contemporary media era” (p. 2) (emphasis original) when college radio station began to transmit their signals as online streams in the 1990s. The recession helped trigger both an industrial and formal swing toward deconvergence and the intensification of rhetoric that distanced “new media” from their predecessors. Google’s pivot to streaming coincided with a growth spurt for the emergent on-demand music service sector: Spotify had launched its beta version in 2007 and would emerge as a majority venture capital owned platform company by 2009 (Eriksson et al., 2019). Google launched its own music streaming service in 2011. Renamed as Google Play Music and revitalized by the acquisition of Songza—an independent player founded in 2010 that had navigated and installed cultural parameters key to a more widespread integration of on-demand music into American life (Baade, 2018)—the service would spend several years competing seriously with offerings from Apple, Amazon, and Spotify before Google migrated it into YouTube in 2020.

The 2006 YouTube acquisition, of course, had set in motion another possible future for Google’s participation in media industries. This vision, wherein user-generated content would fuel new media that could subsume and replace broadcast media, seems like an obvious bet from the vantage of 2023, when YouTube remains a powerhouse in a crowded field of media platforms that combine amateur production, professional licensing, and social engagement (Poell et al., 2022). Yet even in the recession’s wake, it was not yet clear that this vision would win out. Garon (2010: 435–436) wrote:

The question remains whether Google’s acquisition of YouTube was a short-sighted folly or a long-term strategy for dominance. If the broadcasters continue to decline, a central site that delivers content to computers, portable music/video devices, cellular phones and Internet-equipped televisions could become the new media hub.

For Google’s new media vision to succeed, in other words, broadcast media needed to fail decisively. Deregulation, consolidation, and financial crisis had ensured that they would hardly need Google’s help in doing so.

Conclusion: automation, deconvergence, disjuncture

The sudden change in strategy for Google—from treating radio as a proving ground for its ad systems to distancing its audio projects from radio—points to an intriguing alternate scenario where companies like Google would have instead pursued more integration between broadcast radio and the then-emerging streaming model. Had the 2008 recession not wiped out the imperative for advertising platforms to diversify across media and had Google succeeded in building a general auction platform for audio ads, one can imagine a radio landscape with (at the least) a very different advertising makeup and also an Internet audio landscape where radio stations (as privileged ad space vendors) act as more significant interface nodes. Perhaps most intriguing for radio studies is the potential that, with Google Radio Automation, radio and Internet media would have continued their convergence through the interior of broadcasting’s infrastructure and yielded broader shifts in programming rather than just in advertising.

But the elimination of these possibilities through Google’s bifurcation of streaming from radio also highlights how and why streaming platforms became new media. It reinforces a fact that will be familiar to readers of New Media & Society: that supposedly “new” media are always more entangled with previous media industries and infrastructures than their boosters would have it (Peters, 2009). Media historians do well to treat claims to novelty as their own kind of media production in an always shifting yet always continuous apparatus (Gitelman, 2008). Nevertheless, it is significant that new media—or new media regimes, as we might call the platformized, on-demand media products of the 2010s and 2020s—tend to arise through particular moments of crisis rather than through an even-paced and consistently determined churn. These crises invite powerful actors to amplify and entrench novelty claims that they may have already been testing for some time. The break between radio and streaming was produced under very specific conditions that were not inevitable; it was a deliberate artifice on the part of Internet industrialists, but external factors outside their control motivated and timed it. Among these factors, automation and advertising held a special status as both the foremost areas of Google’s expertise and as longer media industry trajectories that helped thwart it.

Radio automation software represented the kind of “technological hybrid” that Leah Lievrouw (2023: 23) has located as an important juncture: one where computerization in established media’s inner workings helped datafication become an organizing principle. Google’s active investment in updating the SS32 automation suite reflected the process Lievrouw describes where a new “media logic” gradually emerged out of this juncture as data-driven business models remolded user experiences around those points of juncture. Yet Google’s sudden divestment from radio and pivot to streaming audio reflected a point of disjuncture: a moment when technological resources were ported over to a separate media context so as to protect them from the first’s industrial liabilities. In radio, automation resulted from and contributed to the same trends—consolidation and regularization under an advertising-driven broadcast model—that produced those liabilities. The processes that destabilize “old” media are essential components in the production of “new” media. New media scholarship has identified automation as one of the defining principles of a present media regime (Andrejevic, 2019; Manovich, 2002); by further understanding automation as both an infrastructure and a vector for the interests that corrode infrastructure, media studies might better anticipate the regime that will supplant this current one.