Automated content recognition (ACR),smart TVs,and ad-tech infrastructure

Abstract

Smart TVs, which enable unprecedented viewing data to be collected at device level, are reconfiguring business practices in television advertising and audience measurement. Using a video fingerprinting technology known as automated content recognition (ACR), most smart TVs can now recognize virtually all content displayed on the TV screen, allowing precise targeting and re-targeting of ads to users based on their viewing behaviour and prior ad exposure. This article examines how TV manufacturers including Samsung, LG and Vizio integrated ACR technology into their businesses between 2012 and 2022. Drawing on infrastructural theories of connected TV and digital advertising, I show how ACR has contributed to TV manufacturers’ wider goal of becoming digital ad platforms, shifting the underlying power structures of both the consumer electronics and television advertising industries.

Introduction

Most TVs sold today are internet-connected, app-enabled smart TVs. Among the distinctive features of smart TVs is their capability for advanced surveillance, including collection and analysis of the user’s viewing history to enable precise targeting of ads (Varmarken et al., 2020). Blurring the boundaries between television and online media, smart TVs have effectively moved television advertising into the realm of digital ad-tech – the infrastructures that distribute, target and monetize advertising online. No longer a one-way receiver, the TV is now a sophisticated tracking device that closely watches the user, while the user watches the TV.

This article investigates a little-known technology that is integral to the wider process of smart TV surveillance. Automated Content Recognition (ACR) is an algorithmic capability that enables smart TV manufacturers to identify what content is being shown on-screen. ¹ Built into smart TVs at the chipset level, ACR can be used to identify almost any kind of content – including movies, TV episodes, ads, games, and live sport – regardless of source. Smart TV manufacturers work with their ad-tech partners to construct detailed consumer profiles based on ACR data and to target future advertising on this basis. As such, ACR-enabled tracking goes far beyond the familiar capability of apps to track in-app viewing to personalize the service to the individual user; instead, the purpose of ACR is to identify any content shown on the TV, across all apps, channels, and external devices, thus building up a detailed history of content viewed over potentially many years of use, and to share this information with advertisers and ad-tech intermediaries. ACR is also commonly used by analytics, metadata and hardware providers in the connected TV industries (e.g. iSpot, Gracenote, Tivo); however its integration into smart TVs is the focus of the present article.

To realize this constant surveillance, ACR-enabled smart TVs use video and/or audio fingerprinting to identify the content shown on screen. A Samsung TV, for example, ‘takes glass-level screenshots every 500 milliseconds […], converts them into “unique patterns,” and compares these visual snapshots against others in the [reference library]’ (Samsung Ads, nd: 7). Vizio smart TVs use a similar approach, as described in their press kit for advertisers:

When a piece of content runs across any input or ancillary device on opted-in VIZIO TVs, ACR detects image pixels and audio as they play in real-time. We match those to our library of known fingerprints, giving us meaningful viewership insights, which then can be easily applied to VIZIO media buys. (Vizio 2024)

Variations on this approach are used by other TV manufacturers, which either build a proprietary ACR tool into their TVs or integrate with a third-party ACR provider, such as Samba TV. Unless the user opts out – which is uncommon given ACR disclosures are buried deep in the terms and conditions ² – the TV will continue to collect viewing data whenever it is turned on, building up a rich history of the users’ interests and personality over time. Importantly, the object of surveillance for ACR is the ‘glass’, the TV screen, rather than the viewer; in this sense, ACR is fundamentally different from earlier audience measurement tools such as diaries which track the viewing activities of individuals. However, when triangulated with first-party data from advertisers, ACR becomes a powerful tool for knowing the audience, contributing to what Ien Ang (1991, 18) famously described as ‘the stipulated, official and generalized knowledge produced by the discourse of audience measurement’.

From the perspective of television industry research, ACR forms part of what Mark Andrejevic (2004: 35) famously called digital enclosure, or ‘the process whereby activities and transactions formerly carried out beyond the monitoring capacity of the Internet are enfolded into its virtual space’. Additionally, ACR is closely related to datafication in audience measurement, as theorized in recent television audience and industry scholarship (Doyle, 2018; Kelly, 2019; Kosterich and Napoli, 2016; Webster, 2014). Yet smart TV ACR has barely been mentioned in the existing literature on these topics (for exceptions, see Heuguet, 2019; Navar-Gill, 2019; McGuigan, 2021). It is therefore timely to consider the emergence of this now-pervasive technology. In this article I address two research questions: What were the industrial arrangements that facilitated ACR’s entry into our TVs, and our lounge rooms? And what have been the implications of ACR adoption for the consumer electronics and television advertising industries?

This article’s approach to ACR is informed by a rich tradition of work on television audience surveillance, which includes foundational theorizations of audience commodification (Meehan, 1990; Smythe, 1977), the ‘work of being watched’ (Andrejevic, 2004), and the related role of audience measurement technologies, from diaries and peoplemeters to biometric and online tracking (Ang, 1991; Buzzzard, 2012; Napoli, 2011). These studies, and others, have contributed to what is now a sophisticated theorization of audience surveillance as an extractive process whereby audience attention is recorded, monetized, packaged, and sold, with measurement technologies providing the enabling infrastructure for these transactions. While mindful of the value of surveillance critique, I want to stress that the primary purpose of this article is not to theorize ACR as a surveillance technology (that much is already clear). Instead, my aim is to do something different: to explain how ACR came to be integrated into business practices in the television and consumer electronics industries, and to reflect on what this can tell us about present and future relations between the consumer electronics and television advertising industries. To do this, the article locates ACR in relation to existing research on digital content recognition (Eriksson and Heuguet, 2021; Heuguet, 2019), television audience measurement (Buzzard, 2012; Hessler, 2021; Napoli, 2011), digital ad-tech (Braun, 2013b; McGuigan, 2021, 2023; Turow, 2012), and TV manufacturing (Chandler, 2005; Galperin, 2004; Hart, 2004), approaching ACR as a boundary object that connects these histories, which are mostly treated separately by scholars.

The key claim of this article is that ACR is contributing to a power shift in television advertising, cementing the position of TV manufacturers as vital data intermediaries in the attention economy. As I will show, leading TV manufacturers are now pursuing a platform strategy that seeks to capture a slice of advertising revenues, leverage emerging currencies of audience measurement, and control their own FAST (free ad-supported TV streaming) channels. No longer mere hardware providers, TV manufacturers are now attempting to move up the value chain by becoming media and advertising companies in their own right – representing a major realignment in the political economy of television advertising that requires critical investigation. ACR, I will argue, is central to these ambitions as it supplies the data needed for manufacturers to participate in digital ad markets. In this respect, my analysis seeks to advance media industries analysis by underscoring the need ‘to correct understandings of digitalization and of recent changes in television that are excessively Internet- and web-centric’ by reaffirming the importance of ‘consumer electronics and IT companies that develop, manufacture and market devices’ (Hesmondhalgh and Lobato, 2019: 959).

I use a critical media industry studies framework, drawing primarily on trade press sources. To assemble an archive for analysis, I read all articles in the trade publications Broadcasting and Cable, TWICE (This Week in Consumer Electronics), Ad Age, AdWeek, and Multichannel News published since 2008 which contained the keywords ‘ACR’ or ‘content recognition’ (more than 200 articles in total). Keyword searches in academic databases (Factiva, ProQuest, EBSCOHost) were used to locate these articles which were then arranged into chronological order to provide a record of trade press coverage. To contextualize this core data set, I also consulted major business news (Financial Times, New York Times), connected TV trade sources (TV[Rev], Streaming Media, Broadband TV News), policy and regulatory reports (FTC, Ofcom), and technical documents (whitepapers and patents from Samsung, LG, Samba, Zapr, Viant, and Cognitive Networks). From these sources I constructed a timeline of the key technologies, product launches, mergers, acquisitions, public statements, and PR strategies that helped to define ACR industrially as a media technology. Like other archival methods, trade press analysis is limited by the biases of its sources, which often perform a promotional function. The approach used in this article is to gather and reflect critically on the information trade sources provide, scouring documents for factual information while also understanding those documents as discursive artefacts that are ‘coded, managed, and inflected’ (Caldwell, 2008: 2) and which must be analyzed on that basis. Accordingly, this article does not claim to access any behind-the-scenes ‘truth’ about ACR. Its more limited ambition is to assess how the consumer electronics and advertising industries have publicly reflected on ACR as a disruptive technology reshaping parts of those industries over the last decade and a half.

Infrastructural approaches to connected TV advertising and content recognition

Scholars in television, digital media and advertising studies have recently developed helpful concepts that can guide our analysis of ACR as a form of digital media infrastructure. While addressing diverse topics, this research offers a common way of thinking about digital media as systems for systems for the accumulation, storage and circulation of data, which enable new opportunities for value creation by media platforms and digital intermediaries. As such, the focus of such research often falls on the ‘infrastructures of identification’ (Frith 2019) that identify both individuals and the digital media they interact with. In the following literature review, I discuss several key studies that apply this mode of analysis in generative ways.

Lee McGuigan’s sociotechnical histories of advertising are particularly instructive here (McGuigan, 2021, 2023). McGuigan’s work seeks to locate ad-tech technologies within a wider critical context informed by histories of science and technology. For McGuigan (2021: 127), ad-tech is an ‘infrastructure for administering discrimination’, which has enabled new market arrangements based on the identification of individuals, including their interests, attributes, and buying behaviours. In a recent, short intervention, McGuigan (2021) begins the work of locating ACR within a history of audience surveillance extending from early peoplemeters and audimeters, via the set-top box, through to online cookies, trackers and pixels. McGuigan argues that such technologies not only provide a basis for audience measurement but are also ‘important sinews and circulatory systems for a data- and surveillance-intensive form of capitalism’ (123). His work carefully documents the industrial structures, rationalities, and discourses that have emerged around these technologies, providing a powerful lens for understanding the integration of technologies into industrial practice.

Joshua Braun’s studies of connected TV infrastructure similarly emphasize the sociotechnical dimensions of advertising technology (Braun, 2013a; 2013b, 2015). Braun’s work on online television distribution in the early 2000s considers the role played by what he calls transparent intermediaries – a category that includes ad-tech providers, content delivery networks, and other obscure parts of the video value chain ‘whose technologies and activities increasingly enable, but also structure, the distribution of video online and, by extension, our experiences of connected viewing’ (Braun, 2013b: 125). Braun is one of the few media scholars to have paid close attention to these business-to-business intermediary firms, whose services are often ephemeral, obscure, and sold on a white-label basis. Another example is Jeremy Morris, whose concept of infomediaries (Morris, 2015: 452) helpfully refers to ‘organizational entities that monitor, collect, process and repackage cultural and technical usage data into an informational infrastructure that shapes the presentation and representation of cultural goods’. While these studies do not address ACR specifically, they provide a foundation for understanding providers of ACR companies as intermediaries/infomediaries, whose work is integral to online video economy.

A parallel strand of research investigates the cultural histories of content-recognition technology across video, text and music. Notably, Guillaume Heuguet and Maria Eriksson (Eriksson and Heuguet, 2021; Heuguet, 2019) have begun the important work of locating content-recognition technologies within the history of visual, cultural and communication industries. Their co-edited introduction to a recent special issue of Internet Histories uniquely applies a genealogical approach to content recognition. For Eriksson and Heuguet (2021), content-recognition tools are not merely cameras mutely processing flows of cultural data; they are also engines that act on culture in subtle but important ways:

content identification tools do much more than simply identify: at a technical and semiotic level – and under the pretext of recognizing singularities – they model, interpret, calculate, classify, and filter. This is especially the case when content identification techniques acquire infrastructural features, become naturalized, and blend into the technological background of the everyday. (Eriksson and Heuguet 2021:2)

Their account emphasizes, in a Foucauldian fashion, the connection between identification technologies and the surveillance of populations. In other work, Heuguet (2019) has also excavated the more specific impacts of a particular content-recognition technology – YouTube’s ContentID. Heuguet is careful not to approach such tools through a determinist lens of flat and dominating effects; instead, he raises the possibility that such tools might represent a sociotechnical ‘trade-off between the circulation of cultural forms and their control as works and goods… a technological compromise for the management of copyright and related rights online’ (Heuguet, 2019). This idea provides a useful lens for understanding ACR also, as I will argue below.

Together, these interventions provide suggestions for how we might frame a study of ACR. The aforementioned scholars pay careful attention to the industrial evolution of advertising and content-recognition technologies, showing how their path is never pre-ordained but is instead produced by contingent arrangements of technical, regulatory, commercial and discursive elements. Additionally, the infrastructural orientation of these scholars – the focus on back-end technologies – provides an evocative model for how to study the now-pervasive infrastructures of connected TV ad-tech. In the next section, I draw on these concepts to show how ACR has become integrated into the consumer electronics and connected TV advertising economies, and how it has contributed to shifting industrial norms within those sectors.

Institutionalizing ACR in consumer electronics: The role of TV manufacturers

The roots of ACR lie in three general-purpose technologies which, in various combinations, enable the identification of digital content. These are: hashing (the algorithmic technique of creating short, unique identifiers, such as a string of letters and numbers), watermarking (embedding invisible identifiers within digital content), and fingerprinting (identifying digital content based on an algorithmic calculation of its properties, checked against a reference database). Each technology has a longer history. Hashing is used widely in cryptography, while video watermarking systems, such as Macrovision, have long been used to identify copyright-protected content to restrict infringing uses. Fingerprinting is the most important part of contemporary ACR, and it is the only technology that can effectively be used at scale to identify a very wide range of video content.

Two kinds of fingerprinting – video and audio fingerprinting – have been used by ACR providers to identify content shown on TVs. Audio fingerprinting techniques analyze time-frequency range and tempo, among other variables, and perform various mathematical calculations on those data to create a unique identifier. Companies offering ACR solutions based on audio fingerprinting include Audible Magic – a Silicon Valley firm founded in 1999 whose fingerprinting method has been licenced by platforms including YouTube, Facebook, Dailymotion, Twitch, SoundCloud, and Tiktok (EUIPO, 2020:18) – and Shazam, a content-recognition company famous for its song recognition app released in 2008. As transparent intermediaries providing back-end infrastructure to monetize online video, these two firms have played a substantial and but largely unrecognized role in the streaming and ad-tech economy. Shazam, in particular, was involved in advertising from its earliest days: its free song recognition app was also a covert monitoring tool, recording ambient sounds via the microphone and sharing this data with advertisers to determine which ads and content the user had been exposed to. As Rozlagova (2018: 260) observes, this ‘feature creep transformed Shazam into an effective surveillance and marketing engine’.

Video fingerprinting involves identifying content based on the arrangement of pixels in video frames. Some of the key providers of video fingerprinting tools for smart TVs include: the Silicon Valley firm Cognitive Networks; Flingo (now SambaTV), a social TV and ACR company founded by ex-Bit Torrent staff; the Dutch firm Civolution, a spin-off of Philips; and the Korean firm Enswers. These companies worked with TV manufacturers throughout the 2010s to develop various ACR solutions capable of identifying whatever was shown on the TV screen (‘the glass’). Importantly, this included not only movies and TV shows but also ads – a capability that opened up new commercial applications for television advertisers. For example, Cognitive Networks’ ACR tech was designed so content providers and advertisers could initiate interactive advertising features – polls, pop-up ads, quizzes, and so on – when certain ads appeared; an interactive order pop-up might appear during a Pizza Hut ad. Cognitive Networks’ ACR also had a range of other uses in television advertising including ‘match[ing] data to other data sets, building targetable audience segments, and [enabling] measurement of return on investment in programmatic efforts’ (Winslow, 2015).

These early years of ACR experimentation correspond to what Kosterich and Napoli (2016: 261-2), in their analysis of social media metrics in advertising, refer to as the ‘preinstitutionalization’ period – the time when ‘relatively few organizations […] have adopted the new structure or practice, and are doing so in an ad hoc, or experimental, manner’. As Kosterich and Napoli (2016: 262) observe,

Preinstitutionalization also represents the stage at which organizations begin to evaluate the utility of this alternative approach to measuring and valuing television audiences… As preinstitutionalization occurs, legacy stakeholders and organizations attempt to make sense of this new playing field with some degree of exploration and potential participation.

In the case of ACR, this preinstitutionalization period proved to be a time of experimentation, when the potential gains, risks, and opportunities of ACR remained unclear, and stakeholders were exploring the best way to integrate it into their businesses. Between 2008 and 2012, when the ACR hype began to peak in the television and consumer electronics industries, a flurry of partnership agreements between manufacturers and ACR providers were signed. Trade reports from the 2012 Consumer Electronics Show, the major trade fair for TV technology held annually in Las Vegas, reported that a host of ACT providers, including Audible Magic and Gracenote, were actively showing their wares and seeking to partner with TV manufacturers to embed ACR capabilities into TV sets (Ng, 2012; Nuttall, 2012). During this preinstitutionalization period, many manufacturers were considering whether to build up ACR expertise inside their own companies or buy-in that expertise through acquisition and licencing.

By the middle of the 2010s, ACR had become firmly integrated into the institutional apparatus of consumer electronics manufacturing. Most TVs being shipped by the major manufacturers now had some form of ACR built-in, and ACR capability had also been specified within the ATSC3.0 NextGen TV broadcast standard in the United States. Tellingly, ACR was number one on Broadcasting & Cable’s list of buzzwords TV industry professionals need to know about for 2015, with the venerable trade magazine crowing that ACR ‘could make advertising much smarter and give programmers real-time measurement of what is being viewed’ (Winslow, 2014). Together, these accomplishments signalled ACR’s arrival as a viable advertising technology. However, more work needed to be done to find a place for ACR within the structures of consumer electronics manufacturing. How should this disruptive capability be assimilated into the corporate, technical and industrial processes used by the largest consumer electronics firms?

TV manufacturers took different approaches in this regard, with some outsourcing their ACR operations and others preferring to build up internal capacity. Samsung initially partnered with ACR providers including Enswers and Yahoo Broadcasting Interactivity, but eventually built up its own in-house ACR capability. Over time, it has acquired a number of other firms, including the Indian provider Zapr, to boost its ACR tech stack. LG was a later arrival to the ACR party. After partnering initially with Cognitive Networks, LG took a controlling stake in the ACR provider Alphonso in 2021 which it eventually absorbed into its own in-house operations. This provided the basis for LG’s ad-tech operation, now known as LG Ad Solutions. A similar approach was taken by the US TV manufacturer Vizio, which acquired Cognitive Networks in 2015. The strategic importance of ACR to Vizio was reflected by the fact that, after the acquisition, Cognitive Networks CEO Zeev Neumeier became Vizio’s Chief Innovation Officer and served in that role for almost a decade.

While the largest TV manufacturers could afford to acquire ACR start-ups, second-tier TV manufacturers often preferred to outsource the task of content recognition to external providers. For example, the US-based company SambaTV supplies Sony, Toshiba, Panasonic, and other smaller TV manufacturers with their ACR back-end. Samba’s ACR is often preinstalled into smart TVs and presented, somewhat disingenuously, as a personalization service and recommendation agent. Through these partnerships with manufacturers SambaTV has grown into a sizeable operation: its tracking technology is built into more than 40 million devices globally (SambaTV, 2024), providing a global infrastructure for big-data surveillance of TV viewing.

In each case, manufacturers and their ad-tech partners were able to leverage ACR for different purposes including advertising, audience measurement, and optimization. Whitepapers from ACR providers give a sense of the uses of ACR, which seek to monetize viewing data in ways both creative and concerning. For example, the US-based ACR provider Viant (nd 8) describes how its ACR product was used to measure ad effectiveness for a luxury retailer; the client provided ‘a nightly feed of their in-store sales data, which was then matched back to the brand’s existing customer profiles’ to reveal how many customers that had seen smart TV ads for the brand ended up visiting the retailer’s stores. Here, ACR data – when combined with first- or third-party transaction data – effectively link ‘viewers’ cross-channel ad exposure to their offline and online purchases’ (ibid).

Another use for ACR is to create audience segments based on video consumption. For example, the Indian ACR provider Zapr (now owned by Samsung) allowed its clients to target and segment individual viewers based on their viewing choices:

Retail Investors are audiences who watch business news during the first half of the day since it means they require this kind of information for professional purposes. These audiences are a good segment to have for premium products and investment/banking related products… Rural Audiences are those who heavily watch FTA (Free to Air) channels. Digital planners can use this segment to reach rural consumers for products like tractor brands… (Zapr, nd 18)

Here we see how private viewing choices provide the basis for market segments (high net-worth professionals versus rural farmers) which then enable further targeting and tracking. While ACR’s effectiveness is restricted by a range of factors, including multi-member households, return path reliability, and audiences’ ability to view content that might not align to marketers’ assumptions, ACR’s promise to target advertising to the individual has helped manufacturers to distinguish their ‘advanced’ advertising product from traditional broadcast TV advertising, which infers viewers’ interests based on programming characteristics.

There is also a related use-case for ACR in audience measurement. By tracking what people are watching on their smart TVs, ACR-enabled data collection can replace traditional audience ratings tools such as peoplemeters and set-top boxes. For this reason, ACR is often touted by its champions as the solution to the longstanding problem of audience fragmentation, accelerated by the recent drift to streaming services. As one Samsung executive puts it, ‘in a world where there’s more fragmentation, where agencies and advertisers are finding it really hard to understand and reconcile what’s happening across all these different channels, the manufacturer [using ACR] is really able to provide that holistic view’ (The Drum, 2022). This capability has not gone unnoticed by established players in audience measurement. Global industry leader Nielsen entered the ACR business in 2017 via its acquisition of the music and video metadata firm Gracenote, and has since signed major ACR deals with Vizio, LG and Roku. Their long-term aim is to combine ACR data ‘with Nielsen’s representative panel […] to provide the most accurate modern audience measurement of linear ads as well as increase coverage’ of connected TV viewing (Nielsen, 2023). In recent years, Samsung and Vizio have also both announced the launch of their own ACR-based panels for audience measurement. As these examples suggest, ACR has not replaced conventional forms of television audience measurement but now coexists and competes with them, and can be combined with panels and other traditional methods as part of multi-method measurement systems. This reflects the tendency towards the ‘layering or bundling of distinct systems’ noted in previous infrastructural studies of media technologies (Parks and Starosielski, 2015: 9).

Looking back at this integration of ACR into consumer electronics and advertising, I would make two general observations. First, the history of ACR in television was by no means a linear process. ACR, as I have shown, has never been a single ‘thing’; rather, it is a bundle of identification tools offered by different providers, which manufacturers integrated into TVs in different ways and to different extents. In this sense the story of ACR bears some resemblance to the wider logic of ad-tech, described by McGuigan (2021: 125) as ‘a site of contest and compromise, as the interests of buyers and sellers are negotiated in the construction and stabilization of a product whose tangible existence is defined precisely through those exchange relations’. What we now call ACR was a product of these exchange relations, which determined how content identification could be integrated into business practices.

The second point to note here is that the integration of ACR into smart TVs was a largely unregulated process. As a result, manufacturers and their ad-tech partners were free to develop ACR products as they wished. While the ability to spy on TV viewers and sell this data to third parties would seem to be a step change requiring close policy scrutiny – especially given that such practices mark a break with existing legal precedents about the privacy of video viewers (witness the United States’ Video Privacy Protection Act of 1988, which guaranteed that video rental histories could not be made available to third parties) – regulatory oversight was the exception rather than the rule. The Federal Trade Commission launched in 2017 an action against Vizio, resulting in a US$2.2 million fine and public rebuke; but the issue at stake there was insufficient disclosure and transparency about ACR tracking, not the right of manufacturers to track viewing in the first place. Equally, controversy over Samsung’s smart TVs in the mid-2010s focused on the exceptional instance of webcam/microphone-enabled spying on users rather than the pervasive, and arguably more concerning, practice of ACR-enabled mass surveillance of viewing. In Europe – where the GDPR (General Data Protection Regulation) has been in force since 2018, providing a degree of transparency and accountability for smart TV data collection – regulators have generally taken a firmer hand in the regulation of digital markets. However, TV manufacturers in Europe remain free to develop business models based on ACR tracking, evidenced by the launch of Samsung Ads Europe in 2018 and LG Ads Europe in 2022. In this sense, ACR has been able to quietly establish itself under the radar as a mainstream technology in the television advertising and consumer electronics industries, with relatively little direct regulation.

So far I have described how ACR was institutionalized throughout the 2010s as TV manufacturers scrambled to integrate ACR tech into their smart TVs. Now it is time to consider some longer-term industrial transformations associated with ACR. Below, I describe two key shifts: the platformisation of TV manufacturers, and their ACR-supported investment in TV services and content.

Shift 1: TV manufacturers become platforms

The leading consumer electronics firms – Sony, Samsung, LG, and others – have long seen themselves as hardware businesses. They excelled at producing physical devices at scale, efficiently, and with sufficient innovation to keep users replacing these devices at regular intervals (Chandler, 2005). Until the 2000s, this recipe was enough to ensure steady profits in TV set manufacturing. But the entry of Chinese manufacturers including Hisense and TCL into the TV business in the 2000s changed this cosy arrangement, leading to unprecedented price competition. As the retail price of TVs fell, the margins of TV manufacturers became wafer-thin. Average margins in TV hardware are now less than 1%, whereas margins in connected TV advertising are 50% or more (Thomson, 2024). Faced with these shrinking hardware margins, manufacturers have increasingly turned to services – advertising, content sales, and data brokerage – to improve their bottom lines. Just as Apple’s App Store established a lucrative services business alongside Apple’s hardware business, TV manufacturers are similarly keen to move up the value chain to higher-margin activities.

The strategy here is to become economic platforms – institutions that can extract rent from multi-sided marketplace transactions (Evans, 2011). The smart TV is a platform in several different senses: its operating system is a software platform for third-party services; its app store is an economic platform enabling firms to transact within the smart TV’s user interface; and its home screen, menu interfaces and free advertising-supported streaming TV channels (FASTs) are platforms for paid advertising from business partners. In this sense, the smart TV combines several platforms in one.

For example, Samsung staff refer to the company’s smart TV operating system Tizen as ‘a homegrown platform that continues to evolve’ (Samsung, 2024) and to Samsung TV Plus as ‘a next generation media platform’ for third-party content (Samsung, 2023). Meanwhile LG Ads is described as a ‘connected TV media and measurement platform’ (Cahillane, 2001). While use of the term platform is endemic in digital media industries, adoption of a platform vocabulary nonetheless reflects a shift in the self-understanding of these consumer electronics firms, which now regard success as premised on interoperability, cooperation, and revenue-sharing with third-party services, alongside traditional strategies of supply-chain efficiency, strategic branding and competitive pricing.

In a platform business model, revenue is no longer limited to a one-off hardware sale but can also involve a years-long extractive process based on micro-payments and commissions from transactions and ad exposures that the TV will generate over its lifetime. In an unusually candid interview, the CEO of the leading US-based TV manufacturer Vizio, Bill Baxter, explains the strategy as follows:

It’s about post-purchase monetization of the TV. This is a cutthroat industry. It’s a 6-percent margin industry, right? I mean, you know it’s pretty ruthless. … I need to make money off those TVs. They live in households for 6.9 years – the average lifetime of a Vizio TV is 6.9 years… [O]ur strategy – you’ve seen this with all of our software upgrades including AirPlay 2 and HomeKit – is that we want to make things backward compatible to those TVs. So we’re continuing to invest in those older TVs to bring them up to feature level comparison with the new TVs when there’s no hardware limitation that would otherwise prevent that. And the reason why we do that is there are ways to monetize that TV and data is one, but not only the only one. It’s sort of like a business of singles and doubles, it’s not home runs, right? You make a little money here, a little money there. You sell some movies, you sell some TV shows, you sell some ads, you know. (in Patel, 2019)

By 2021, Vizio was making more money from its platform business than it was from hardware sales – by a factor of two, and with a much higher profit margin. FTC filings in 2021 noted that Vizio generates, on average, an annual revenue of US$12.99 from each user of its SmartCast TV operating system, via ad insertions in Vizio channels, ad placements on the TV home screen, sale of personal data to third parties, commissions on SVOD signups, content sales, and branded buttons on Vizio remote controls (Vizio, 2021: 3-4).

As the examples cited above show, ACR is a crucial element in the platformisation strategy employed by TV manufacturers including Samsung, Vizio and LG. By knowing the user’s tastes, habits and ad exposure, ACR allows ‘better’ targeting of ads, thus allowing manufacturers and their partners to increase the CPMs (cost per mille) of ads, on the basis of this claimed precision targeting. ACR also enables personalized marketing for manufacturers’ in-house services, such as digital content sales and FAST channels, and diverse opportunities to integrate ACR data with first-party data from advertisers and other business partners. All of this has contributed to manufacturers coming to see themselves in a new way – as platforms.

This change cannot be entirely attributed to ACR, of course; it is equally the result of longer-term cost pressures in TV manufacturing that have eroded hardware profit margins as well as management theories that emphasize value-added services over hardware. But what is significant for our purposes is that this transformation has effectively blurred the boundary in consumer electronics between television hardware, software and content. Manufacturers are no longer merely device producers; they are also data platforms, ad brokers, and analytics providers. ³ In other words, what appeared to be stable boundaries between industry sectors – hardware, content and services – now become fluid. The boundaries of television as a ‘cultural industry’ become more difficult to make out.

Shift 2: TV manufacturers become content providers (FAST channels)

A second noteworthy outcome of ACR is the entry of TV manufacturers into content and programming, achieved through the launch of FAST (Free Ad-Supported Streaming TV) channels. Since 2015, several manufacturers have started offering their own FAST platforms, such as Samsung TV Plus, LG Channels, TCLtv+, and Hisense’s VIDAA TV. These FAST platforms consist of dozens (sometimes hundreds) of linear streaming channels that can be zapped through using the remote control, just like traditional broadcast or cable TV channels. As Seijin Woo, a Samsung product manager, observes, Samung’s FAST service Samsung TV Plus ‘was designed to resemble broadcast TV’ in its simplicity and accessibility (Samsung 2023). However, unlike broadcast TV FAST channels are heavily promoted on the smart TV’s home screen and in the EPG (electronic programme guide), making them more discoverable within the user interface than their broadcast competitors (Scarlata and Dharmawardhana, 2024). ⁴ This strategic integration reflects manufacturers’ ambition for their own FAST services to compete with, and possibly replace, legacy linear broadcast and basic cable television (Wolk and Damata, 2023).

The content offered by FASTs is diverse, and includes everything from movies, sitcoms, and music videos to niche sports, news, and user-generated content – but unlike premium pay-TV channels, most FAST channels offer back-catalogue content (older movies and TV shows) or low-budget content (weather, pets, minor sports channels, etc.). As of 2023, there were more than 1600 different FAST channels in the US, offered by TV manufacturers and video platforms (Bridge, 2023). Other countries have a smaller but growing number; our 2023 analysis of the Australian market found more than 600 FAST channels (Scarlata, Lobato, and Dharmawardhana, unpublished research). These FAST channels are typically licenced by manufacturers from film and TV studios, content providers, cable channels, and multi-channel networks (Lobato, 2016). For example, the Korean provider New ID offers dozens of off-the-shelf FAST channels including BabyShark TV, World Billiards TV and My Secret Romance that are available on various brands of smart TVs; while the French-owned multinational production house Banijay offers dedicated FAST channels for many of its popular shows, including Mr Bean and 8 Out of 10 Cats that are licenced by Samsung, LG and TCL.

Manufacturers, when licencing FAST channels, negotiate a commercial deal for access to their smart TVs. This typically comprises an agreed percentage of ad inventory within a FAST channel – for example, 30% of the ad slots – which it can then sell to its own clients through an in-house ad team (LG Ads, Samsung Ads) and/or on programmatic marketplaces. Alternatively, an equivalent share of the channel’s ad revenue may be negotiated. In either case, the manufacturer always extracts some kind of rent in return for distribution of the FAST channel across its installed base of smart TVs. In this context, ACR plays a multifaceted role. First, ACR enables granular targeting based on prior viewing history, as explained earlier. Second, there is a marketing play: ACR, by allowing the manufacturer to know the viewer(s) intimately, means the manufacturer can promote its FAST channels on the user interface as replacements for competitor services; for example, viewers that stream a lot of broadcast news can be offered a FAST news channel instead. Third, insights from ACR data inform manufacturers’ decision-making about which FAST channels to licence, allowing them to optimize their content investment. These capabilities come together to give manufacturers a distinct advantage as streaming content suppliers to smart TVs. As an executive from LG Ads Solutions explains,

As OEMs [original equipment manufacturers] we know what people like to watch on linear TV via our ACR data. And that has allowed us to pick and choose the type of content we want to put in front of users. That amount of data doesn’t really exist with anybody else and it’s given us the ability to experiment too with the type of linear [FAST] channels we put up. (Raghu Kodige, President, LG Ad Solutions, in Wolk and Damata, 2023)

As Kodige acknowledges here, the rise of FASTs was infrastructurally enabled by ACR, which also provided the monetization model for FASTs. Absent ACR, ads would have to be sold by manufacturers the old-fashioned way, using the program as an index for demographics (target men through sports, women through soap operas, and so on). But with ACR, manufacturers could make inferences about their audiences based on glass-level data collection. In this way manufacturers can leverage their control of the ACR ad-tech and FAST programming to become ad platforms, with ads programmatically traded and targeted to individuals – thus bringing TV advertising closer to the micro-targeting of social media. And because manufacturers control the device itself, they can offer advertisers a solution for accessing high-value, hard-to-reach, cable-cutter TV audiences who no longer watch much network or cable TV.

Kodige’s comments are symptomatic of the tendency among manufacturers to present FAST as the future of TV advertising: a new paradigm of data-driven, nimble, and disruptive advertiser solutions based on the ability to know the user intimately – all powered by the wonders of ACR. This is but one example of a pervasive discourse. Vizio’s Vice-President of Ad Sales describes ACR as ‘one of the most cohesive and impactful signals… in the data marketplace today’ (Wolk, 2021). This all aligns with McGuigan’s (2021: 128) astute diagnosis of the rhetoric of digital ad-tech, with its ambition to ‘tell the difference between users, to make bets about their worth…, and to verify outcomes with greater confidence’.

Still, the business model behind these FASTs clearly represents a shift in the business of television advertising. For the first time, manufacturers have become content providers and programmers, and get to keep the ad revenue generated by their channels (minus licencing payments or revenue-share arrangements). Just as importantly, manufacturers keep the ACR viewing data which can be used to enhance programming and target future advertising. In other words, manufacturers have used ACR to gain a foothold in the content business – amounting to a partial vertical integration move. The cultural implications of this shift are still unclear, because FAST is an emergent phenomenon and audience awareness is low. But over time, it is possible that FASTs may become a permanent and possibly even dominant feature of the television landscape in many countries, ceding further power to the manufacturers who play the role of gatekeeper and rent-extractor for those services.

Conclusion

By analyzing the uses of ACR in smart TV advertising, this article has posed a number of critical questions about the infrastructural basis of connected TV. I have argued that ACR – a little-known but now-pervasive technology in video platforms and devices – has been adopted enthusiastically by TV manufacturers who see in content recognition the prospect of capturing higher-margin services revenue, offering a pathway to diversify their revenue base away from hardware. Related to this strategy is the disruptive capability of ACR-enabled TVs to deliver and target ads, making the TV manufacturer a powerful actor in the emerging data economies of internet-connected television. Here, we see glimpses of a new industrial order emerging in which the global consumer electronics firms shore up their position against Big Tech (Amazon, Google, Apple) by asserting control over the data and revenues generated by their devices.

The use of ACR also marks a rupture with past practices in audience measurement. ACR-enabled measurement offers potentially greater scale than sample-based in-home monitoring systems. ACR is also distinctive, compared to earlier forms of television audience measurement, in that it requires no cooperation, and only the most minimal consent, from viewers; as such, it represents an interesting challenge to what Hessler (2021: 411) describes as ‘the “problem” of cooperation’ in audience measurement. But perhaps the most distinctive feature of ACR is the fact that – once again – the device manufacturer is ultimately in charge of data collection and monetization. In this sense, critical attention must once again turn to the manufacturer as the key site of power and control in the political economy of the smart TV.

A key challenge for future research will be to understand what all this means to television audiences in different countries. Whether the compulsory reinvention of the TV as a surveillance device is acceptable to users is still unclear. Public awareness of ACR is low due in part to manufacturers’ practice of burying information about ACR deep in terms and conditions. As a result it is not yet possible to meaningfully assess how people feel about their TVs spying on their viewing and selling data to advertisers, or to draw firm conclusions based on the limited number of people who opt out of ACR. And the privacy paradox of mass resignation alongside mass mistrust will make it difficult to establish actual levels of concern.

Nonetheless, the rupture with television history that the ACR-enabled smart TV represents is significant. For most of the last century, audiences have known the TV as a one-way receiver – a dumb device capable of little more than decoding and delivering content. Now the TV is both smart and surveillant, but our normative frameworks have yet to adapt. ACR providers, as transparent intermediaries supporting this process of mass surveillance, have played a crucial role in this sociotechnical reconfiguration of the TV, and their activities will require critical analysis in the years to come.