From audiences to data points: The role of media agencies in the platformization of the news media industry

Introduction

Viewers of the Netflix series Mad Men will probably remember Harold Crane. A young employee at the advertising agency with dark, brilliantined hair and heavy glasses, working in the media buy department, where his job is to choose the right media for the advertisements and campaigns designed by the cool consultants in the creative department. The character of Harold Crane was a supporting role, and the same can be said of the function and status of media buying at advertising agencies in the 1950s: mirroring the cast, storyline and set-up of Mad Men, the creative departments of advertising agencies were the largest and most prestigious functions in the early days of the advertising industry. Media buy was a subordinate and supportive function. However, this began to change during the 1960s, when the large advertising agencies set up specialized media buy departments. By the 1980s and 1990s, a number of international media agencies had branched out and established themselves as a growing industry in its own right (Pratt, 2006, 2007). From the late 90s, media agencies have grown side by side with the digital advertising market: in 1997, global quarterly Internet advertising revenue was 130 million US dollars; by 2019, it had risen to 29.9 billion US dollars (IAB/PwC, 2019), and in 2018 the turnover of the top 30 media agencies in the UK surpassed 1 billion UK pounds (Carder, 2018). Media buy is no longer something done by the young man in the office with no windows, but a billion-dollar global industry laying the tracks of the public sphere (e.g. Ferree et al., 2002; Rauchfleisch, 2017) by channelling advertisements to some media and some platforms and not to others. However, research on media agencies is sparse and mostly from the perspective of marketing or management studies (e.g. Bulearca and Bulearca, 2009; Green and Trevarsis, 2002), leaving us with very little knowledge of the societal role of media agencies, and how their practices influence the digital media ecology and different sectors. As a starting point, we can understand media agencies and media agency employees as cultural intermediaries as they ‘construct value, by framing how others – end consumers, as well as other market actors including other cultural intermediaries – engage with goods, affecting and effecting others’ orientations towards those goods as legitimate’ (Maguire and Matthews, 2012: 552, see also Boyle, 2018; Cronin, 2004; Loacker and Sullivan, 2016). Media agencies sell their services to many different sectors from online banks to fashion webshops and everything in between, and the largest media agencies are global companies with regional/local national offices. This article focusses on a particular sector, the news media industry, in order to describe and discuss the role of media agencies. At the same time, the study also serves as a more general case study of platformization, and how programmatic buying are transforming audiences to data points.

About the news media sector, we know that it has been challenged in the past 20 years as a result of the increasing digitization and datafication of the media industry (Arsenault, 2017). This puts pressure on the traditional business model of journalism, which is selling advertising space (Newman et al., 2019). Newspapers traditionally had a privileged position in a two-sided market where they would connect advertisers with consumers, but with the introduction of online platforms the market becomes multi-sided, with a range of different opportunities for advertisers to reach consumers without connecting via news media (Nieborg and Poell, 2018). This has again led to news media trying out different forms of collaboration with online platforms (Rashidian et al., 2018, 2019), which has been described as a conversion of online news and social media (Braun and Gillespie, 2011). Digital advertising is much cheaper than buying material space in a printed newspaper or magazine, which also has an effect on the size of national advertising markets, since most advertising revenue goes to global platforms and tech companies rather than national news media companies: when a citizen opens her daily online newspaper, more than 200 companies will on average be working with the data she generates, and only a very small amount of the revenue generated will go to the news company (Lindskow, 2016).

This article explores the role of media agencies in this new digital media infrastructure (Plantin, 2019; Plantin and Punathamekar, 2019) where social platforms – especially ‘the Big Five’ – are at once technological, economic and ideological forces that shape our everyday lives, our social practices and our institutions and companies (Van Dijck et al., 2018). The study contributes to the theory of platformization (Van Dijck et al., 2018, see also Gillespie, 2010) and digital media infrastructures (Plantin and Punathamekar, 2019) by looking at one of the less visible actors in this new platform ecosystem, the media agency, asking: Do media agencies contribute to creating public value, do they enhance or curtail ‘platform mechanisms’, and are they in fact ‘redefining media’, as one company’s tagline says (Carat, 2021)?

The platformization of the news industry

Like many other industries, the news industry has become increasingly digital in recent years, and both media companies and advertising companies work increasingly with big data. In a discussion on whether big data is an opportunity or a threat to communication in the public sphere, Holtzhausen (2016) argues that ‘one of the major purposes of datafication is communication’ because it ‘will enable strategic communicators to micro-target their communication to specific individuals without making use of any mass communication platforms’ (p. 23, see also Binet and Field, 2007, 2009). This naturally poses a serious threat to the existence of ‘mass communication platforms’ such as news media and therefore potentially a threat to democracy. In the quote below, Couldry and Turow (2014) argue that this individualized communication is a threat to the ‘shared social and civic space’ and thereby a threat to democratic deliberation:

[. . .] big data’s embedding in personalized marketing and content production threatens the ecology of connections that link citizens and groups via information, argumentation, empathy, and celebration as members of a shared social and civic space. Unless challenged, these developments risk eliminating the connective media necessary for an effective democracy.

For Couldry and Turow (2014), the development is not the result of evil intent on the part of big corporations, but ‘[. . .] an unintended side effect – a negative externality – of how advertising, big data, and content production have come to coexist over the past two decades’ (p. 1710). Others are less concerned with the development of this fragmentation of the public sphere, and for instance Harper (2017) argues that ‘[. . .] fragmentation can encourage participation from the otherwise disempowered’ (p. 1424). However, we have very little empirical research on the changes that the public sphere might be going through in these years. Research is still testing the existence of ‘filter bubbles’, a term describing different segments of the population isolated from each other in digital opinion silos, rarely being exposed to news or views other than those of their peers (Pariser, 2011), and at the moment there is evidence both for and against, as well as different interpretations of the phenomenon (e.g. Bechmann and Nielbo, 2018; Colleoni et al., 2014; Flaxman et al., 2016).

The relatively new theoretical perspective of platformization is promising in relation to researching and understanding the role of digital platforms in shaping the public sphere. Platformization has been applied in various areas of business studies (Nieborg and Poell, 2018) but is also used in other fields such a healthcare (Bygstad and Hanseth, 2019; Charitsis, 2019) and the financial sector (Westermeier, 2020), and can be defined as ‘the rise of the platform as the dominant infrastructural and economic model of the social web and its consequences’ (Helmond, 2015: 1) or as ‘the penetration of economic, governmental, and infrastructural extensions of digital platforms into the web and app ecosystems, fundamentally affecting the operations of the cultural industries’ (Nieborg and Poell, 2018: 4275). Whether speaking of a ‘rise’ or a ‘penetration’ and whether including an explicit governmental aspect or not, both definitions point to platformization as a process where platforms become the dominant infrastructural and economic model of the web, bringing with it fundamental changes – also for the cultural industries and thus the news media industry. But how does platformization take place? Van Dijck, Poell and De Waal (2018) point to three so-called platform mechanisms: (1) datafication, (2) commodification and (3) selection. Central to their analysis are the powerful Big Five – Alphabet Inc., Facebook, Inc., Apple Inc., Microsoft Corporation and Amazon.com, Inc. – which they label ‘the infrastructural core of the ecosystem’ (p. 29), as they own the vast majority of the most important infrastructural platforms such as search engines, social media sites, advertising services etc. and are also behind the many sectorial platforms that serve specific communities, for instance the platforms used in newsrooms for A/B testing headlines online in real time. In fact, Napoli and Caplan (2017) convincingly argue that these companies increasingly produce and increase content to such a degree that they cannot claim that they are merely ‘tech companies’ when they are indeed what we should understand as media companies (see also Hess, 2014; Napoli, 2014). Following this, we can assume that the three platform mechanisms; datafication, commodification and selection, are productive in many different sectors but in the following we will look specifically at the news media sector.

In the news media industry, the first platform mechanism, datafication, is not least descriptive of how news organizations work with audiences. Data on audiences has always been vital for news organizations (see Ang, 1991; DeWerth-Pallmeyer, 1997; Ettema and Whitney, 1994), as their business model is not selling journalism but in fact selling audiences as ‘products’ (Napoli, 2016; see also Smythe, 1981). The development of the way in which news media construct their audiences and the effect this has on democracy are closely connected to developments in the advertising industry, which has been dominated by an increasing use of segmentation and targeting tools since the 1950s, leading up to the individualized digital marketing strategies of today (Turow, 1998, 2008, 2011). Harper (2017) argues that, with big data and algorithmic processing, the construction of media audiences as consumers becomes the dominant ideology. Where news media historically could keep up the so-called ‘state and church division’, constructing ‘readers’ or ‘viewers’ in the editorial departments while at the same time selling ‘consumers’ in the marketing departments (Author removed), these audience constructions are no longer exclusively the marketing tools of content-producing media outlets, but now increasingly a platform service provided by the Big Five. A potential problem is that the Big Five companies, products and services do not carry with them the same publicist concerns or public values that characterize journalistic media. In fact, we might assume the opposite: that the Big Five have an interest in constructing us as consumers and products and not in constructing us as citizens by providing us with the best tools for public deliberation or information gathering. In this way, datafication is closely linked to – and according to Van Dijck et al. (2018: 37) is also intensifying – the second platform mechanism, which is commodification. Commodification is never neutral or democratic, but is almost by nature a stratifying process, as some customers will always be more attractive than others to some advertisers. Turow has warned against ‘marketing discrimination’ where some customers are offered some (read: better) deals that others are not offered (Turow, 2008, see also Craig, 2004), and Fourcade and Healy (2017) argue that modern organizations increasingly follow a ‘data imperative’ to collect as much data as possible, making it possible to classify and stratify audiences, which leads to an ‘economy of moral judgement’. Ruckenstein and Granroth makes the case that ‘[. . .] the digital infrastructure, of which personal data collection is an integral part, has become naturalized’ (p. 14) and call for collective responses to this ‘undermining of public culture’ (p. 12). Parallel to these arguments, Harper (2017) proposes that we are witnessing a transformation of the public sphere where big data induces a ‘tyranny of the majority’ and produces a ‘hegemony of consumption’.

Commodification is also related to the third platform mechanism of selection or ‘curation of most relevant topics, terms, actors, offers, series, etc.’ (Van Dijck et al., 2018: 40). Selection of news stories and angles used to be the professional privilege of journalists and editors, but, with more and more people getting their news online, and with average citizens not always distinguishing between news sources, this professional practice of selection or curation is challenged by the many platform agents personalizing, recommending and moderating the flow of ‘news’. In a recent strategic case study, Nielsen and Ganter (2018) show how news media are increasingly becoming dependent upon new digital intermediaries and thus document how the relationship between news media and platforms is asymmetrical and a case of platformization. The argument of platformization can also be read in Tambini and Labo (2016) in their research on digital intermediaries. On the basis of secondary analysis of publicly available data, they argue that Google and Facebook have become ‘power brokers’ in the news industry as news consumption is increasingly online and as more and more news audiences are referred via digital intermediaries (Tambini and Labo, 2016: 44). This gives digital intermediaries power in ‘the ability to affect the content consumers see through a process of filtering, omitting and prioritizing certain types of information’ (Tambini and Labo, 2016: 44).

For Van Dijck et al., there is no doubt that platformization is taking place in the news industry, and they encourage ‘scholars concerned with the realization of key journalistic values [. . .] to squarely focus on the interplay between the different actors in the contemporary news process. Above all, this means shifting the attention from individual new organizations and single platforms to the ecosystem as a complex dynamic between infrastructural platforms and sectorial players’ (p. 71). One of these sectorial players that we need to know more about are the media agencies. On the one hand, we know that platforms are creating a new infrastructure, and on the other hand we know that journalism is suffering from this platformization, as it disrupts the news media’s business model. Media agencies are the missing link between these two observations, because they advise advertisers on where and how to put their advertising budgets. As such, the research questions guiding this explorative study of media agencies are whether media agencies contribute to creating public value, defined as ‘the value that an organization contributes to society to benefit the common good’ (Van Dijck et al., 2018: 22), whether their practices enhance or curtail platformization, and whether they are in fact ‘redefining media’, as one company’s tagline says (Carat, 2021).

Understanding media agencies: Empirical material and methods

According to Livingstone (2019), a job for critical media research is to question the claims of the data analytics industry. Performing critical research when the object is the commercial media industry can of course be a difficult task, but is not impossible, according to the advice given by Turow and Couldry (2018): ‘It is, nevertheless, quite possible to unpack the essential aims and workings of most of the activities through wide, careful reading of industry trade magazines, attendance at industry conferences, close analyses of the technologies used, and interviews of relevant executives’ (p. 422). The methodology of this study takes its lead from Livingstone, Turow and Couldry and is based on publicly accessible documents, both web pages of media companies and printed material such as branding magazines picked up in the reception areas of the companies where interviews were conducted, articles from trade journals (for instance Advertising Age and Adweek), online industry news sites (such as Campaign, The Drum, etc.) as well as conversations with people from the media and advertising industry. Not least, formal interviews were conducted with eight executives (CEOs, strategic directors and heads of digital departments) from six of the largest global media agencies with offices in Copenhagen in charge of strategic development of their organization, alongside around 10 informal interviews with other media agency employees and media industry people. The interviews focused on the role and practices of media agencies in relation to the news media sector in order to understand the news media sector in particular, as well as the role and practices of media agencies in general. Data for the study was collected in the period from 2015 to 2019, with most formal interviews conducted in 2016/2017 and desk research as well as informal interview conducted both before and after. The interview sample was designed to represent the big players in the industry. Accordingly, six media agencies were chosen from a list of the 10 largest global media agencies in 2016; they remained so during the entire data collection period (Campaign, 2019). The decision to conduct interviews in the Copenhagen offices was a logistical choice but with two weighty advantages. Firstly, the case has a representative aspect, as all companies have global and European strategies rather than national or regional strategies. As such, it is unlikely that the findings of the study would be very different had the interviews been conducted in, for instance, Spain or the UK. Secondly, the case might also have a ‘best practice’ aspect: Denmark is one of the Scandinavian countries where journalism is understood as a ‘public good’ (Allern and Pollack, 2019), which translates into a media policy where politicians traditionally support a strong public service sector and also provide public media support for privately owned newspapers. This means that an explicit ‘public value’ approach to news media and journalism is part of the culture in Danish society and the Danish media industry, and therefore Danish interviewees might be more reflective about the societal role of social platforms than interviewees from a country with a more commercial media landscape.

The semi-structured interviews lasted around an hour and had three themes: (1) the strategies of the media agencies; (2) the services provided and developed, which at the time of the interviews were very much focused around programmatic buying; and (3) the executives’ understanding of media audiences in general and news media in particular. Informal interviews varied in duration and alternated between themes. Questions were refined during the research project in accordance with the author’s accumulated insight into the media agency business and marketing lingo. All interviews were conducted by the author and transcribed by a professional transcription service. The subsequent analysis was structured using thematic coding informed by the analytical framework, and all interviews have been anonymized by agreement with the interviewees.

The findings in the next section are primarily based on the empirical material mentioned above, but also on informal conversations with people from media agencies, advertising agencies and news media (both marketing and editorial staff) met at industry conferences and professional gatherings or in private settings during the research period. These informal conversations were helpful not least as an ongoing testing of research questions, assumptions and understanding of the business.

From audiences to data points

So, what do media agencies actually do? Their core business is advising clients on where to place their advertisements and marketing budgets and executing the actual buy of media space. During the past decade, their business model has changed fundamentally as a result of digital technologies, from selling space in paid printed and electronic media to advising clients on media buy and enabling it on paid, earned and owned media on all platforms (Jensen and Sund, 2018). In a rare case study of a media buying process in Sweden based on observations and interviews, Ots (2009) finds different constructions of ‘customer value’ at play: ‘[. . .] the two customers in this study – media agency and advertiser – appear to value different aspects of the media offering and construct their understanding of value through different processes’ (p. 131). Following on from Jensen and Sund (2018) and Ots (2009), this study of media agencies asks what media agencies are advising and enabling, and, from Bolin and Schwartz (2015), how media agencies work with big data.

Pinpointing the business of media agencies, one of the interviewees explains: ‘Today, we are less of a buying machine and more about strategic planning’. For all of the media agencies, advising and planning are very much focused around programmatic buying. Programmatic media buy, or programmatic buying, is a relatively new phenomenon which was introduced less than a decade ago and which can be described as a new marketing approach where our digital micro-behaviour is traded in real time (see Gonzalvez-Cabañas and Mochón, 2016). It is sometimes used as a catch-all phrase covering different digital technologies and services (Adweek, 2013), but most definitions put real-time bidding (sometimes abbreviated RTB) at its core (see for instance O’Sullivan, 2015). For media agencies, digital advertising has many advantages for the advertiser compared to print and electronic advertisements: digital marketing can be measured; it comes at a lower price; there is less ‘waste exposure’, because you can fine-tune your target group, etc. This means that digital marketing has become the primary strategy for most of the major global media agencies today, and that the power of the media agencies has increased with the growth of the industry and with digitalization. One of the strategic managers puts it this way:

What has happened with digitalization is that there are crazy amounts of data out there which we can begin to use to see people’s actual and precise behaviour online at cookie level. We know what content [they like], what they actually consume, and all that data can be used to do meta-analyses. With this kind of data, we can do interfaces and database solutions and get a structure in the data. And this gives the media agencies more power than we used to have when advising clients. Our role is definitely more important today than back in the old days, when all we really did was to book a TV commercial.

The quote above addresses the new role and power of the media agencies, but also touches upon a change in the core service. In the old days – notably no more than 15 or 20 years ago – media agencies booked TV commercials or bought pages in newspapers and magazines. Today, they are in the business of structuring data and translating big data into social data (Bolin and Schwarz, 2015). As an illustration of the relational power of print vs digital, 14 out of 16 employees worked with digital media in one of the Copenhagen branches of a global media agency in 2017, one worked with print media and one with electronic media. Because digital media – and the way we use them and what we allow companies to see and do with our data – have become the major focus of the media agencies, the way that media agencies work with data has changed fundamentally. Until very recently, which is no more than 5 or 10 years ago, media agencies used to sell audiences as socio-demographic segments (for instance, high-income city singles). Now, they sell micro-behaviour in real time. As one of the strategic developers puts it:

We used to do surveys and ask people all kinds of things, their political opinion and a lot of other things which were defining for whether they belonged to a certain segment or not. Now, it is actually your behaviour in relation to a certain product that will define which segment you belong to.

When the media agency executive above speaks about ‘behaviour’, it is of course not our offline life he is referring to, nor is it any person in the sense of an identifiable human being, but the activity of a given IP address. This is a fundamental shift in how media agencies think about and work with consumers, and not least a fundamental shift in the knowledge that lies behind the construction of different target groups. Segments are no longer people sharing socio-demographic characteristics such as age, town, income etc., but people sharing buying profiles and likeliness to need or want a certain product or service. ‘Today, a segment can be people in the market for buying a car. Or a segment can be people who will soon be in the market for buying insurance for their house’, as one of the executives puts it. The fundamental change in the data processing of the media agency industry is visible in both examples in the quote above. Ten or fifteen years ago, car dealers or car producers would market their cars to a segment which they knew would be interested in their brand, but they would not have a clue whether the people who were exposed to their commercial on TV or in print would (1) notice the ad, (2) read the ad, (3) react to the ad, for instance by discussing the car ad with their spouse, or even (4) go to the car dealer and finally (5) buy a car. With digital media, media agencies can sell ad space directed at people who are in the market for a car, or even a car of that specific brand, and track their exact online behaviour from interest to final buy. Together with other big data, for instance on geography, income group, favourite car colour etc., digital advertisements can be targeted more precisely, cheaply and measurably than a decade ago, when electronic and printed advertisements were targeted at segments. Segments, though, are still an important way of communicating the products and services of media agencies to clients, even though the data behind them is increasingly digital: ‘Segments are still called something funny and fancy which you can communicate, but the underlying data is behaviour’, as a media agency executive puts it. Or, put another way:

‘Old school segments’ is a communicative thing. We have to communicate who they are and who they would like to be and all that communicative stuff but how we hit them is based on their behaviour.

In the days of print and electronic marketing, news and entertainment media were vital to advertisers, as they offered not just neutral space for ads but a branded environment where ads could be placed. News and entertainment media were thus part of the message sent to potential customers. A highbrow luxury brand selling jewellery, for instance, would need an upmarket magazine, newspaper, television channel or television show which matched its values. Today, the same jewellery brand can reach us anywhere online and does not have to place ads with a specific media brand/outlet. This of course challenges the classic business model of most legacy media. Where media used to be context for advertising, that context is becoming less important as programmatic buying is increasingly based on behaviour: ‘The value of context is definitely decreasing. It is behaviour that is becoming important. Why is that? Because we can see that behaviour often trumps context’, as one of the interviewees put it. When media agencies direct advertisers towards programmatic buying, the effect on legacy news media and news outlets is that they lose their traditional function in the marketing chain as an attractive environment for advertisements. The term ‘context collapse’ describes the lack of context on SNSs such as Facebook and Twitter, where we communicate to an imagined audience which might represent very different contexts such as close family, distant friends and professional network, but we might also understand the role of media agencies with inspiration from the concept. Context collapse has been explained as an affordance of social media (Beam et al., 2017; Marwick and Boyd, 2011, see also Costa, 2018). Looking at the practice of media agencies, we might understand their current focus on digitizing and commodifying the audience as effectually producing context collapse: when advertisements are increasingly targeted to our IP addresses and not to the media brands that we are using, this means that news media’s traditional role as context for advertisements collapses.

As argued earlier, digital marketing tools have many advantages compared to other techniques. From 2015 to 2019, the industry was very much dedicated to developing new digital tools and products, and algorithms were considered by many to be the new black. One executive boldly said that ‘In the future, all media buy will be based on algorithms’, and whether this proves true or not, it illustrates the preoccupation with digital marketing tools. At the same time, a certain ethical wariness could be detected in the interviews. In the quote below, an executive explains in detail how they target specific segments:

We tend to become a little NSA-like. . . For us, it is a question of tying as much technology as possible together so the data we have on you – not your name, but your IP or a hashtag – this becomes a total picture so that we can draw you and communicate to you in the right way in all of your touchpoints.

The CEO above is very descriptive when trying to explain how the company works with data, trying to gather and tie together as many data points as possible. It is also interesting that he uses the example of the NSA when trying to characterize the methods – and smiles when he says it. The quote and situation are typical of how all of the executives expressed a two-sided understanding: on the one hand, what might best be called a romantic ‘data crush’ and a determined preoccupation with digital marketing tools, and on the other hand a detectible reservation towards the same strategy –expressed either as an ironic distance or as more explicit ethical reflections.

In interviews, democracy is naturally not ‘top of mind’, to use a marketing expression; least of all is it a concern of the media agencies, who are commercial companies. But when prompted a bit, some of the media agency executives offer critical reflections on the societal role of the industry that they are part of. Asked about the legal frameworks and regulation around big data (at the time of the interviews, a new European regulation on data protection – the GDPR – was first discussed then introduced which became a theme in many of the interviews), all interviewees felt that laws were either unclear or not up to date with technological development, which led to a feeling of sometimes working in a grey zone. One executive was especially reflective and (self-)critical about this because the global media companies, in his observation, thus operate above the law:

It is interesting that the large global media companies who are heading the technological development seem to work above the law. [. . .] I think the problem is that the political system is in no way involved any more.

In the quote, the executive describes how he sees the media agencies working above the law because the politicians are not able to follow developments sufficiently, perhaps because of their complexity, perhaps because of the pace of developments or perhaps a combination of the two. This might be understood as the result of political de-synchronization, a concept from Rosa (2010) describing the mismatch between the high rate of technological development on the one hand, and the slowness of political and democratic processes on the other, a phenomenon which, according to Rosa (2010), poses a threat to democratic societies: ‘If we want to be basically democratic societies, this means that politics regulates the frames and broad directions within which science, technology and the economy operate. However, this [. . .] is based on the assumption that political decision-making and social evolution are, or at least can be, synchronized’ (p. 71). In relation to the media industry, Rosa’s definition of democratic societies as social spaces regulated by politics points to the lack of political focus on the tech giants in general, and more specifically to the fact that the core business of media agencies is changing as a result of a technological development which is hardly regulated. This political de-synchronization is of course not only the result of technological innovation and sparse regulation (the EU Commission’s discussions about regulating tech giants might serve as an example of how big the challenge is even for politicians and political institutions with good intentions), but is also the result of consumer agency and driven by the public accepting digital cookies and allowing their micro-data to be resold again and again – or using ad blockers. A media agency executive points to the paradox that the public give away their data for free on social media without understanding the business model of platforms or the consequences for their personal lives, while at the same time believing that the Internet and social media are the best thing that ever happened to them:

People almost see Facebook as a human right [. . .] and in reality, it is just a machine selling your data.

At the same time, media agencies are also worried about the increasing awareness of data policies and digital marketing in the public. They see a growing tendency for the public to use ad blockers, which makes them concerned about future access to data. Needless to say, the use of ad blockers is seen as an obstacle, an annoyance and a threat to media agencies.

Discussion

Returning to the research questions guiding this study, do media agencies contribute to creating public value, do they enhance or curtail ‘platform mechanisms’, and are they in fact ‘redefining media’, as one company’s tagline says (Carat, 2021)? If we take the question of platform mechanisms first, the analysis above has illustrated how media audiences, segments and target groups are slowly becoming data points in the digital practices of global media agencies. Today, media agencies are deeply engaged in digitizing and tracking audiences by advising advertisers on how and where to place advertisements on or via different infrastructural and sectorial platforms. In this way, media agencies enhance the platform mechanisms of datafication, commodification and selection and contribute to the platformization of the news sector as well as other sectors. As to the question about public value, the interviewees are – not necessarily surprisingly, as they are serving not the public but their owners and shareholders – not very reflective or explicit. On the one hand, none of the interviewees offer a formal or explicit strategy about the public value or societal role of media agencies. On the other hand, all interviews show a concern or reflection about the rapid development of platforms and platform technologies: when interviewees speak about working in a grey zone, under the political radar, or with methods that might seem ‘NSA-like’ to outsiders, and when they express sincere surprise or concern that the majority of the public give away their data for free, it can be argued that this does in fact point towards a concern for public values. What they express is a concern that doing business in a grey zone is not ethically ideal, that it is problematic if business is conducted under the political radar, that surveillance methods are not legitimate in all cases, and that people should be more aware of their individual role in the platform economy. The last question, whether media agencies are really ‘redefining media’, can be answered positively: Yes they are. As the analysis has clearly shown, the media agencies are well aware that they are part of the digital reshaping of the media industry. Media agencies might not understand themselves as cultural intermediaries but as this explorative study of their practice has shown, media agencies are indeed cultural intermediaries producing symbolic value (e.g. Maguire and Matthews, 2012) and mediating culture and economy, and culture and consumption (Cronin, 2004). It would be productive with more future research on the work of media agencies seen as cultural intermediaries on both an individual level and on a sectorial level (as for instance Boyle, 2018).

The analysis has also shown, that media agencies neither understand themselves as media or content providers, but as Napoli and Caplan (2017) have argued, they are – and it matters that we understand them as such. Understanding platform companies as media, we can begin to discuss issues such as transparency, pluralism and accountability. On a structural level, relatively few companies dominate the media infrastructure, operating globally and adhering to different laws in different countries, making business practices less transparent and making accountability a moving target (Soederberg, 2010; Taplin, 2017). Mansell (2015) argues that the lack of transparency and accountability is a problem for researchers and public alike: ‘Before digital intermediary platform owners took centre stage, it was comparatively easy, though not uncontroversial, to assess which companies had incentives to operate in ways that would limit media plurality or abuse citizen’s fundamental rights’ (p. 13). The same concern is raised by Couldry (2017: 187), who argues that the surveillance aspect of the new digitally driven media industry imposes an external, corporate logic on social and individual processes, and Tambini and Labo (2016), who warn us about the implications of digital intermediaries for news plurality. Recent research has also pointed towards constructive solutions for countering the negative externalities made by platformization, for instance ‘algorithm transparency’ (Bracha and Pasquale, 2008) and ‘cooperative responsibility’ (Helberger et al., 2018), which are beyond the scope of this article. But, hopefully, this study has contributed to highlighting the importance of including media agencies in future research of negative externalities of platformization as well as in the discussion of constructive solutions.

Conclusion

In the past two decades, media agencies have grown to become key actors in the media industry advising advertisers on how and where to spend their budgets and thus shaping media markets and the digital media infrastructure. This article has studied the strategies of ten global media agencies through analysis of interviews with strategic managers, trade journals and publicly available material. The analysis shows how media agencies are deeply engaged in digitizing, quantifying and commodifying media audiences while at the same time aware of ethical challenges of the digital media infrastructure and in the process, they are fundamentally changing the infrastructure of the public sphere. Theoretically, we can understand these developments as a case of political de-synchronization (Rosa, 2010) where technological development moves faster than the political system, and also as a result of negative externalities of the media industry (Couldry and Turow, 2014). This makes it very important to begin discussing possible political responses to the development. On a general and international level, it is vital to address the business models of ‘the Big Five’ and how they affect the business model of media agencies as well as the business models in different sectors, and thus ultimately how they affect cultural production and cultural consumption. Regarding the news media sector in particular, we need more research on how digital technologies shape the public sphere, as well as more critical research on the practices of cultural intermediaries as media agencies in order to understand platformization and the effects on the new digital media infrastructure on democracy. Last but not least, the study might hopefully inspire to critical research on the complex relationship between media, platforms, advertisers and audiences and the role media agencies play in relation to different sectors.