Streaming culture and a new wave of institutionalisation of audience measurement in China

Abstract

This article examines the unfolding re-institutionalisation of audience measurement as the streaming culture reshapes the television industry in China. Following a brief introduction of the institutionalisation of the ratings culture, I examine how streaming platforms are reconfiguring audience measurement by tracing how various metrics are defined, used and valued by different platforms, in different contexts and across time. The analysis reveals continuing and new forms of informalities despite signs of convergence towards the use of algorithmic metrics, which are closely connected to streaming service providers’ self-serving interest in a multi-sided market. Next, I discuss how the disruption of the state’s near-monopoly in audience measurement in the traditional television industry breaks the regulatory inertia. I explain the dual-approach to institutional intervention by the state – the regulation of metric commodities through mandated calibration, and the launch of an official system as market alternative and policy instrument. I argue that the state intervention – through governance of and with algorithms – constitutes adaptive cultural governance serving ‘mass’ audience construction, and the regulation doubles up as a market response to serve the state’s intention to reclaim power in and over the market.

Keywords

algorithm audience measurement China institutionalisation metrics platform streaming television

Introduction

The rise of streaming culture has irrevocably pushed the boundaries of television. Changes in how audiences access and use television has rendered inadequate the established approach to audience measurement, which have long underpinned the media economy (Ang, 1991; Bolin, 2011; Napoli, 2010). In the history of television industry, ratings services have paralleled the development of commercial television and played a crucial role in audience commodification. As broadcasters, content producers and advertisers use ratings to measure content success and facilitate advertising deals, ratings construct what has been termed as ‘institutionally effective audiences’ (Ettema and Whitney, 1994: 5) for various stakeholders.

The rapid growth of video streaming services and their penetration into the traditional television industry have not only reshaped screen content consumption, distribution and increasingly production (Cunningham and Silver, 2013; Evens and Donders, 2018; Holt and Sanson, 2013; Jenner, 2018; Johnson, 2019; Lobato, 2019; Lotz, 2017, 2022), but relatedly, unsettled the market information regime. Streaming services have become a key site of audience measurement with a dizzying array of metrics. Traditionally focused on exposure metrics, audience measurement has increasingly incorporated various metrics of engagement and affect (Kosterich, 2016; Kosterich and Napoli, 2016; van Dijck, 2014). These metrics play an important role in meaning production and identity construction, to the extent their absence may cause user disorientation (Bolin and Velkova, 2020). The expanding scope of capturable data fuel the growing range of metrics, and increasingly, algorithmic mechanisms for measuring and seeking the audience (Fisher and Mehozay, 2019). As the streaming culture continues to reshape television, the negotiation over the value and meanings of various metrics, and approaches to audience measurement more broadly, is complicated and far from settled.

The institutionalisation of market information regimes has significant bearings on how participants make sense of the market and how the field might be changed (Anand and Peterson, 2000). As the streaming culture continues to reshape television, the changes and continuities in audience measurement, the politics of institutionalisation, as well as their cultural implications in various sociocultural, industrial and political contexts are understudied. These questions become more urgent considering streaming platforms’ role not only in audience measurement but also in content production and distribution.

To pursue this line of inquiry, this article foregrounds the politics of market information regimes in a wave of re-institutionalisation of audience measurement in the ongoing transformation of the television industry in China. China offers a productive site to study not only for its world’s largest audience, but more importantly because of the complexities and disruptions brought by the proliferating streaming platforms to the industry. While there is a parallel development in the rise of new market participants as is seen in the global market, China has its idiosyncratic cultural system, regulatory conditions, and industrial practices. It therefore offers opportunities to understand not only the common challenges in the changing field of audience measurement but also how the socially constructed process unfolds in the specific context. In the broader context of a turn towards state regulation of digital platforms across the globe (Flew, 2021), the case of China contributes to the literature which increasingly focus on how platforms govern, how they are governed, and what this means for societies. This article builds on existing literature on audience measurement, combined with platform studies, television studies and critical algorithm studies to examine the unfolding re-institutionalisation of audience measurement amidst the rise of streaming culture in China. The analysis utilises an array of data sources from corporate websites, promotional materials, industry reports, trade publications and policy documents. These materials are approached with a critical perspective, taking into consideration their inherent biases. Different data sources produced for various audiences and purposes are triangulated to yield new knowledge.

The rest of the article begins with a brief historical account on the institutionalisation of ratings culture in China in the context of media commercialisation and globalisation, and the stubborn ill of informalities as manifested in data manipulation and fraud in audience measurement market due to institutional inertia in a state-dominated market structure. It then focuses on the rise of streaming service providers and their approaches to audience measurement in the evolving television market. By tracing how various metrics are defined, used and valued by different streaming platforms, in different contexts and across time, I reveal continuing and new forms of informalities – inconsistencies, algorithmic opacity, arbitrary changes and data manipulation – in audience measurement despite signs of consensus in the shift towards algorithmic metrics. I explain how these informalities are closely connected to streaming service providers’ self-serving interest while navigating a multi-sided market. Next, I explain how the development in the platform market disrupts the state’s near-monopoly in audience measurement in the traditional television industry and breaks the regulatory inertia. I discuss the state’s dual approach to what constitutes the apparatus of adaptive cultural governance – the regulation of algorithmic metrics in the platform market, and the establishment of an official algorithmic measurement system. The analysis reveals the politics where the platforms’ attempt to draw on metric power and algorithmic governance to solidify their power yet meets the state’s institutional attempt to recalibrate it in both market regulation and cultural governance.

The institutionalisation of television ratings and its discontents in the near-monopoly era

Ratings as a western lexicon did not enter China’s television market until the mid-1980s. Before then, the main devices to gauge audience interest were letters and calls from the audience as well as sporadic interviews conducted by broadcasters in the context of limited competition. This scenario changed with intensified market competition (Bai, 2005; Keane, 2015). The decentralisation policy in the 1980s saw a proliferation of television stations through the establishment of a four-level (national, provincial, municipal and county) broadcasting system. In the following years, the development of cable and satellite TV technologies gave rise to a growing number of TV stations and channels, which intensified competition. Broadcasters started to survey audience viewing patterns based on statistical sampling, perceived as a more ‘scientific’ method (Zhang, 2009).

The introduction of what Ang (1991) calls a ‘market feedback technology’ (p. 7) was not without controversies when media operated solely as organs of the party-state apparatus. A major shift came in the 1990s with the commercialisation of the television industry unleashed in the broader context of cultural system reform (Keane and Zhao, 2014; Zhao, 2008). A strong demand for ratings service emerged as advertising, rather than state subsidies, became the main revenue source for broadcasters (Zhu, 2008). The commercial imperative in the market aligned well with the state’s intention to understand people’s viewing behaviour. In 1995, what was previously the arm under CCTV General Editorial Office coordinating the national audience surveys transformed into the China Viewing Survey Centre (CVSC), an independent enterprise fully responsible for its own business performance. This signalled an ideational shift from audiences as masses to be enlightened to consumers to be satisfied, although the state’s power to exercise ideological and moral oversight remains firmly in place (Zhao, 1998; Zhu, 2008).

The development of the ratings industry was further propelled with the entry of international market players and the introduction of people metre technologies in the late 1990s. Notably, AC Nielson entered the China market in 1996. This was quickly followed by the establishment of CVSC-Sofres-Media (CSM), a joint venture between China Central Television’s research arm and the London-based Taylor Nelson Sofres (TNS). Both ratings firms have played a crucial role in the development of the television industry in China as the country reintegrates itself into the global capitalist economy. Having said that, the state-associated CSM has dominated the market with a near-monopoly. Indeed, the history of audience measurement has been characterised by monopoly or near-monopoly in most media industry sectors (Buzzard, 2012).

While ratings were once perceived as an imported mechanism to modernise audience measurement in China, they are prone to flaws. Most notably, the politics of representation in the institutionalisation of ratings has attracted critique from scholars (Napoli, 2003; Zhang et al., 2015). These flaws, often resulting from limited sample size and sampling bias, are deeply entwined with the commercial logic of the ratings industry. Typically, as the middle class in urban areas are sought-after consumers for advertisers, they are represented more in the samples while other demographics are overlooked (Braun, 2014). This scenario is exacerbated by CSM’s near-monopoly in the market, which leaves little room for alternatives yet without any third-party audit. While CSM hopes to establish an objective image, its ties with CCTV pose significant challenges in gaining market trust. Despite this, the near-monopoly market structure has maintained its stability.

Further, data manipulation and fraud have given rise to a parasite industry which exploits the vulnerable loopholes in the ratings methodology. While industry self-governance and state regulation are in place, they are largely ineffective. It was not until 2009 that China Radio and Television Association (CRTA) (later renamed as China Alliance of Radio, Film and Television in 2014) established Television Audience Measurement Guidelines. This was followed by the national standard released by Standardization Administration of China, which came into effect in mid-2014. However, weak enforcement means samples are all too often polluted and viewership data directly manipulated (Liu, 2012).

The central role of ratings in programming decisions, as a symptom of broadcasters’ risk aversion, further drives the proliferation of data manipulation and fraud. To counter the highly risky nature of content acquisition, broadcasters tend to utilise valuation adjustment mechanism (VAM, also known as bet-on agreement) in the contractual agreement with content production firms (Ren, 2015). Under such arrangement, broadcasters directly link the transaction price to ratings and reserve the right to receive an agreed amount of rebate if the ratings of the purchased content fall short of the agreed target. The policy of ‘one series, two satellite stations’ (yiju liangxing) enacted in 2015 exacerbated the scenario. With one series only to be allowed to broadcast on two satellite stations, it essentially nullified the long existing practice of cost sharing in content acquisition among multiple broadcasters, therefore raising the stakes for them in a typically uncertain market. Further, the growth of content offerings has led to intensified competition, which has led many production firms to resort to data manipulation. While data manipulation is widely known yet rarely publicly discussed, veteran TV director Guo Jingyu broke the silence in 2018. Guo exposed on Sina Weibo the prevalence of corrupt and shady practices to manipulate ratings in the television industry by referring to his own experience when a TV station demanded that he pay 72 million RMB to boost the ratings as a condition to air his 80-episode drama series Mother’s Life (niang dao) (China Daily, 2018). The exposure stirred a storm in the industry where the long-standing ills become worse as competition for air time intensifies.

The institutionalisation of television ratings in China is thus a process paralleling the commercialisation of the television industry, the dominance of the state-media associated ratings provider, and the burgeoning growth of data manipulation and fraud. Ratings, while initially perceived as a scientific and objective market information regime to modernise the television industry, are fraught with flaws which generates informal data practices. In spite of this, the collusion of multiple stakeholders including content producers, broadcasters, advertisers, ratings firms as well as regulators means ratings maintain the status as an industry currency across the boundary between formal and informal spheres in the near-monopoly market.

Reconfiguring audience measurement in the streaming culture

The proliferation of streaming service providers has introduced significant changes to the television industry in China. They have not only become a key part of distribution infrastructure of television content but also wield significant power in content production. Leading streaming service providers such as iQiyi, Tencent Video and Youku Tudou have not only deployed licensing strategies to acquire local, regional and international titles, but also made aggressive investment in original and commissioned productions, most notably drama series and variety shows (Fung, 2019; Keane, 2015; Lin, 2022; Wang and Lobato, 2019; Zhao and Keane, 2013). These platforms usually derive their revenue from advertising, user subscription as well as on-demand transactions. Audience measurement plays a crucial role in negotiating advertising deals and content acquisition, commissioning and production strategies to attract and retain viewers.

The longstanding near-monopoly market of audience measurement in the legacy television industry has seen streaming platforms complicating the status quo. Indeed, both technological and cultural changes constitute the ‘precipitating jolt’ which leaves space for alternative approaches in institutional changes (Greenwood et al., 2002). Given the imperative to monetise audience attention, streaming service providers have introduced platform metrics and expanded the scope of data capture. From view count to watch time, from likes to skips, from subscriber size to fan engagement, these metrics attempt to capture a spectrum of viewing practices and various imagined audiences. The expanding possibilities to measure audiences is often claimed to boost reliability, the reality is however one of emerging and continuing forms of informality. More broadly, the process of platformisation in China is characterised by various circuits of informality and their complex interactions with the formal media economy, which prompts different regulatory responses (Zhao, 2019).

As streaming service providers navigate the multi-sided market, the meanings and value of metrics are context contingent. Informalities are manifested in inconsistencies, algorithmic opacity, arbitrary changes, and data manipulation. These forms of informality are closely connected to how metrics are defined, used and valued by different streaming platforms, in different contexts, and across time. They reveal how platform evolution in business models, features and interfaces, as well as governance approaches impact on the metric culture.

Frist, the meanings of metrics are varied and inconsistent in different contexts and across different platforms. Consider a widely used metric of content popularity – video views. The metric typically disregards watch time in platforms’ promotional material. In other words, a complete view through to the end of the video can be equated with a brief view followed by skipping the video, with little explanation of how this metric is defined. For reporting purpose in the annual report, the meaning of the metric receives a note. For instance, in iQiyi’s (2020, 2021) annual reports, video views refers to ‘the number of times a video is launched on our platform, regardless of time spent viewing the video’. However, in the context of revenue sharing with content partners, the value of video views in audience measurement becomes contested, and what count as effective video views differ across various platforms. Take online drama series for example. iQyi regards a viewing session over 6 minutes as an effective video view, while Tencent Video draws the line at 5 minutes (Sina, 2021). Likewise, while the metric of total watch time may include time spent watching trailers and extras in the promotional discourse, revenue sharing mechanisms for online drama series excludes the time spent watching trailers and extras. Clearly, inconsistencies arise as platforms use metrics in different contexts. Some serve promotional purposes with a trumpeting tendency while others involve reassessment of the value and meanings of metrics, especially when multiple industry participants negotiate economic interest at stake.

Moreover, the value and meanings of platform-specific metrics change as a result of platform evolution, including adjustment in business strategies, launch of new features or interfaces, or shifts in governance approaches. For instance, when new user growth showed signs of losing momentum, audience measurement started to focus more on maintaining existing users. In this context, total watch time becomes a key indicator of platform stickiness. Subscriber count becomes a key metric when streaming platforms start to expand revenue sources by charging subscription fees, often for users to avoid advertising or unlock premium content. As van Dijck and Poell (2013) point out, new forms of quantification, as facilitated by various features, are intimately bound up with value production of digital platforms. In a similar vein, the value of metrics changes as the mode of value production shifts.

Another illustrative example is the introduction of various versions of algorithms-informed popularity indexes by leading streaming service providers in China. This not only responds to the growing role of social media and fandom culture in the audiencing process, but equally importantly, accentuates the will to solidify platform power. The indexes measure the extent of popularity of series, shows, celebrities or idols, which are closely linked to audience viewing behaviour and social media engagement as captured by data. The rise of these popularity indexes coincides with the streaming service providers’ move to build satellite platforms as social engines in the process of platformisation to capitalise on datafied fan labour (Zhao, 2021). These social engines, most notably iQiyi’s Paopao and Tencent Video’s Doki, are situated at the intersection of the cultural and economic logic of idol fan culture and platform expansion. From check-ins, likes, comments, shares to platform-assigned everyday data tasks, various features cajole fans to draw on affective labour and datafied practices to boost the rankings of their idols or the content featuring their idols. This indicates a turn towards metrics prompting engagement to create the audience commodity. In this process, the integration of datafied fandom practices into audience measurement is closely intertwined with streaming service providers’ expansion into the idol industry in China for capital accumulation. It serves these organisations’ will to enhance platform power in content distribution, production and talent cultivation in the broad context of interindustry dynamics.

The shift towards algorithmic measurement has its own shade of informality. Algorithmic opacity remains a feature of popularity indexes on platforms. This extends the logic of selective transparencies and strategic opacities characteristic of the digital media economy (Lobato and Thomas, 2015; Napoli, 2014). Despite guidelines about how datafied fandom practices can contribute to idol rankings, little information exists about other aspects of how algorithms operate. Selective disclosure of algorithmic measurement mechanisms serves the strategic purpose of translating fan affect into capturable engagement by operationalising the affective meaning and power of data. The metric power serves the platform power in the context of platform evolution and interindustry dynamics (Zhao, 2021). This is underpinned by the metric power of bringing possibilities and producing uncertainties (Beer, 2016) and media circulation power of metrics on screen interfaces (Hesmondhalgh and Lotz, 2020).

It is worth noting that the turn to algorithmic metrics follows closely the move to hide view counts on public interfaces, allegedly to strengthen platforms’ governance of pervasive informal practices – data manipulation and fraud. At the end of 2018, iQiyi announced the decision to remove view counts from its public interface, followed by Youku in 2019 and Tencent Video in 2022. Indeed, a profitable ancillary industry consisting of click farms plays a prominent role in audience measurement. Operating in a grey area, these enterprises typically hire low-paid workers and sometimes draw on automation technologies to generate internet traffic. This is similar to the longstanding problem haunting conventional ratings, although taking a new form. These intermediaries often promote themselves as solution providers who help their client to attract viewers and advertisers. In the platform-mediated attention economy, the imperative to increase visibility creates demand for these informal practices.

While the move to hide view counts is claimed as a measure of platform governance, where to draw the parameters of data manipulation is more complicated than it seems. For example, as fans and fandom practices are reduced to measurable data points, datafied fandom defies the black or white distinction. Enthusiastic fans could put themselves to hard work in generating data, or they could resort to crowdfunding to raise money and delegate the tasks to third-party data brokers. While the platforms stress their determination to crack down on click farms, they paradoxically mobilise datafied fandom with intensifying effort. The latter is evident in how fandom practices are integrated into algorithmic metrics introduced by streaming platforms. This undermines platforms’ self-claimed governance initiatives over informal data practices. Therefore, as streaming service providers mobilise datafied fandom to retain users, bolster content offerings and extend power in the broader screen ecology, the selective focus in platform governance and the associated metric design works to cement their self-serving interest. Underneath the claim for a moral high ground as a pretext for platform governance is what Petre et al. (2019) call platform paternalism that labels a nebulous practice and evokes double standards.

As audience measurement is being reconfigured in the streaming economy, emerging signs of convergence are complicated by contingent and contested meanings and values of metrics. They are characterised by inconsistencies, algorithmic opacity, arbitrary changes and data manipulation. These informalities are closely intertwined with the context-contingent use of metrics, and platform evolution in business models, features and interfaces, and equally importantly, governance approaches. Considering streaming platforms’ role in content distribution, production and increasingly idol cultivation and fandom mobilisation, the strategic design, use and governance of metrics can be seen as an attempt to enhance its market power through audience measurement as infrastructure of audience making. In this sense, their mobilisation of metric power is situated in the process of both platformisation and infrastructuralisation at a time of re-institutionalisation of audience measurement.

Dual approach of algorithmic governance by the state as adaptive cultural governance

State regulation of metric commodities through mandated calibration

The informalities, and more importantly, streaming platforms’ role in reconfiguring audience measurement in internet television have raised regulatory concerns. Both regulatory authorities and state media have censured online viewership measurement for their negative impact on cultural production. Overreliance on click ratings and the prevalence of data fraud is specifically highlighted. In 2016, the State Administration of Press, Publication, Radio, Film and Television (SAPPRFT), the media regulator that preceded the National Radio and Television Administration (NRTA), highlighted the imperative to investigate click ratings of online audiovisual programs and rectify the overreliance on ratings and clicks in evaluating cultural content. The regulator stressed the importance to ‘crack down on content that is cheap (disu), vulgar (yongsu) and tasteless (meisu)’, and to ‘ensure the development of culture in the politically correct direction’ (People.cn, 2016).

Mounting criticism from the state media against data inflation and fraud followed in the ensuing years. In April 2017, CCTV-6 broadcast a special edition on the inflation of viewership data on streaming platforms. People’s Daily (2018) also drew attention to fake clicks on streaming platforms and called for more stringent regulation over audience measurement in internet television. In 2019, CCTV News revealed the ‘alarming’ data fraud behind the rise of highly-paid, internet traffic-dependent celebrities (liuliang mingxing), driving up the cost in screen content production.

The reprimands over data inflation and fraud extends to fan culture, with a moralistic framing of datafied fandom and pay-to-vote practices in particular. Media commentaries highlighted ‘blind’, ‘irrational’ and ‘pathological’ fandom practices (e.g. Xinhua, 2020, 2021). The rhetoric indicates a grave concern about the capacity to formulate close-knit communities with a shared purpose and organise networked actions as shown in datafied fandom practices. Indeed, existing evidence shows the potential of online affective communities in mobilising civic engagement and political participation (Wu, 2017; Zhang and Mao, 2013). It therefore becomes crucial that these networked communities and their practices are aligned with the interest and values of the party-state. This also explains the connection between the moralistic framing of datafied fandom and the censure against digital platforms for misleading youth.

The extensive censure against audience measurement on streaming platforms are closely connected to the regulatory will to rein in digital platforms and the entertainment industry. In July 2020, the Cyberspace Administration of China (CAC, 2020) in its ‘Clear and Bright’ Campaign started to highlight the focus to crack down on ‘chaotic fandom’ among minors. Targeted focus of the campaign included rectifying datafied fandom practices such as voting for idols to boost their positions in various rankings, inflating click counts and manipulating comments. The crackdown stepped up when the Publicity Department of the Communist Party of China Central Committee (hereafter the Publicity Department) stressed in September 2021 the imperative for relevant regulatory authorities to make coordinated efforts in regulating the cultural and entertainment sector, highlighting specific problems including data fraud and chaotic fandom and stressing the platform responsibility in content governance and algorithmic optimisation (People.cn, 2021).

In response to the Publicity Department, the National Radio and Television Administration (NRTA) launched a dedicated campaign targeting internet variety shows. The media regulator called for digital platforms and production companies to curtail idol cultivation shows, overhaul the voting mechanisms, and rectify the tendency of data worshipping (NRTA, 2021a). Within a month, the NRTA (2021b) issued the Notice on Further Regulation of Arts and Cultural Programs as well as Related Personnel, prohibiting all broadcasters and streaming platforms from distributing idol cultivation shows. It reiterated the stance against ‘using online traffic data as the only evaluation criteria’ and called for ‘professional and authoritative arts and cultural critique’.

Specific regulatory measures targeting algorithmic audience measurement on digital platforms followed. As a part of ‘Clear and Bright’ Campaign in 2021, the CAC (2021) ordered platforms to stop publishing celebrity-ranking lists and called for an ‘optimisation of ranking lists of cultural products’. The latter involves reducing the weight of fan check-ins, likes and comments, prohibiting platforms from encouraging fans to vote for their idols and allowing fans to boost engagement metrics via payment, and increasing the weight of value orientation and professional evaluation. In line with the Publicity Department, the internet watchdog stressed the imperative and urgency to ‘safeguard political and ideological security online’ (CAC, 2021).

Clearly, the state’s move to overhaul metric commodities in the internet television market becomes a crucial part of cultural governance, closely tied to the political and ideological purposes. The focus on data manipulation and fraud, with a noticeable extension to a moral framing of datafied fandom, highlights the role of algorithmic metrics in reshaping audience measurement. The regulation over algorithmic metric commodities by way of mandated calibration drives the re-institutionalisation of audience measurement.

In this process, the regulatory mandate to reconfigure algorithms introduce changes to two dimensions of institutionality of algorithms. As Napoli (2014) points out, institutionality of algorithms can be understood from two perspectives – the operation of algorithms as code, and the development of algorithms as a process subject to institutional forces. The above analysis shows how informalities arising from the operation of algorithms as code in the market trigger regulatory intervention, which in turn impact on how algorithms are designed and deployed.

Ironically, from a historical perspective, informalities in the television ratings market have been a longstanding and widely recognised problem yet little substantial regulation has been put in place. The near-monopoly of the state-associated ratings provider renders the problem less of a concern for the state. The institutional inertia is however disrupted in the streaming era, triggering intensified regulatory responses. As cultural governance is enacted through market information regimes, algorithms become a regulatory object and instrument at once.

State-initiated measurement regime as market alternative and policy instrument

The regulation of the emerging metric commodities cannot be fully understood without discussing the introduction of the state-initiated algorithmic audience measurement regime. In fact, NRTA (2021b) has called for intensifying efforts to promote and use the official measurement regime – China’s Audio-Visual Big Data (CVB). The enamourment with big data, a problematic term in itself and in its connotation of ‘bigger is better’ (boyd and Crawford, 2012), is evident. In the context where ‘authority is increasingly expressed algorithmically’ (Pasquale, 2015: 8), not only the state regulation of algorithms on the market but also the development of an official system complicates the ways in which authority is expressed, enabled and negotiated.

Launched in December 2019, CVB is a system for audiovisual program evaluation that has been under development since 2016. The deployment of CVB is in response to the party-state’s call to establish a new, authoritative audience measurement system, as well as addressing stakeholders’ need in a rapidly developing industry. The purpose, according to NRTA (2020), is to combat ratings manipulation, to guide the healthy development of the industry, and to support wide dissemination of quality programs. By the end of 2021, it covered over 300 million cable, IPTV and OTT households across 36 administrative areas across different levels from municipalities, provincial capitals to prefecture-level cities, as well as counties in China (Jiang, 2021).

Although grounded in the recognition of industry flaws, the attempt to legitimate the officially sanctioned measurement system is in itself problematic. The promotional discourse accentuates several characteristics including a vast coverage of data sources, a diversified range of metrics to capture various viewing behaviour, and a ‘completely automatic’ process free from manipulation (NRTA, 2020). However, the purported advantages fall short upon a closer examination.

Firstly, the emphasis on data volume obfuscates the politics of representation in relation to how audiences access television. Despite the self-claimed vast coverage, proportional alignment to nationwide data in terms of different ways to access television remains a challenge. CVB at initial launch focused on cable and IPTV users, with OTT included at a later stage (Zu, 2019). So far, OTT users remain a relatively small part of the coverage in CVB while far exceeding cable and IPTV users on a national scale. This distortion however is hidden in the self-claimed vast coverage rhetoric.

Moreover, the emphasis on metric diversity to capture various viewing behaviour eludes the key question of how the metrics are used to inform evaluation. What becomes evident is the role of audience measurement in ideological governance, a point reiterated by the regulator. NRTA pointed to the role of CVB in ‘strengthening the capacity to guide public opinion’ and ‘promoting select programs reflecting “the new era”’ (NRTA, 2021c). The latter specifically serves the project of developing theme programs (zhuti jiemu) in the new era, a term specifically connected to Xi Jinping Thought on Socialism with Chinese Characteristics since the 18th CPC National Congress in 2012. According to the media regulator, the move to publish the data of good-performing theme programs helps promoting these programs and mainstream values more broadly to the audience. Such reference to – and mobilisation of – the logic of visibility begets visibility shows the state’s attempt to draw on algorithmic governance to serve the purpose of cultural and ideological governance. The ongoing re-institutionalisation of audience measurement in the evolving screen ecology is thus underlined by the imbrication of the market logic and the political logic.

Finally, the attempt to trumpet computational neutrality of the automatic process conceals the underlying complex dynamics. Equating automation with neutrality provincialises the measurement process as it neglects multiple interlocking stages that inform the computing process and impact the result. The rhetoric turns automation into a synecdoche for the entire process. This works to establish an imagined binary between computers and humans, yet the reality is more complicated. In fact, as humans and nonhumans entangle within an assemblage, they translate one another and formulate technological-social hybrids (Couldry, 2008; Latour, 1994). Bracketing off mediation from both humans and machines is a thinly veiled attempt to shore up computational sublime.

The introduction of an official audience measurement mechanism constitutes the state’s response as a market participant and regulator at once. It serves as both a market alternative and a policy instrument. As the streaming culture renders the legacy audience measurement mechanism inadequate, the space opens up for alternative approaches. For the state, therefore, it becomes a clear imperative to capture the emerging market to maintain its dominance. The attempt to capture the market doubles up as an adaptive approach to cultural governance. By reinserting itself into the audience measurement market in the streaming era, the state reinforces its political logic via metric power. In doing so, the legitimacy building rhetoric points to the tech industry’s flaws, and paradoxically, evokes a sense of technological sublime around big data and automation permeating in the industry. This however obfuscates the politics of representation and various forms of informality that continue to trouble audience measurement in the streaming ecology.

Conclusion

As streaming culture reshapes the television industry, audience measurement becomes a crucial locus of media power. This article has examined the changing approaches to audience measurement as both digital platforms and the state grapple the complexities brought by the streaming culture in China. The analysis shows some signs of institutional convergence in the industry, in terms of championing the ‘big data’ approach, incorporating social media metrics and moving towards algorithmic metrics. However, much like the institutionalisation of the ratings culture is fraught with methodological flaws and questionable data practices, informality continues to exist as the new market information regime is in formation. As illustrated, despite the often self-claimed break from the stubborn ill of the ratings service, use of new metrics manifests inconsistencies, algorithmic opacity, arbitrary changes, and data manipulation. In addition to context-contingent and cross-platform fragmentation, various moments such as business model adjustments, new feature or interface launches and shifts in platform governance approaches may introduce tweaks to the design, use and equally importantly, governance of metrics. This is closely connected to the politics of the platform market, and often in service of streaming service providers.

As streaming platforms reconfigure audience measurement, increasingly through algorithms, they command more power to shape cultural production. This, together with the insufficiency of platform self-regulation, has triggered regulatory concerns which are underpinned by the attempt to invoke moral and pragmatic legitimacy. Streaming platforms are highlighted as vectors of informal practices including data inflation and fraud, undermining public trust and industry development. Datafied fandom practices, increasingly nurtured and mediated by platforms in the streaming culture, are brought into the regulatory remit with a moralistic framing despite the often vague and penetrable formal-informal divides. The regulatory will over the audience measurement market is closely connected to the state’s resolve to rein in digital platforms and the entertainment industry more broadly.

At a time when the new measurement regime is in the making, the state becomes a prominent actor in driving a new wave of institutionalisation of audience measurement in the streaming era. The state responds in two ways, by regulating metric commodities on platforms and by launching an official measurement system. The former places the emerging measurement regime within institutional parameters; the latter operates both as a market alternative and a policy instrument, showing the state’s attempt to stay relevant in the new market as its near-monopoly status in the legacy industry is unsettled. Algorithms become regulatory objects and instruments at once. The regulatory response doubles up as a market response, revealing the state’s intention to reclaim power as both market regulator and participant.

The dual approach – governance of and with algorithms – constitute pathways to adaptive cultural governance. Given algorithms are inescapably value-laden, outlining institutional boundaries for algorithmic design and mobilising the demonstrative effect of the official system enable the state to shape the re-institutionalisation process. The attempt to embed authority in algorithms is however compromised to some extent. The purported rigor and objectivity based on data size, metric diversity and automation of the state-initiated measurement system is illusory. Under the computational sublime are the hidden politics of inclusion at both data and metric levels as well as in the algorithmic workings, which defy the synecdoche of automation. Despite the tenuous claims around rigor and objectivity, the state’s responses demonstrate its attempt to drive re-institutionalisation of audience measurement in the streaming era to stay relevant in the market and hone its approach to cultural governance.

As the new wave of institutionalisation of audience measurement unfolds in the context of streaming culture, the trajectory in China so far indicates how platforms have challenged legacy institutions, and how the disruptive effect can be limited as the state mobilises metric power and algorithmic governance to reassert its authority in and over the market. The expanding boundaries of television and fragmentation in access and consumption parallels the ‘mass’ audience construction, and the politics of metrics underpin the negotiation of cultural power. While the platformisation of the television industry in China shares some similarity with the unfolding process elsewhere in the world, both regulatory approaches and industrial practices have distinctive aspects that warrants further research. As the re-institutionalisation process continue to evolve, future research should investigate further how various versions of the new market information regime introduced by platforms and the state are understood, used and negotiated by stakeholders, and what consequences they bring to cultural production as the state and corporate platforms become increasingly entangled. Moreover, in the broader context of Chinese platforms’ expansion in overseas markets and the global competition for digital sovereignty (Keane et al., 2020; Kokas, 2022), how the state-platform dynamics play out in geopolitical complexities warrants further research.

Footnotes

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Elaine Jing Zhao

References

Anand

Peterson

(2000) When market information constitutes fields: Sensemaking of markets in the commercial music industry. Organization Science 11(3): 270–284.

Ang

(1991) Desperately Seeking the Audience. London: Routledge.

Beer

(2016) Metric Power. Basingstoke, UK: Palgrave Macmillan.

Bai

(2005) Media commercialization, entertainment and the party-state: The political economy of contemporary Chinese television entertainment culture. Global Media Journal 4(6): 1–54.

Bolin

(2011) Value and the Media. Cultural Production and Consumption in Digital Markets. Farnham: Ashgate.

Bolin

Velkova

(2020) Audience-metric continuity? Approaching the meaning of measurement in the digital everyday. Media, Culture and Society 42(7–8): 1193–1209.

boyd

Crawford

(2012) Critical questions for big data. Information, Communication & Society 15(5): 662–679.

Braun

(2014) Transparent intermediaries: Building the infrastructures of connected viewing. In: Holt

Sanson

(eds) Connected Viewing: Selling, Streaming and Sharing Media in the Digital Era. New York and London: Routledge, pp.124–143.

Buzzard

(2012) Tracking the Audience: The Ratings Industry from Analog to Digital. New York, NY: Routledge.

10.

China Daily (2018) Doctoring of audience ratings has to stop. Available at: https://www.chinadaily.com.cn/a/201809/19/WS5ba188fea31033b4f4656d08.html (accessed 18 March 2022).

11.

Couldry

(2008) Actor network theory and media: Do they connect and on what terms? In: Hepp

Krotz

Moores

Winter

(eds) Connectivity, Networks and Flow: Conceptualizing Contemporary Communications. Cresskill, NJ: Hampton Press, pp.93–110.

12.

Cunningham

Silver

(2013) Screen Distribution and the New King Kongs of the Online World. Basingstoke: Palgrave Macmillan.

13.

Cyberspace Administration of China (2020) Guanyu kaizhan 2020 qinglang weichengnianren shuqi wangluo huanjing zhuanxiang zhengzhi de tongzhi [The notice on special rectification campaign for a clear and bright internet for Juveniles in summer 2020]. Available at: http://www.cac.gov.cn/2020-07/13/c_1596175859026231.htm (accessed 18 March 2022).

14.

Cyberspace Administration of China (2021) Guanyu jinyibu jiaqiang fanquan luanxiang zhili de tongzhi [The notice on further strengthen the governance of chaos in fan circles]. Available at: http://www.cac.gov.cn/2021-08/26/c_1631563902354584.htm (accessed 18 March 2022).

15.

Ettema

Whitney

(1994) Audiencemaking: How the Media Create the Audience. Thousand Oaks, CA: Sage.

16.

Evens

Donders

(2018) Platform Power and Policy in Transforming Television Markets. Basingstoke: Palgrave Macmillan.

17.

Fisher

Mehozay

(2019) How algorithms see their audience: Media epistemes and the changing conception of the individual. Media, Culture & Society 41(8): 1176–1191.

18.

Flew

(2021) Regulating Platforms. Cambridge: Polity.

19.

Fung

(2019) Fandomization of online video or television in China. Media, Culture and Society 41(7): 995–1010.

20.

Greenwood

Suddaby

Hinings

(2002) Theorizing change: The role of professional associations in the transformation of institutionalized fields. Academy of Management Journal 45(1): 58–80.

21.

Hesmondhalgh

Lotz

(2020) Video screen interfaces as new sites of media circulation power. International Journal of Communication 14: 386–409.

22.

Holt

Sanson

(eds) (2013) Connected Viewing: Selling, Streaming, and Sharing Media in the Digital Age. London: Routledge.

23.

iQiyi (2020) Form 20-F. Available at: https://ir.iqiyi.com/static-files/82e37ef4-f1e5-471d-90cb-1107d2a33aba (accessed 6 March 2022).

24.

iQiyi (2021) Form 20-F. Available at: https://ir.iqiyi.com/static-files/b354fb69-a41d-460a-89fd-8112f5300012 (accessed 6 March 2022).

25.

Jenner

(2018) Netflix and the Re-invention of Television. London: Palgrave Macmillan.

26.

Jiang

(2021) Yuanqi, yunxing, fazhan: zhongguo shiting dashuju fuwu xuanchuan guanli, zhicheng hangye fazhan [Origin, operation and development: CVB serves propaganda management and supports industry development]. Guangdian Shiping. Available at: https://www.163.com/dy/article/GPJOJB0A05119UAT.html (accessed 10 April 2022).

27.

Johnson

(2019) Online TV. Abingdon and New York: Routledge.

28.

Keane

(2015) The Chinese Television Industry. London: BFI Palgrave.

29.

Keane

Zhao

, et al. (2020) China’s Digital Presence in the Asia-Pacific: Culture, Technology and Platforms. London: Anthem.

30.

Keane

Zhao

(2014) The reform of the cultural system: Culture, creativity and innovation in China. In Lee

Lim

(eds) Cultural Policies in East Asia: Dynamics between the State, Arts and Creative Industries. New York, NY: Palgrave Macmillan, pp.155–173.

31.

Kokas

(2022) Trafficking Data: How China is Winning the Battle for Digital Sovereignty. Oxford: Oxford University Press.

32.

Kosterich

(2016) Reconfiguring the “hits”: The new portrait of television program success in an era of big data. International Journal on Media Management 18(1): 43–58.

33.

Kosterich

Napoli

(2016) Reconfiguring the audience commodity: The institutionalization of social TV analytics as market information regime. Television & New Media 17(3): 254–271.

34.

Latour

(1994) On technical mediation—Philosophy, sociology, genealogy. Common Knowledge 3(2): 29–64.

35.

Lin

(2022) Convergent Chinese Television Industries: An Ethnography of Chinese Production Cultures. Cham, Switzerland: Palgrave Macmillan.

36.

Liu

(2012) Zaitan shoushilv zaojia: yuanqi, jishu yu jianguan [On ratings manipulation: Causes, technologies and regulation]. Xiandai Chuanbo 10: 1–6.

37.

Lobato

(2019) Netflix Nations: The Geography of Digital Distribution. New York, NY: New York University Press.

38.

Lobato

Thomas

(2015) The Informal Media Economy. Cambridge and Malden: Polity.

39.

Lotz

(2017) Portals: A Treatise on Internet-Distributed Television. Ann Arbor, MI: Maize Books.

40.

Lotz

(2022) Netflix and Streaming Video: The Business of Subscriber-Funded Video on Demand. London: Polity.

41.

Napoli

(2003) Audience Economics: Media Institutions and the Audience Marketplace. New York, NY: Columbia University Press.

42.

Napoli

(2010) Audience Evolution: New Technologies and the Transformation of Media Audiences. New York, NY: Columbia University Press.

43.

Napoli

(2014) Automated media: An institutional perspective on algorithmic media production and consumption. Communication Theory 24(3): 340–360.

44.

National Radio and Television Administration (2020) Zhongguo shiting dashuju yinling shoushi shuju touming shidai de daolai [China’s Audio-Visual Big Data heralds an age of transparency]. Available at: http://www.nrta.gov.cn/art/2020/1/22/art_114_49594.html (accessed 18 March 2022).

45.

National Radio and Television Administration (2021a) Jinyibu jiaqiang wangluo zongyijiemu guanli de tongzhi [Notification on further regulation of internet variety shows]. Available at: https://news.sina.com.cn/c/2021-08-03/doc-ikqciyzk9339872.shtml (accessed 16 March 2022).

46.

National Radio and Television Administration (2021b) Guojia guangbo dianshi zongju bangongting guanyu jinyibu jiaqiang wenyi jiemu jiqi renyuan guanli de tongzhi [The notice on further regulation of arts and cultural programs as well as related personnel]. Available at: http://www.nrta.gov.cn/art/2021/9/2/art_113_57756.html (accessed 18 March 2022).

47.

National Radio and Television Administration (2021c) 2020 niandu zhongguo guangbo dianshi hangye shida keji guanjianci pingxuan jieguo gongbu [Announcing ten tech keywords of China’s radio and television industry in 2020]. Available at: http://www.nrta.gov.cn/art/2021/4/2/art_114_55663.html (accessed 10 April 2022).

48.

Pasquale

(2015) The Black Box Society: The Secret Algorithms That Control Money and Information. Cambridge, MA: Harvard University Press.

49.

People.cn (2016) Zhonggong guojia xinwen chuban guangdian zongju guanyu xunshi zhenggai qingkuang de tongbao [The notification of the party leadership group of NRTA on the progress of inspection and rectification]. Available at: http://fanfu.people.com.cn/n1/2016/0829/c64371-28672869.html (accessed 15 March 2022).

50.

People.cn (2021) Zhongyang xuanchuanbu yinfa tongzhi, bushu wenyu lingyu zonghe zhili gongzuo [The notice of the publicity Department of the Communist Party of China Central Committee on integrated governance of cultural and entertainment sector]. Available at: http://politics.people.com.cn/n1/2021/0902/c1001-32215990.html (accessed 16 March 2022).

51.

People’s Daily (2018) wangboju dianjiliang zaojia dang jingti [Beware fake click ratings of internet dramas]. Available at: http://it.people.com.cn/n1/2018/0116/c1009-29766683.html (accessed 15 March 2022).

52.

Petre

Duffy

Hund

(2019) “Gaming the system”: Platform paternalism and the politics of algorithmic visibility. Social Media + Society 5(4): 1–12.

53.

Ren

(2015) Dianshi zhibo “shoushi duidu” de zaishenshi [Re-examining valuation adjustment mechanism in television production and distribution]. Nanfang Dianshi Xuekan 5: 71–73.

54.

Sina (2021) You Ai Teng: San pingtai fenzhang youbie yu jufang zhuiqiu gongying [Youku Tudou, iQiyi and Tencent Video diverge in revenue sharing approaches in pursuit of win-win scenarios with content producers]. Beijing News, 4 March. Available at: https://finance.sina.com.cn/tech/2021-03-04/doc-ikftssaq0557339.shtml (accessed 22 March 2022).

55.

van Dijck

(2014) Datafication, dataism and dataveillance: Big data between scientific paradigm and ideology. Surveillance & Society 12(2): 197–208.

56.

van Dijck

Poell

(2013) Understanding social media logic. Media and Communication 1(1): 2–14.

57.

Wang

Lobato

(2019) Chinese video streaming services in the context of global platform studies, Chinese Journal of Communication 12(3): 356–371.

58.

(2017) Entertainment and Politics in Contemporary China. Switzerland: Palgrave Macmillan.

59.

Xinhua (2020) Mitu “fanquan” ruhe zhifan [How can the lost “fan circles” return to the right track]. Available at: http://www.xinhuanet.com/politics/2020-09/03/c_1126445828.htm (accessed 16 March 2022).

60.

Xinhua (2021) Ba niunai dao gouli? Bie ba qingnianren daigouli! [Dumping milk down the drain? Do not mislead the youth!]. Available at: http://www.xinhuanet.com/politics/2021-05/04/c_1127408247.htm (accessed 16 March 2022).

61.

Zhang

(2009) Chinese television audience research. In: Zhu

Berry

(eds) TV China: A Reader on New Media. Bloomington and Indianapolis: Indiana University Press, pp. 168–180.

62.

Zhang

Mao

(2013) Fan activism sustained and challenged: Participatory culture in Chinese online translation communities. Chinese Journal of Communication 6(1): 45–61.

63.

Zhang

Zhao

(2015) The people’s choice? Class and representative politics behind ratings. Open Times 3: 158–175.

64.

Zhao

(2019) Digital China’s Informal Circuits: Platforms, Labour and Governance. London: Routledge.

65.

Zhao

(2021) Reconfiguring audience measurement in platform ecologies of video streaming: iQiyi’s pivot toward data-driven fandom and algorithmic metrics. International Journal of Communication 15: 3671–3691.

66.

Zhao

Keane

(2013) Between formal and informal: The shakeout in China’s online video industry. Media, Culture and Society 35(6): 724–741.

67.

Zhao

(1998) Media, Market, and Democracy in China: Between the Party Line and the Bottom Line. Urbana and Chicago, IL: University of Illinois Press.

68.

Zhao

(2008) Communication in China: Political Economy, Power and Conflict. Lanham, MD: Rowman and Littlefield.

69.

Zhu

(2008) Television in Post-Reform China: Serial Dramas, Confucian Leadership and the Global Television Market. Abingdon: Routledge.

70.

(2019) Zongju dashuju xitong shangxian: guanxuan de shoushilv xin zai na’er [The launch of official big data system: What’s new with the new measurement system?]. Beijing Qingnian Bao [Beijing Youth Daily]. Available at: http://media.people.com.cn/n1/2019/1220/c40606-31514595.html (accessed 5 April 2021).