Audience-metric continuity? Approaching the meaning of measurement in the digital everyday

Introduction

Digital ‘media life’ (Deuze, 2012) is increasingly permeated by metrics. Social media platforms, apps and personal gadgets quantify an ever-expanding array of social life that has ‘never been quantified before – friendships, interests, casual conversations, information searches, expressions of tastes, emotional responses’ (van Dijck, 2014: 198). Metrics trigger complex processes of value, politics and worth accumulation within the framework of a digital economy (Mackenzie, 2018; Ruppert and Savage, 2011). Measures and metrics have become ‘environments in which we live [. . .] an “air” that we breathe, an atmospheric component of society’ (Brighenti, 2018: 25). As a naturalised, atmospheric component of media life, metrics form the wider context of what we in this article will call ‘the digital everyday’, that is, the wider routinised and taken-for-granted digital media landscape.

While metrics penetrate many spheres in society, they have long since been the basis for the business models used by commercial mass media companies. Digitisation, however, has brought with it a range of new business models, as well as new features for existing models, all of which build on algorithmically processed audience monitoring and tailored targeting of advertising based on predictive behavioural analysis (Bolin, 2011). Traditional forms of mass media have partly adjusted to these new business models (Bjur, 2009), while digitally born media were already based on measured audience profiling. With the latter has emerged a new type of jargon around the ‘data analytics industry’ (Beer, 2018), with emerging terminologies centred on ‘big data’ (Andrejevic, 2013; boyd and Crawford, 2012), ‘bio-metrics’ (Gates, 2011), ‘data doubles’ (Haggerty and Ericson, 2000), ‘gamification’ (Whitson, 2013), ‘social profiling’ (Gould, 2014), ‘data-base marketing’ (Turow, 2006), and so on. Digitisation has privileged new ways for the media industries to relate to their audiences, with the development of innovative textual strategies, such as click-bait journalism (Wahl-Jørgensen et al., 2016), and more specialised markets, such as ‘like economies’ (Gerlitz and Helmond, 2013). At the same time, the black-box nature of audience profiling (Pasquale, 2015) and the general level of abstraction in the handling of media user data have led to new translation practices, in which the subtle mechanisms of algorithmically based behavioural targeting are converted into the discourse of old audience targeting techniques (Bolin and Andersson Schwarz, 2015).

While digitisation has intensified the penetration of metrics in all spheres of life, as media researchers ‘we are still limited in our understanding of the lived realities of data circulation that are to be found within the mundane routines of everyday life’ (Beer, 2016: 86). Since metrics are so deeply embedded in the digital everyday, they pose methodological challenges for such an understanding, making it also difficult to attend to historical changes in the social, cultural and economic meanings of metrics. Our contribution with this article is to explore methodological avenues for understanding the role of metrics in digital everyday life. We do this based on experiences from a pilot study, in which we used artist Ben Grosser’s Facebook Demetricator as an experimental method, in addition to traditional qualitative methodologies, to trace user experiences. We also want to consider the contemporary relevance of these experiences against a longer history of audience metrics and measurement.

In the following, we first review previous research approaches to metrics as a part of everyday digital media use, in order to then, historically frame these in the context of audience metrics in the traditional mass media and advertising industries. Third, we discuss our pilot study. Using reflections from the interviewees in the pilot study and drawing on examples from a spectrum of responses, in the conclusion we argue for the value of further exploration of such issues in larger studies and for the use of experimental methods more generally when trying to understand everyday engagement with digital media.

Approaches to metrics and everyday digital media use

Metrics as a standard of measurement – or, following the Oxford English Dictionary, ‘a criterion or set of criteria stated in quantifiable terms’ – have been around, and evolving, since the early precursors of writing (Schmandt-Besserat, 1978). However, the rise of statistics in the mid-19th century introduced a new phase in the development of measurements – for example, for the administration of populations (Hacking, 1990; Porter, 1986) – and has increasingly set guidance for social behaviour and privileged the ways in which humans have oriented and acted in social space (Beer, 2016). Digitisation has arguably fuelled this development, introducing real-time algorithmic measurement in online spaces, building on continuously larger datasets, and devising new ways of calculating data, which have allowed for the synchronisation of sets of data on unprecedented scales.

Today, there have emerged several areas of empirical enquiry which have advanced theoretical approaches and methodological designs for the analysis of metrics. The literature in this area is rapidly expanding and covers large ground, ranging from more general methodological discussions on ‘digital methods’ (Marres, 2017; Rogers, 2013) to more specialised debates on ‘interface methods’ (Dieter et al., 2018; Marres and Gerlitz, 2016), along with ‘algorithm methods’ (Bucher, 2016), which both advance analytical tools and define interfaces, platforms, apps, and so forth as objects of analysis. Others have adopted processual approaches, such as clicking through an app’s functionality through a ‘walkthrough method’ (Light et al., 2018), the visual mapping of digital affordances of networked technologies like the blockchain (Velasco, 2016), or the production of ‘eventfulness’ through media experiments as a method to reveal events of novelty and disjunction in media production and circulation that would otherwise remain obscured in habitual everyday media use (Despard, 2016). Yet others have designed projects that are focused on the social effects of metrics, for example, by disrupting ‘the daily stream of online engagement’ through abstention from connectivity, in order to sensitise media users to the naturalised dimensions of digital media life (Kaun and Schwarzenegger, 2014). Such disruptive techniques have also been used by other researchers, mainly through forced abstention from social media use (e.g. Roberts and Koliska, 2014; Tiidenberg et al., 2017).

When it comes to the more specialised area of researching the impact of metrics on social life, the approaches can be roughly divided into two types: those which focus on the technology itself and those which focus on the users of technology. In the first group, diverse forms of platform-centred analysis are dominant. For example, Gerlitz and Lury (2014) analyse the numbers and measures used in the interface of the influence measurement platform Klout, in order to suggest that they perform the role of ‘participative metrics of value’. Similarly, Graham (2018) studies the numbers and ratings of TripAdvisor to argue how choice is influenced by rankings. Mackenzie (2018) focuses on coding practices as co-constituents of a process of creating the economic value of platforms, where metrics are central for the discursive and social construction of this value.

In the second group, most predominant across a wide range of empirical domains have been qualitative interviews and ethnographies in which scholars discuss with media users about their experiences and understandings of metrics. Using interviews with musicians who promote their work on social media, Baym (2013) emphasises the flawed objectivity and representativeness of metrics and the broad spectrum of interpretations found in the various contexts of social media use. This argument is reiterated in the vast body of scholarship on self-tracking, where applications and devices generate ‘truths about the body and the everyday [. . .] offering a standardised base for interpretative efforts’ (Pantzar and Ruckenstein, 2017: 7). Ethnographic research on self-tracking has problematised the discrepancies that exist between the individual everyday construction of meaning around personal data aggregation and the rigid, inflexible algorithms which enable it. Nafus and Sherman (2014) call the experiences of this discrepancy ‘soft resistance’, whereas Pink et al. (2018) reveal the anxieties and emotions of media users that emerge in the process of making metrics meaningful in their digital everyday. Similar arguments are advanced by Kennedy and Hill (2018), who use ‘experimental visualisation practice, social semiotic analysis, focus groups, interviews and diary-keeping’ (Kennedy and Hill, 2018: 834) to suggest that living with metrics means also developing a ‘feeling of numbers’. Such a feeling produces everything from an ‘intimacy of surveillance’ (Ruckenstein and Granroth, 2020), or fears and pleasures of being tracked online, to attempts by users to manipulate algorithms and guide content display by imagining how algorithms may work (Bucher, 2017; Velkova and Kaun, 2019).

Pantzar and Ruckenstein (2017) have adopted experimental methodologies based on the quantitative tracking of heart rhythms and qualitative diary-keeping to further reveal the flawed concept of ‘mechanical objectivity’, suggested by numbers in tracking technologies, against the lived realities of practice. They propose the term ‘situated objectivity’ to advance what they call an eclectic understanding of the everyday which accounts for the variations in individual interpretations and importance of metrics, measurement and personal data. Didžiokaitė et al. (2017) discuss how self-tracking and metrics are instrumentally used in everyday life to pursue short-term individual goals based on interview data, while Frigo (2017) uses an autoethnographic approach to argue that everyday self-tracking projects can inform lifelong goals and become forms of ‘life-stowing’. Metrics generated through everyday self-tracking can also be understood as communicative acts oriented towards technologies of tracking and towards people, as Lomborg and Frandsen (2016) reveal through interviews with Danish self-trackers.

Most of these studies approach the entanglement between everyday life and metrics from a phenomenological perspective, focusing on the lived experiences of users and metrics. Further studies include post-phenomenological approaches that address the entangled co-evolution of humans and technologies of measurement in the course of everyday engagement with metrics, co-creating ‘laboratories of the self’ where ‘calculative practices and metrics become part of the interpenetration of technological and human forces and agencies and how they might be challenged’ (Kristensen and Ruckenstein, 2018).

In sum, we can group the perspectives and methods that seek to analyse metrics into two broad categories. Software- and platform-centred studies focus on the technological platforms in themselves in order to analyse the affordances on offer for the user. This approach is (infra-) structural in that it suggests how technologies of metrication structure users’ everyday media use. Second, we have interview-based studies, often in combination with diaries or log data (e.g. from fitness trackers), which problematise these embedded assumptions about the meaning of numbers for everyday media use. They reveal complex practices of meaning-making evoked by numbers which individuals encounter. These practices are affective, sometimes irrational, and often counter-intuitive to the assumed metric functions. They are further nuanced through ethnographic approaches, including interviews, in combination with longer term observations of informants.

In this article, we suggest an approach that reconciles both these vantage points by a creative combination of traditional methods (e.g. interviews) with digital methods (in this case, metric manipulation software). In addition, since few have historicised metrification practices (for exceptions, see Crawford et al., 2015; Frigo, 2017), we also wish to analyse metrics in light of metrification practices by the mass media, in order to understand which media user reactions could possibly be new. To this end, we distinguish between operational and representational metrics. In the following section, we give a historical overview that will serve as a background to the testing of our method, by means of which we relate user experiences of metrics to historical continuities in audience management.

Representational and operational metrics in audience management

Arguably, there are important observations to be made in situating contemporary practices of metrification in a wider historical context. Metrics have always been important for the advertising industry, given its aim to obtain more refined tools to map target groups (Turow, 2011). We wish to concentrate on the fact that metrics always played a central role in the programming of commercial broadcast mass media content, serving as indicators for its popularity among specific audience segments. Television broadcasting has used a spectrum of technologies for audience measurement, such as sampling, audiometers, diary data and peoplemeters (Bermejo, 2009; Bjur, 2009). Most of these technologies within the media industries built on representational metrics (such as age, sex, occupation, education, income, household income and so forth), and were used to produce economic value from audiences and inform content programming, through which the experience of ‘watching television’ – or what Williams (1979[1974]: 86) calls ‘planned flow’ – used to be crafted. This uninterrupted stream of content encouraged continued watching. Thus, metrics became operational in that they became an organising element, as they informed the creation of schedules around what was expected to be most popular among certain target audiences: ‘the schedule is the locus of power in television, the mechanism whereby demographic speculations are turned into a viewing experience’ (Ellis, 2000: 26). ¹ Scheduling was a source of competition between channels and the locus of attempts to establish a cross-channel continuity in order to retain audiences (Ytreberg, 2002). However, since metrics and scheduling were primarily of interest to the media industries, audiences had a rather remote relation to these numbers.

With the proliferation of online communication from the 1990s and onwards, new types of metrics for measuring media consumption emerged. One of these was exposure on the web, which eventually became website visits, tracked through server logfiles and tags (Bermejo, 2009: 143). With this metric, advertisers came to be charged in relation to their reach into online audiences; for example, CPM (the cost of reaching a 1000 users) and performance of an advertisement (such as cost-per-click) (Bermejo, 2009: 139, 144; see Turow, 2011) were two measures that were further developed by search engines, including links as an additional metric for authority and relevance on the web through algorithms such as PageRank, which led to the formation of a ‘link-and-hit’ economy (Rogers, 2002).

In the recent context of social media, the capture and measurement of user interactivity by algorithms remain at the centre of production of the audience commodity. Algorithms track patterns of media use (Bucher, 2012) and generate metrics, providing not just ‘information to users but also users to their algorithms’ (Gillespie, 2014: 173), analogous to how broadcast mass media provide audiences to advertisers. Hence, algorithms have replaced manual content scheduling through buttons and counters that ‘metrify and intensify affects’ (Gerlitz and Helmond, 2013: 1349) and link media use with different web economies. Buttons (e.g. ‘Like’, ‘Notifications’) represent ‘composite metric[s]’ (Gerlitz and Helmond, 2013), which work with the affective register of user interaction. They concentrate and distribute data traffic, producing satisfaction rather than relevance (Gillespie, 2014: 175), or, perhaps more accurately, they produce satisfaction through perceived relevance on part of the user. Historically, this is not new. Television broadcasters have always worked towards audience satisfaction by aggregating metrics that inform scheduling. The difference is that if content used to be important for mass media companies in order to attract and sell audiences to advertisers, in the context of social media platforms, metrics are important to prompt users to craft their own content and continue engaging with the medium, in turn producing the data needed for targeted advertising. Thus, the two major developments in the context of social media are the move towards the automation of programming and making certain metrics visible to users in order to prompt engagement.

When metrics are made visible, users are prompted to make judgements about content or people they encounter in their social media feed, and to express these judgements through the means that the platform offers (Gerlitz and Lury, 2014). Simultaneously, user reactions and judgements are captured, valorised and used to make further predictions about users’ potential engagements with the platform (see Berman and Hirschman, 2018; Espeland and Sauder, 2007). As Gerlitz and Helmond (2013) note, ‘a like is not a means in itself, but designed as an ongoing and potentially scalable process’ (p. 1359).

When discussing metrics in social media settings, it is thus instructive to distinguish between representational metrics, which have the function of driving user engagement and encouraging activity, and operational metrics, which are the basis for algorithmic processing and behavioural targeting. Representational metrics are the numbers the user sees on the interface and acts in relation to. Operational metrics are the ‘under the hood’ metrics that are hidden from the user but nonetheless instrumental for managing the advertising flows and other things displayed to the user.

But how do visible, representational metrics re-shape the use of social media? Espeland and Stevens (2008: 416ff) argue that numbers are granted authority. But they are also a ‘reference point’, and part of the way in which people value themselves as well as others (Esposito and Stark, 2019). However, such social processes of valuation are naturalised in everyday media use, and they can be very hard to get at through traditional methods alone.

The meaning of audience metrics in social media: a pilot study of demetrication

If metrics comprise the environment in which we live, and the air we breathe, as Brighenti (2018) argues, how can we reflect on the role that they play in the digital everyday? The methodological designs that have dominated the study of users’ experiences of everyday metrics have taken their point of departure in the presence of metrics. These approaches assume the use of representational metrics as a reference point against which user experiences can be studied, but if metrics are largely naturalised, and in addition hidden from the user (as operational metrics), we would need methods that could sensitise our informants to their workings. Therefore, we experimented with an approach based on the absence of representational metrics.

In the spring of 2018, we conducted a pilot study based on artist Ben Grosser’s browser extension Facebook Demetricator, which removes all numbers from Facebook. Introduced in 2012, the extension strips the amount of ‘“likes”, comments, shares, friends, mutual friends, pending notifications, events, friend requests, messages waiting, chats waiting, photos, places’ (Grosser, 2014), merely indicating that ‘others liked this’ without mentioning how many. One of the questions Grosser (2014) posed was, ‘When users see everyone else’s “likes”, how does this knowledge affect their propensity to give a like?’ Facebook Demetricator is also supposed to trigger self-reflection:

Demetricator invites Facebook’s users to try the system without the numbers, to see how their experience is changed by their absence. With this work I aim to disrupt the prescribed sociality these metrics produce, enabling a network society that isn’t dependent on quantification. (Grosser, 2012)

Our goal was to explore in a more systematic way what kinds of reflections and experiences would be generated by the absence of these metrics. For this purpose, we recruited experienced social media users willing to instal the Facebook Demetricator extension for at least 3 days, after which they participated in a semi-structured interview. The interview guide covered four areas: first, it posed questions about contextual factors, such as how long the informant had used social media and if they considered themselves passive or active users (i.e. if they were mainly reading or if they were frequent contributors of content). Second, the guide featured questions related to the informants’ experience of using Facebook Demetricator. Third, there was a more focused set of questions around the phenomenon of ‘fear-of-missing-out’ (FOMO). Finally, the informants were asked if they would continue using Facebook Demetricator – or, if not, why. All in all, seven interviews were conducted with informants between the ages of 20–30 who had used Facebook for 7–9 years. Two were male, four were female, and one did not want to disclose their gender. Each of them had Facebook Demetricator installed for 3–9 days.

The interviews were held within the framework of a Masters course on the ‘Metric mindset’ (given by one of the authors). After being trained by us, students recruited the informants and conducted and transcribed the interviews. Informed consent was secured, including the right to discontinue the experiment and/or the follow-up interview at any point, should participants so wish. Informants were recruited through the snowball method of sampling (‘friends of friends’). Interviews were anonymised by the interviewers before presenting them to us (resulting in greater anonymity on the part of the respondents but allowing us less control over the collected data). Interviews were done in several different languages: English, Swedish and German. The student interviewers translated the transcripts into English. The quotes used below have been double-checked with the interviewer/transcriber.

The pilot aimed at testing a research design to study social media metrics. We sought to learn if the method could produce data that could be analysed meaningfully. Hence, we used the interviews as material that could suggest how users make sense of metrics. As an example of such meaning-making among the interviewees (the informants are distinguished as I1, I2, I3, etc.), we can point to a significant sense of disorientation that the removal of metrics caused. Such disorientation can be analytically divided into three dimensions: temporal, social and value-related, although these are not mutually exclusive.

Metric continuity and methodologies that disrupt media use

Social media are naturalised in the everyday flow of unreflected practices that many people engage in, and there is thus a need for developing methodological approaches that can capture the nuances of the sense-making practices of everyday media users. This pilot study used a combination of disruptive technology (Facebook Demetricator) and traditional interviews, analysed within a framework of the historical development of audience metrification. The merits of this combination of methods are that it first sensitises media users to the meaning of metrics, in order to provoke reflections on them. By situating these meanings in a longer trajectory of audience interaction with metrics, the historical framework can help in identifying what may be new types of audience-metric relations. Pilot investigations such as this can thus generate more precise ways to formulate questions around social media use, which can then be followed up in larger studies. The interviews we conducted reveal some such questions around types of audience disorientation and metric continuity, and we shall briefly discuss these below.

Qualitative audience approaches like this one do not, of course, strive towards establishing structural regularities; instead, they seek to identify a spectrum of nuanced responses to a given phenomenon. Although pilot studies can never fully cover this spectrum, the different forms of disorientation that we have identified among the interviewees indicate the importance of representational metrics for social, temporal and value-related orientation to social-networking media, indicating a felt need for curated content. This felt need can be explored further. In the age of mass media, curating was conducted by programming and scheduling staff, but in the world of social-networking media the curating has been delegated to users, facilitating their production of the representational metrics against which action in digital space occurs. The felt need for content to be curated seems to be one of the continuities that can be observed among our interviewees. We hypothesise, then, that representational metrics fill a similar function for social media use as the programmed ‘flow’ (Williams, 1979 [1974]) in traditional broadcast media: an effect that is actively sought and expected by users. While the platform structures user behaviour and prompts reactions through metrics, so do users seem to use metrics to craft and give an imagined structure to their own experience and dynamics of the platform. Such dynamics are hard to capture through ethnographic observation, whereas they can be grasped in interviews through reflections prompted by the absence of metrics. We further hypothesise that, contrary to top-down scheduling of programmed flow in mass media, in social media it is not the platform but users who establish a continuity in using media content (albeit within the technological framework set up by the platform).

A second continuity concerns the role of metrics in defining the experience and ‘identity’ of the medium – as a social-networking medium, as a news medium, and as a digital content-sharing platform. The use of Facebook Demetricator made apparent the difficulty to identify social media in relation to a specific experience of a medium, which our users would otherwise do with the help of metrics. If television schedules were produced by aggregating metrics of anticipated watching, and in this way defined the experience of watching television, we hypothesise that social media users today use representational metrics to produce an individuated experience of social media as a medium. Facebook Demetricator exposed a sense of disorientation when our users could not define what kind of medium Facebook was: when it no longer behaved as a news medium due to the lack of timestamps, when it was not social as it did not measure sociality, and when it was difficult to estimate the value of its content.

Facebook Demetricator also made apparent the ‘investments in numbers’ that Espeland and Stevens (2008: 417) discuss. The growing literature and methodological approaches to metrics have already accounted for the broad spectrum of meanings that users might apply when they interpret representational metrics. Facebook Demetricator further illuminates how valuable certain representational metrics are when it comes to maintaining the individual’s own understanding of what social media is about. Other representational metrics also have an operational meaning, in particular those related to producing reactivity, such as the number of notifications. From the perspective of the platform, notifications are made visible as a representational metric of different content-related events, aggregated by algorithms and suggesting relevance with the aim to engage (Gerlitz and Helmond, 2013). From the perspective of the user, however, the same metrics have an operational meaning; when they disappear, some individuals feel forced to find their own ways to measure eventfulness and engagement, in order to create meaningful navigation within the platform.

It is helpful to make such aspects apparent, as they indicate a persistence of older patterns of audiencehood that might shape the ways in which users orient when using social media. As audience profiling and scheduling have become automated and algorithmically driven in social media, some users derive assistance from numbers in creating their own consumption repertoires. In that sense, a major change which we can hypothesise is the fundamental role that numbers – and not types of content – play in the creation of meaningful experiences on social media. Therefore, the major significance of our experiment is in pointing to certain continuities against the longer trajectory of the historical development of the meaning of metrics in the media industries. This could help to create a better understanding of the lived realities of people immersed in new metricated environments, and at the same time keep such understanding historically contextualised.

Conclusion

In this article, we have discussed a methodological approach which combines a form of digital experiment with traditional, qualitative methods to gain insights into everyday user experiences of metrics in social media contexts. As argued in the ‘Introduction’ section, the predominant research on digital metric cultures to date has either focused on technology or its users as the vantage point to study experiences of metrics in different digital contexts. Our methodological approach suggests one possible way to reconcile both of these, in order to gain new empirical insights into the experiential dimensions of a deepening quantification of social life in the digital everyday, first by de-naturalising the affordances of a digital medium and then by qualitatively tracing user experiences in such a changed technological context. This approach resonates with Kennedy (2016) and Marres’ (2017: 112ff.) observation that contemporary digital research requires both a critical and creative approach, which combines and experiments with traditional and emergent methodologies with the aim of advancing social enquiry related to digital contexts.

Through our use of Ben Grosser’s Facebook Demetricator, we reconsider digital methods to not just include ‘methods of the medium’ (Rogers, 2013), but also playful, artistic tools that do not require computational or coding skills on the part of researchers or users; instead they represent ‘ready-mades’ that can be integrated and potentially repurposed to different ends in research. The main advantage which we see with this approach is to expand and complicate the range of questions that can be answered with digital methods, while also contributing to a methodological renewal of digital media research. As Venturini et al. (2018) admit, digital methods as they are currently used and defined make it possible to answer only certain questions, which do not necessarily generate interesting research even though they may be societally relevant. For example, digital methods struggle to address questions of experience, and their proper functioning is often reliant on the aggregation and use of metrics from online media use. Here, we demonstrated instead how digital methods can be creatively employed as a disruptive method in research that departs from a rejection of quantification and metric cultures and productively employs this rejection to study the very experience of living in metric cultures.

In order to extend the knowledge on the role of metrics in the digital everyday, we have focused on the domain of social media metrics and proposed an analytical approach which brings together audience studies, platform studies and digital experimental methods in a historically grounded methodological framework. Our point of departure was that deeply rooted cultural perceptions within the media industries might sensitise continuities that could be followed into new business models based on behavioural targeting. Our goal has been to develop an approach for analysing metrics and social media use, in relation to earlier models of organising content and to historicise the meaning of metrics in the context of a longer trajectory of media and measurement. We hope that this approach could be further tested and enriched to produce a greater understanding of the social meaning of metrics across digital contexts, as well as the role it plays in crafting experiences of mediated sociality.