Direct marketization: Platform surveillance and the contextual integrity of democratic media

Abstract

This essay uses the theory of privacy as contextual integrity together with critical research on the role of media in democracies to critique platform surveillance and a related process that I call direct marketization. It focuses on the case of advertising attribution, a paradigm of audience and marketing measurement that attempts to determine advertising effectiveness by observing people as both media audiences and marketplace consumers. Advertising platforms have recently promised to implement ‘privacy-preserving’ methods of attribution measurement. The paper argues that these efforts to legitimize attribution make implicit claims about the values and purposes of media systems. It then introduces the concept of direct marketization to explain how these claims relate to shifts in the social and institutional norms of ad-supported media. The analysis exposes direct marketization and advertising attribution, which both fuel and depend on platform surveillance, as contradictory to the contextual integrity of democratic media.

Introduction

The advertising industry structures cultural production and the mediation of social life in many capitalist economies (Griffin, 2023; Joseph and Bishop, 2024; Leiss et al., 2018; Mattelart, 1991; Sinclair, 2012). That industry is now wrestling with a dilemma. On one hand, as surveillance and data-analytic capacities have been threaded everywhere into internet-enabled media (Burrell and Fourcade, 2021; Crain, 2021; Turow and Couldry, 2018; West, 2019), adtech companies – especially large platform companies like Google, Meta, Amazon, and Apple – have claimed an impressive power to attribute marketplace outcomes to specific advertising events (Pybus and Cote, 2024; Smith, 2019). Leveraging partnerships and software integrations with publishers, marketers, retailers, device makers, mobile-app developers, and data brokers (Khan, 2019; van der Vlist and Helmond, 2021), platform companies channel massive flows of information purporting to represent the complex pathways people take toward their purchasing decisions. These ‘attribution’ services promise to measure and optimize advertisers’ return on investment (ROI) by identifying causal links between audience members’ exposure to advertisements and their eventual consumer behaviors. Though attribution is an imperfect science, its institutionalization marks a signal achievement in the expansion of commercial surveillance and the platform enclosure of social mediation (Andrejevic, 2020; Napoli, 2019; Wu and Taneja, 2021), driven by continual demands for evidence of advertising effectiveness (McGowan et al., 2024; McGuigan, 2023).

On the other hand, though, adtech’s means of surveillance and data processing increasingly offend cultural and legal norms regarding privacy (Draper and Turow, 2019; Veale and Borgesius, 2022). Many routine adtech processes are becoming disreputable, unlawful, and/or obsolete (McGuigan et al., 2023a). Stakeholders in digital advertising are trying to finesse this dilemma by disavowing any intentions to ‘track’ users, while asserting their right to ‘measure’ advertising effects. They defend personalized advertising as an essential mode of patronage for ‘free’ or subsidized communication services (Interactive Advertising Bureau, 2022), and they characterize access to detailed data about the behaviors and economic value of consumers as a necessary entitlement for serving that function (Deighton and Kornfield, 2020; Interactive Advertising Bureau, 2024a). Platform companies stake these claims with vigor. They tout their unique capacities to determine advertising effectiveness within privacy-secure environs and with the help of artificial intelligence (AI) tools that can compensate for new limitations on marketers’ abilities to identify and observe individuals (Hercher, 2021; Schiff, 2023). Bolstered by their infrastructural and market power, platform companies are nudging the advertising ecosystem to depend even more on their authority to construct – and keep secret – the ‘facts’ of audience-consumer behavior.

This paper analyzes the antidemocratic implications of commercial platform surveillance. It draws on the theory of contextual integrity (Nissenbaum, 2010), which defines privacy as the appropriate flow of information, where appropriateness is determined by the norms, values, purposes, and priorities of specific social spheres, or contexts. I use this framework to examine advertising attribution, a paradigm of audience and marketing measurement that attempts to recognize how advertising events influence marketplace outcomes. Joining records of media audiences’ exposure to ads with records of consumers’ purchasing behaviors (or other valuable activities), attribution comprises an information flow that combines observations or inferences derived from two distinct, albeit porous, social contexts – media systems and marketplaces. The advertising industry’s turn toward ‘privacy-preserving attribution’ provides an opportunity to assess this ‘context collapse’ using critical scholarship about the roles of media in democracies (Cammaerts and Mansell, 2018; Couldry and Turow, 2014; Golding and Murdock, 2023; Napoli, 2019; Pickard, 2020).

The essay makes two contributions. First, I argue that adtech’s commitment to privacy-preserving attribution advances subtle but significant normative claims about the priorities of media systems and, in practice, contributes to a project of corporate-platform enclosure whereby platform companies exercise an authority to observe, coordinate, and extract data and value from more social and economic relations. Second, I suggest that the normalization of attribution is related to a long-run process that I call direct marketization, of which the platform-mediation of cultural production, circulation, and analytics is the most forceful contemporary movement (Carah et al., 2023; Helmond et al., 2019; Poell et al., 2021; Wu and Taneja, 2021). I use the term direct marketization to reference both the centrality of direct-marketing metrics and techniques within contemporary media systems, as well as the ongoing collapse of those media into shoppable marketplaces. Direct marketization has quietly altered the relationships among audiences, publishers, and advertisers, and so it deserves critical scrutiny. This scrutiny exposes direct marketization and attribution, which fuel and depend on platform surveillance, as contradictory to integral values of democratic media. Anyone sanctioning the legitimacy of attribution needs to either contend with those contradictions or admit that they regard media and marketing systems as coextensive.

Advertising attribution

Attribution services assemble knowledge claims about the effects of advertising events and the productivity of advertising venues (Smith, 2019). They belong to the long history of audience and marketing measurements – such as pantry audits and single-source data – devised to ‘close the loop’ between ad-supported-media usage and buying behavior. Attribution is now used to ‘optimize’ the allocation of advertisers’ budgets toward opportunities that appear likely to maximize ROI, and to determine the payments owed to media channels that sell ad inventory based on ‘conversions’ (e.g., purchases, app installs, store visits). In short, attribution is an outcomes-focused mode of analysis, fundamental to ‘performance marketing’, which, according to an article in Harvard Business Review, ‘has become the dominant approach companies use to connect with consumers’ (Stengel et al., 2023). This paradigm helps materialize a relationship wherein the advertising industry treats media and audiences as quantifiable investment opportunities.

Attribution is orchestrated in many ways and by many vendors, including demand-side platforms (e.g., The Trade Desk) that bid into programmatic auctions on behalf of advertisers; marketing cloud providers (e.g., Adobe, Salesforce); data brokers (e.g., LiveRamp, TransUnion); e-commerce service suppliers (e.g., Shopify); and platform giants (e.g., Google, Meta, Amazon) that operate integrated adtech stacks within and beyond their own walled-gardens. Almost any company that sells and/or distributes ads – from broadcasters like Tegna, to cable operators like Comcast, to streaming services like Hulu – may produce attribution metrics for their customers through partnerships with firms that track audiences or consumers. Companies that amass behavioral data with the most scale and scope – like large search, social, and e-commerce platforms – are best positioned to discern connections across media and marketing touchpoints.

The basic mechanics of attribution involve generating records of advertising exposure, cross-referencing those with records of relevant consumer actions, and assigning credit for causing those actions to one or more advertising events. Adtech platforms log an extensive bundle of data points for each impression served, including a timestamp, the user’s IP address, a unique user identifier (e.g., a cookie, device, or account ID), a campaign identifier, the publisher’s domain, geolocation, and more. A similar inventory is created for conversion events that take place on advertisers’ websites or apps (Clearcode, 2022: 142). The connection between impressions and conversions can be operationalized in several ways. In last-touch attribution, the measured outcome is considered to have been fully produced by the most recent ad event. By contrast, multi-touch attribution tries to account for the impact of all the ads someone encounters on their customer journey. The latter involves ‘tracking and normalizing all of the audience, device, media and sales/conversion data that go into media campaigns’ and then extracting a story about someone’s ‘path to purchase’ (Interactive Advertising Bureau, 2017: 2). Multi-touch attribution models can allocate credit based on pre-defined rules (e.g., equally across all ads; weighted according to time) or based on ‘fractional’ algorithmic analyses that estimate the value of each advertising event (Interactive Advertising Bureau, 2016: 5).

Attribution gets more complicated, and enrolls more means of identification and information processing, as its sphere of accountability increases. It is relatively simple to make a last-touch attribution claim about an ad engagement and a conversion that both take place within one web browser: a user clicks an ad beside a news article and is directed to the advertiser’s website, where a piece of software called a ‘pixel’ records user behavior on the destination site, associates that with a user identifier (i.e., a cookie ID), and sends the data to companies involved in serving the initial ad (e.g., ad networks run by Google, Meta, or Amazon). The URL for the advertiser’s site would likely include ‘UTM’ parameters that document the source of the traffic, including the news site’s domain and some details about the ad. ¹ Pixels and cookies are also used to attribute conversions to ads that are seen but not clicked. Google, Meta, Amazon, and others embed conversion pixels on countless sites across the web to observe the shopping activities of audiences to whom they deliver ads. Mobile-app advertising involves similar processes, except that software development kits (SDKs) facilitate user identification, behavioral record-keeping, and data sharing, in place of cookies and pixels.

Attribution methods that consider cross-device activity and in-store purchases require extra efforts to fasten the links in this audience-to-consumer chain, often using mathematical models to guess about relevant variables, such as whether ads served to multiple devices were seen by the same user and how much those events contributed to the eventual outcome. In some cases, the detection of personally identifiable information – such as an account log-in tied to an email address – allows users to be ‘deterministically’ matched to multiple devices. In other cases, ‘probabilistic’ conjectures about identity are made using proxies such as IP addresses. Companies providing ‘identity resolution’ services help marketplace actors coordinate their abilities to recognize and valuate users. Comprehensive attribution needs to draw offline behaviors into the frame of accounting, as well. Inferences about advertising performance are often made by using phone location data or in-store beacons – which collect data from phones via Bluetooth – to notice when audience members visit stores. Merchants and credit-card companies also share transaction data with advertising partners, who then match it against the consumer profiles they have attached to cookies or account IDs.

Attribution processes have material implications for media systems because they determine what gets counted, rewarded, and prioritized in the future. Last-touch attribution, the industry standard for years, incentivized direct-response tactics and the targeting of people classified as ready to convert (Interactive Advertising Bureau, 2017). Now, the industry has shifted toward ‘data-driven’ approaches to multi-touch attribution that fabricate claims about ad effects in the absence of complete observational data (Hercher, 2021). Attribution increasingly relies on AI models to cope with general complexities of omnichannel marketing and acute anxieties about ‘signal loss’, the industry’s term for talking about data protections that prevent deterministic matching of conversion events to identifiable individuals. Practitioners feel mounting pressures to delegate ad bidding, personalization, and delivery processes to platform-owned AI tools that optimize conversions with minimal predefinition of which audiences and media to target (McGowan et al., 2024). This state of the art cements the primacy of direct-marketing and performance advertising, and further institutionalizes platform-based attribution within the logistical circuits of the online economy. Platform companies dominate that economy, in part, because of their apparent capacities to determine the value of advertising.

To measure or to track: The subtle shift from audience measurement to advertising measurement

Attribution is a load-bearing pillar of digital advertising, attending to advertisers’ age-old anxieties about their uncertain power to influence consumers. And, yet, it depends on controversial, privacy-invasive forms of surveillance. To navigate this tension, parts of the ad industry are trying to differentiate tracking from measurement. Most famously, the consent management system in Apple’s app ecosystem now lets users reject ‘tracking’ by companies such as Meta that want to observe conversions across apps. Apple itself attributes conversions for advertisers and app developers, but, crucially, this is considered ‘measurement’ and exempted from the same consent requirement. Meta and Google promote their own ‘privacy-preserving’ attribution systems with similar rhetoric (McGuigan et al., 2023b).

The term ‘measurement’ is significant. It implies a legitimate purpose for surveillance. To put it coarsely, ‘tracking’ is exploitative spying by companies who follow users without consent; ‘measurement’ is necessary observation of users by a purpose-authorized entity. The latter borrows credibility from audience measurement – a longstanding institution in media systems (Ang, 1991; Meehan, 2005) – to help justify activities that shift the norms of the advertiser-media-audience relationship. Companies often describe attribution as advertising measurement, which collapses together record-keeping about the distribution of ads (audience exposure) with record-keeping about the effects of ads on consumer behavior. This sleight of hand helps the ad industry assert its right to implement a regime that not only quantifies audiences, but follows them into the marketplace. Industry spokespersons insist that since advertisers are indispensable patrons of public culture and information (Bindra, 2021; Mudd, 2021), advertisers need and deserve to know how their ad campaigns perform – not just whether ads are served and seen, but how efficiently they achieve business objectives. The implied bargain whereby audiences pay attention in exchange for access to content and services is giving way to an arrangement that evaluates audience members and media channels based on marketplace outcomes.

This framing of advertising measurement is part of a broader effort to sanitize adtech. Adtech companies are reacting to the discomfort people feel about commercial surveillance by incorporating privacy-enhancing technologies into attribution processes. These promise to anonymize the personal data that companies collect, reduce the number of parties accessing personal data, and limit the volume or detail of those data (McGuigan et al., 2023b). However effective in terms of cryptography and information security, privacy-preserving attribution nevertheless elides problems of surveillance. ‘To attribute is, by definition, to track’, write Pybus and Coté (2024: 6; original emphasis). ‘[R]egardless of a person’s anonymity, attribution cannot work without invasive and intimate tracking and reporting about all our behaviour to legitimise ROI for ad spending’ (6). Privacy-enhancing technologies do not change the fact that attribution assembles information flows from observations of both media audiences and marketplace consumers. Indeed, the premise that attribution could be ‘private’ should be taken as an invitation to interrogate how platform surveillance serves the mediation of social life.

The contextual integrity of democratic media

Adtech companies defend controversial data practices by appealing to normative values about media in society (Deighton and Kornfeld, 2020; Interactive Advertising Bureau, 2024a). The stakes in these disputes are accentuated by theories that consider privacy to be about the social relations surrounding the circulation and use of information (Nissenbaum, 2010; Viljoen, 2021). Helen Nissenbaum (2010) argues that privacy violations arise when information flows in ways that defy expectations about particular social spheres or ‘contexts’. She defines contexts as ‘structured social settings’ with historically situated purposes, priorities, norms, values, practices, relationships (including roles and identities), and power dynamics (130-132). People often feel uneasy when information produced about their roles, identities, and actions in one context is transmitted and put to use in another. This sort of cross-context flow must be justified with recourse to the values and priorities of the context from which the information derives. Health care provides a stark example: if its integral value is maintaining individual and public health, then any transmission of health-related information into, say, the commercial context of marketing, would be legitimate only if it served that or another defining priority of the health-care context.

My argument starts from the proposition that ‘media systems’ are social contexts. Media systems are complex but discernable arrangements of industries, institutions, and infrastructures of cultural production and circulation (Hesmondhalgh, 2019). They encompass many of the means by which people create, transmit, and receive ideas and representations of the world, including performances of personal identity (van Dijck, 2013). I do not pretend that it is uncomplicated to consider media as contexts, nor do I suppose that their defining characteristics are uniform and permanent – by definition, media come between things. Nevertheless, we can proceed by attending to some patterned ideas about the purposes, priorities, and values that orient our relationships to news, entertainment, and mediated sociality. Those ideas vary across times and places, and the diverse ways people use media prevent universalizing claims. My focus is on North America, acknowledging that some of what follows is applicable elsewhere, while some is contestable even locally.

Context collapse in media systems

Discomfort can arise when contexts ‘collapse’, when the spaces separating multiple spheres of social life dissolve (Davis and Jurgenson, 2014; Marvin, 2013; Shade and Singh, 2016). Carolyn Marvin (2013: 153) suggests that new communication technologies rearrange ‘customary social distances’ and thereby provoke anxieties, since the perceived stability of those social distances is ‘a key guarantor of social trust and a necessary feature of democratic societies’. Frischmann and Ohm (2023) argue similarly that the integrity of social life depends on the maintenance of socially constructed boundaries, or ‘seams’, that separate contexts and create friction against their easy transversal. In cases where seamlessness permits exploitation or other harms, seams can help align governance with contextual values. On the other hand, Davis and Jurgenson (2014) point out that all contexts are porous, with some amount of collapse being expected, tolerated, and even desired.

Media systems are especially porous contexts, since people use media to encounter and participate in many spheres of social experience. Normative debates about media systems, and some important institutional practices, hinge on the negotiation of seams, pores, or social distances. Famously, the journalism profession pursued public trust by establishing a metaphorical ‘wall’ to separate editorial and revenue-seeking processes (Lauerer, 2019). Controversies have erupted as that ‘church-state divide’ has been narrowed due to economic pressures and technological changes (Carlson, 2015; Couldry and Turow, 2014; Petre, 2021). A mix of new and old strategies, including native advertising, branded content, and increased reliance on performance metrics, have turned individual articles into ‘mini profit centres [for] new media capitalists’ (Cohen, 2015: 146), breaching norms regarding publishers’ responsibilities toward readers (Hardy, 2021). While a capitalist press has always overlapped with the market, consternation about seamless integrations of journalism and advertising suggests that people expect media systems to fulfill public-service obligations that exist independently from economic priorities (Napoli, 2019; Pickard, 2020).

Another integral feature of social contexts is the roles and identities adopted by or imputed onto people (Nissenbaum, 2010: 133). This essay is concerned with the complex identities of media audiences. Audiences have a long history of attachments to the ideal and practical roles of citizens and consumers, with political implications that vary in relation to class, gender, race, and private and public interests (Butsch, 2008; Gandy, 2002). Recognizing that most people are waged or salaried workers who must act as consumers to obtain what they need and want, it is clear that the roles of citizen and consumer are entwined. The hybrid citizen-consumer, a sometimes conflicted identity that illustrates ‘the permeability of the political and economic spheres’ (Cohen, 2003: 8), has been a key subjectivity for the construction of audiences within capitalist media (Banet-Weiser, 2012). Measurement is an important mechanism for materializing audiences and assigning value to facets of their identities (Ang, 1991; Hessler, 2021). For instance, as Meehan (2005) shows, early efforts to measure broadcast audiences – to make them institutionalized matters of fact – excluded all but the ‘bona fide consumers’ desired by advertisers. From broadcasts targeting ‘Mrs. Daytime Consumer’ as domestic manager (Spigel, 1991), to software interfaces and classification systems that render identity categories and behavioral propensities as buyable parameters (Bivens and Haimson, 2016), audience construction reflects and participates in processes of social reproduction (Smythe, 1981; Williams, 2003). In particular, the information systems of audience commodification have manifested and exploited measured differences (or similarities) in class, gender, and race (Rosa-Salas, 2025; Shepherd, 2014).

The ad industry treats the citizen-consumer as a conflated (rather than conflicted) identity in advancing the normative legitimacy of attribution. Justifications of advertising as integral to democratic media often presume that ads simply underwrite the communications that shape good citizens, and/or they collapse the poles of this identity by equating consumer sovereignty with democracy. A top executive at Google takes for granted that we appreciate the social value of news when he boasts that ‘our third-party ad tech plays a critical role in keeping journalism alive’ (Levy, 2021). Such vague invocations of public patronage sanction commercial surveillance as a necessary means to virtuous civic ends. That hint of democratic idealism lingers even as policy discourses privilege the private interests of companies and consumers. Advocating against restrictions on commercial surveillance, the Interactive Advertising Bureau (2022: 11) insists that ‘data-driven advertising is fundamental’ to supporting the ‘free and low-cost services that allow consumers to communicate, learn, connect, and access entertainment online’. The IAB centers consumer sovereignty and corporate productivity again in a more recent plea: ‘A privacy framework that prioritizes consumer rights while enabling companies to efficiently target consumers and provide effective measurement and attribution is essential’ (Interactive Advertising Bureau, 2024b: 38).

Basically, adtech’s defenders warn that sponsors who cannot calculate a suitable ROI will be forced to divest from the media that support democracy. Countering these claims should not lead us back to classed and gendered tropes about the irrationality of consumption and the rationality of citizenship. Plenty of consumption choices are careful, and lots of political activity is thoughtless or absurd. But how to balance the citizen and consumer roles, and their respective values and political commitments, remains a key question (Gandy, 2002).

I have no illusions of settling that question, but I will argue that platform surveillance and direct marketization further tilt the organization of media systems toward the latter half of the citizen-consumer identity, deepening the relationships whereby media systems recognize audiences according to customer lifetime value rather than equal rights to political power. Attribution requires users of ad-supported media to be constantly observed as potential purchasing subjects. Attribution systems, and advertising platforms generally, enclose users in an always-on market context that envelopes media systems and the other contexts articulated to them (Manzerolle and Daubs, 2021). The claim that information about media usage and marketplace behavior can flow together under the authorization of advertising measurement is a wedge for opening contextual pores so as to legitimate the exploitation of almost all mediated experience for marketing purposes. Platforms confront audiences as a surveillant and shoppable surround, wrapping porous social spheres in a seamless marketplace membrane. Frictions that help protect values of citizenship get smoothed to afford easier flows of data and commerce.

Media as contexts in democracy

Media serve many social functions, but it is fair to say that a prevailing purpose of democratic media systems is to sustain democratic publics (Ananny, 2018; Couldry and Turow, 2014; Pickard, 2020). As Jamelle Bouie (2023) suggests, ‘A vibrant press is one of the forces that helps shape individuals into members of a community with responsibilities and obligations toward that community’. The notions that that the mass distribution of facts and representations of reality facilitates informed self-governance and ought to serve public interests are engrained in North American political cultures and policy frameworks (Napoli, 2019; Pickard, 2020). Normative theories of the press are rightly contested, of course, so we should not romanticize journalism’s democratic contributions (Nerone, 2012). Nevertheless, commitments to public-service journalism remain widespread and productive of important, if imperfect, ideals (Cohen, 2015; Golding and Murdock, 2023; Pickard, 2023). Furthermore, while journalism provides bright lines for demarcating a media context, we should tolerate the added messiness of considering other fields of cultural production in conversations about the norms, values, and purposes of social mediation – since people find pleasure and identity-formation in news, and people learn about themselves and their worlds through entertainment (Williams and Delli Carpini, 2011).

The key point here is that media systems are critical social institutions. They facilitate mutual recognition among publics and political groups and they provide common symbolic resources and points of reference that people can use to negotiate commitments to shared values – all of which are necessary elements of democracy (Couldry and Turow, 2014). Efforts by marketers and platforms to classify and capitalize on the differential economic value of population segments and social situations run counter to that purpose.

Marketization and platformization

In media systems that depend on advertising revenue, the priorities of marketers influence news and cultural production in variably forceful or subtle ways (Baker, 1994; Griffin, 2023; Joseph and Bishop, 2024; Lauerer, 2019). Since the American advertising industry became insinuated with the mass circulation of information in the late 19^th century, leading publishers and broadcasters have conceived of their operations as ‘extensions of the nation’s marketing system’ (Kielbowicz, 1990: 452). By the end of the 20^th century, George Gerbner (1998: 176) described the industrialized storytelling of dominant media systems, which cultivated publics’ perceptions of reality, as being ‘the byproduct of marketing’. In many ways, the integration of media and marketing has deepened since then (Hardy, 2021).

Many researchers have observed that this market hegemony fails to produce the means and relations of communication required for democratic governance in complex societies (Ananny, 2018; Golding and Murdock, 2023; Hesmondhalgh, 2019; Mosco, 2009; Pickard, 2020). As Murdock and Wasko (2007: 2) put it, marketization undermines the ‘established organization of public culture’ and the democratic imperative for ‘people to acknowledge their common membership of political and moral communities’ (2). The supremacy of marketing priorities within U.S. media should not be mistaken for overwhelming acceptance of their merits. Marketization has been contested, and many accounts consider it to be the outcome of policy choices that placed corporate empowerment and capital accumulation ahead of social values (Gandy, 2002; McChesney, 1993; Pickard, 2014). Matthew Crain (2021) demonstrates that antidemocratic policymaking steered the particular commercialization of the internet toward surveillance advertising. Business-class interests crucially shaped the U.S. policy framework, while public input was largely excluded. The privacy ‘norms’ now encoded in digital media systems were established through corporate maneuvering (Crain, 2021; Draper and Turow, 2019), and their consequences have likewise been antidemocratic (Tufekci, 2014; Vaidhyanathan, 2018).

These concerns intensify as digital platforms stake claims around more social spheres. Platformization describes how platforms insinuate their economic priorities, governance mechanisms, and technological infrastructures throughout other systems of mediation (Poell et al., 2021). For example, news organizations and content creators accommodate to norms dictated by platforms (in policy or code) because the former depend on the latter to traffic audiences and advertiser spending (Caplan and boyd, 2018; Joseph and Bishop, 2024). While platforms ‘provide public services and constitute public spaces’, Nielsen and Ganter (2018: 1603) explain, ‘they do so for private gain as for-profit companies’. The infrastructures mediating ‘a de facto global public sphere with near-monopoly power over the social distribution of attention’ (Burrell and Fourcade, 2021: 230) are governed to realize commercial sociality, often contra public values and interests (Cammaerts and Mansell, 2018).

Advertisers have diverted enormous portions of their spending toward Google, Meta, and Amazon based, in part, on their perceived capacities to engineer consumption-probable situations and to fine-tune marketing efficiency (Carah et al., 2023; McGuigan, 2023). A key vector of platform power is their enclosure of audience and consumer measurement (Birch, 2023; West, 2019; Wu and Taneja, 2021). Platformization extends logics of connectivity and datafication across mediated domains, advancing ‘the underlying transformation of the internet into a space of continuous data extraction through surveillance’ (Turow and Couldry, 2018: 419). Jathan Sadowski describes ‘platforms as infrastructural intermediaries in everyday life’ (Sadowski, 2020: 14). Part of what they mediate (and track) is passage across contexts. Platforms subject the multiple social spheres they enclose to corporate terms of service that authorize commercial surveillance. The software integrations and data arrangements that enable attribution let the companies producing those metrics insert themselves into these various contexts and become party to the relations and activities happening there (Pybus and Coté, 2024). As Sadowski (2020: 13) puts it, ‘digital enclosure provides an insidious way to smuggle the rentier platforms’ ownership claim into things and places that are otherwise owned by somebody else’. With privatization masquerading as privacy, some limitations on advertising data flows actually increase marketers’ reliance on surveillant, shoppable platforms. According to the Interactive Advertising Bureau (2024b: 21), privacy compliance ‘is pushing ad investment to walled gardens where personalized messaging and conversion signals comingle to power closed-loop measurement’.

Contextual integrity helps us recognize adtech’s privacy discourses as statements about the normative values of the social spaces implicated in its dataveillance. If we accept that attribution can be ‘private’ (meaning it is appropriate to combine observations and/or inferences about what people do as media audiences and as marketplace consumers), then we have to admit that accelerating the circulation of commodities (and capitalizing data in the process) is as important to the purpose of media systems as representing and facilitating participation in shared, albeit contestable, social realities. While this has long been implicit in the political economy of U.S. media industries, it is being further materialized through extensions of platform surveillance and direct marketing.

Direct marketization: marketplaces, platforms, and private values

This section argues that platform-mediated shoppability and attribution are part of a process I call direct marketization. I use the term in a double sense. First, it describes the general ascent of direct marketing’s logics and techniques within digital media systems. Second, it not only suggests a greater influence of market forces on cultural production and online sociality, but it names ongoing efforts to make media directly coextensive with retailing environments. The medium becomes the marketplace as a practical matter of design, experience, and de facto policy (McGuigan and Murdock, 2015). Direct marketization reorients the social contract of popular media in a democratic society and helps normalize private corporate governance of more and more social spheres. The following paragraphs synthesize some historical and conceptual points about this process.

Direct marketing and consumer data collection have been entangled since the late 19^th century (Popp, 2021; Robinson, 2012). I will focus on developments since the diffusion of digital information technologies. As Frana (2023) demonstrates, computerized direct marketing cut a wide path for today’s data-driven technics. The intensity of ad personalization and analytics that direct marketing helped establish are ‘central to our understanding of how the attention economy came to be’ (26). A cluster of factors in the second half of the 20^th century – including postage rates, electronic data storage, market segmentation, and the innovation of ZIP codes – helped direct-mail marketing capture a growing share of advertiser spending from traditional media (McKelvey, 2022). ‘Direct mail bypassed newspapers and magazines’, Fenwick McKelvey (2022: 47) explains, ‘offering the possibility of more individualized messaging and, notably, disaggregating consumers from media’. Addressability, customization, and accountability – all hallmarks of direct mail – were integral to the commercial potential of the ‘new media’ that promised a slick convergence of television, print, telecommunications, personal computing, and networked databases (Barney, 2000). ‘With the help of database technologies’, Andrew N. Case (2015: 44) writes, ‘direct mail went from being a quiet backwater to the mainstream of the advertising world in the 1970s and 1980s’.

The expansion of direct marketing contributed to and benefited from movements to marketize communication infrastructures. Driven by pressures from business interests, policy choices in the 1960s and 1970s allowed private, profit-maximizing networking services to compete with common-carriers that had been bound to mandates of public service and universal access (Schiller, 1999). Contemporaneously, dramatic growth in direct-response advertising via toll-free long-distance calling strained the capacities of existing telephone facilities and parcel delivery and created lucrative opportunities for companies to target and monitor upscale-consumer segments (Popp, 2018). Direct marketing, Case (2015: 29) explains, ‘emerged by the 1980s as a highly professional and technologically advanced way to sell products and services to millions of consumers while also studying what they would purchase next’. Direct marketing’s transformation into ‘a system that accrued and studied consumer habits prefigured the type of data-rich marketing that today analyzes clicks on laptop screens and mobile devices’ (Case, 2015: 29). Across the 1980s, advertisers’ total spending on direct mail nearly tripled, from $7.6 billion to almost $22 billion (40). Advertiser spending across all direct-marketing media increased from $16.4 billion in 1977 to $25 billion in 1981 (42).

This growth went along with the hyped adoption of interactive media for personal use within the home and the outfitting of retail businesses with networked databases and various means of tracking customers (Gandy and Simmons, 1986). As Popp et al. (2025: 266) put it, ‘direct marketers looked upon the fast-changing communications landscape of the late-twentieth century and saw a shangri-la in the making’. Cable television, for example, amplified the ‘mobile privatization’ that Raymond Williams (2003) recognized as a catalyzing dynamic for commercial broadcasting and its articulation to consumer culture. According to Kathryn Brownell (2023: 6), cable ‘helped create a privatized public sphere where notions of “efficiency” and “consumer choice” reigned supreme, and earlier expectations that corporations had a civic responsibility increasingly faded’. Discussing how cable and interactive computer services affected the construction and experience of audiences, James Carey (1980) suggests that these ‘new media’ intensified privatization, eroded privacy, and undermined the possibility of public discourse. Carey describes how the double shift of publics into audiences and then into segments evacuates the public realm (see also Couldry and Turow, 2014). Considering how more of cultural experience happens in the private sphere, while more information about what people do in private is collected and exploited, Carey (1980: 43) writes, ‘The achievement of a social identity, that of a citizen, will lack a space within which it can be realized’.

Against the need for media to foster recognition of shared realties and mutual interdependence, segmentation and direct marketing organize cultural production and political participation around personal consumer interests and other ‘private’ identities. This current of direct marketization ‘privatized the very idea of a social relation’ (Carey, 1980: 9). These new media were ‘transactive’ as much as ‘interactive’ (Barney, 2000), and the collapse of social spheres via marketplace connectivity helped integrate nodes in the circuit of capital, opening those spheres for value extraction. As Nick Dyer-Witheford (1999: 81) explains, ‘In consumption, the integration of advertising, market research, point-of-sale devices, and just-in-time inventory control makes the monitoring of the consumer as integral to the production cycle as that of the worker….To be socialized is to be made productive, and to become a subject is to be made subject to value – not only as an employee but as a parent, shopper, and student, as a flexibilized home worker, as an audience in communicative networks’, and so on. Today’s ad-supported platforms advance these aspects of direct marketization on multiple fronts: enclosing public spheres and social life for monetization and datafication; exploiting dependencies on private infrastructures (e.g., cloud services, data centers, AI models, SDKs); and configuring people around consumer-indexed, often proprietary, identities.

Among its many implications, direct marketization renegotiates the implicit bargain in the marketplace for attention, replacing audience measurement (of evident attention) with advertising measurement (of consumer behavior). Advertiser support is often justified through a pretense of patronage: by financing media systems, advertisers act as patrons of public culture; audiences attend to advertising messages and get compensated with discounted access to cultural products, including the news and everyday imagery people use to develop and act out their capacities as citizens. Arguably, this arrangement gives advertisers some legitimate entitlement to surveillance confirming that their messages were delivered to the audiences they purchased. But digital advertising companies stake an even wider claim. The exposure- or attention-based model is being eclipsed by a direct-marketing paradigm wherein strategies and transactions are organized around discernable marketplace outcomes. As Donald MacKenzie (2024) writes, ‘much, perhaps most, advertising on phones is designed to get a “direct response”: an immediate goal, or “conversion,” in the form of a purchase, a sign-up or subscription, an install of a game’. This activates some particular norms and priorities. Unlike mass-media ads packaged within editorial content, direct-marketing messages have not traditionally been embedded in a separate context of news and entertainment. As standalone missives, the quintessential priority of their context is selling – not forming, informing, or inspiring democratic publics. Direct marketers have always worked to join information about individual addressees with evidence that those individuals took revenue-generating actions. This imperative is now central to the surveillance-based business models of ad-supported platforms. Examining Instagram, Carah et al. (2023: 382) point out that a significant dimension of platformization is ‘the historical and cultural process through which our practices of expression, private and public spaces…, and ways of making meaning and affecting each other become integrated into the attention, advertising and retail economy of the platform’. Instagram and other shoppable media (e.g., TikTok) deliberately orchestrate the collapse of media and marketplace contexts, trying to shrink the distance between ‘looking and buying’ (Carah et al., 2023). Attribution is the measurement paradigm for this project; it observes people as both audiences and consumers so as to confirm when looking turns into buying, and to optimize that sequence.

Today, virtually all ad-supported media feel pressure to match platforms’ abilities to demonstrate a ‘robust connection to commerce’, as one TV executive puts it (Steinberg, 2024). Platformization and direct marketization have worked together to realize a mediascape wherein attribution is not just desirable to marketers but essential to the terms of the trade. Platform companies like Google and Meta (as well as forebearers like DoubleClick) courted advertisers by positioning themselves as ideal vehicles for direct marketing (Crain, 2021). They promised to help advertisers assemble bespoke consumer populations and single out individuals with a high probability of conversion, irrespective of the cultural products that attracted their attention. As compared to the image-advertising that has been the focus for most debates about the role of ad-supported media in democracies, direct marketing carries a different set of responsibilities and normative commitments into the center of the digital world. Direct marketers are patrons of commerce (Shapiro et al., 2024), not patrons of public culture. Their objective is the efficient production of sales, not the sponsorship of media content whose topic or style serves as a proxy for the types of people in the audience. Optimization is the ideology, and ROI is the coin of the realm. The reputability, honesty, and social or democratic integrity of media systems hold less practical importance than measurable marketplace performance.

Novel attribution operations deepen these relationships. As mentioned above, digital advertising increasingly relies on AI-based analytics and optimization. Platform companies market themselves as the best stewards of advertisers’ investments within a more ‘privacy-centric’ digital environment, because their walled-garden surveillance and computational force seemingly let them determine advertising effectiveness without resorting to third-party tracking. Google, Meta, TikTok, and others are urging advertisers to let AI models manage campaign spending and personalized targeting. These ‘black box’ services essentially define audiences through attribution; advertisers simply specify budget constraints and conversion goals, and the platforms’ automated systems determine where and to whom to deliver ads, based on estimated conversion probabilities and cost effectiveness. Using attribution as the basis for audience construction and evaluation makes the ‘citizen’ part of audience identity – and the forms of mediation serving its values – completely subordinate to measures of consumption propensity. Many advertisers appear willing to accept this in exchange for proven performance (Scanlon, 2024; Schiff, 2023).

In short, direct marketing funds consumption situations, and attribution metrics determine their worth. This paradigm sits uneasily with the notion of advertising as a beneficent subsidy for democratic media. It not only puts more pressure on ad-supported media to avoid images of reality that are unconducive to a ‘buying mood’ (Baker, 1994; Griffin, 2023), but it lets advertisers avoid journalism altogether, in favor of more efficient and measurable means of achieving marketing goals. Today, retailers – not just Amazon, but also brick-and-mortar merchants – are increasingly active as digital advertising sellers, offering advertisers the benefits of first-party purchase data and proximity to the point-of-sale. Walmart, Target, and other merchants with large customer databases now operate ‘retail media networks’, adtech platforms that distribute ads on their own or other e-commerce sites/apps. Overall, retail media absorbed $114.18 billion of ad spending in 2023 (Rizzo, 2024). This is perhaps the starkest example of direct marketization, diverting funds away from cultural production and toward platforms whose raison d’être is to measure and optimize advertising effectiveness.

Conclusion

Attribution extends the commercial enclosure of social life by privatized platforms. The overhaul of media systems by direct-marketing techniques, metrics, and values has quietly renegotiated the relations among audiences, media, and advertisers. Direct marketization configures audiences as always-on consumers surrounded by an almost ubiquitous marketplace context, and attribution produces operational ‘truths’ for optimizing the circulation of commodities. Considered through the lens of contextual integrity, the claim that attribution can be ‘privacy-preserving’ requires accepting that it is appropriate for information about media usage and marketplace behavior to flow together, and thus that media and marketplaces should integrate seamlessly. Platform companies use privacy rhetoric to grant themselves a special right to administer ‘advertising measurement’, advancing direct marketization and tightening their grip on the production of attribution data.

Fortunately, adtech’s privacy theater provides an opportunity for radical politics. By claiming to do private attribution, platform companies have begged critical questions about the values, purposes, and priorities of the media systems they want to embrace within their scope of ‘measurement’. To judge the contextual integrity of attribution, we need to carefully parse ideas about what media are for in democratic societies. Allowing attribution to be normalized means admitting that media systems exist not to produce democratic publics, but to produce consumption and accumulate private value. This has been implicit in media governance for generations, but recent developments put it nakedly on the table. We can take this opportunity to challenge anyone authorizing these plans to reconcile them with the still-widespread expectation that media should be organized to advance public interests.