When availability is not enough: Content prominence and diversity on a public service VoD platform

Technical affordances of VoD, prominence, and media diversity

The concept of affordances, introduced by Gibson (1979), describes how environments enable or constrain particular actions. Building on this perspective, Nagy and Neff (2015) have argued that affordances should also be understood as sociotechnical constructs that emerge through the interaction between technological properties, human agency, and social context. In this sense, VoD services are not neutral distribution channels but sociotechnical environments that offer specific possibilities for action; their use depends on how these possibilities are perceived, interpreted, and enacted by users within specific social and cultural contexts.

For instance, VoD viewing offers a personalised media experience, allowing users to control what and when they watch, thereby emphasising user freedom and active audiences. The auto-play feature, in addition to making whole seasons available at once, not only facilitates immediate and prolonged viewing but also shapes viewers’ habits and perceptions of series consumption (Markham et al., 2019). The immersive and escapist experience of frictionless viewing has a strong positive impact on user preferences for VoD (Evens et al., 2023). However, service architecture enables certain types of content visibility and actions while constraining others. For instance, VoD platforms promote content on prime shelves and employ interface designs that steer users towards perceiving the platform as a continuous, channel-like flow (Matthew, 2020). Additionally, VoD content presentation is predominantly structured by genres and periodic or functional categories. Ultimately, the technical affordances of VoD translate into systems of visibility that determine which titles or genres gain prominence, also shaping how audiences understand and perceive VoD services.

Content prominence refers to the visibility of content within the interface and functions as an operational tool shaping how easily audiences can encounter specific titles, notably through placements in key areas such as hero, top, or featured banners. The most visible area has been described as “prime space” (Bideau, 2020), analogous to prime time in linear broadcasting. Yet, prominence is a multidimensional concept that extends beyond placement, also encompassing the frequency and duration of a title’s appearance in highlighted areas (Bruun & Lassen, 2023; Kelly & Sørensen, 2021; Sørensen, 2025) and, in personalised interfaces, its presence in algorithmic recommendations and search results (García Leiva & Mazzoli, 2023).

The latter dimensions somewhat overlap with the concept of discoverability. However, while discoverability concerns the ease with which users can find content, prominence captures the degree of priority assigned to specific content by the platform itself (Hesmondhalgh & Lotz, 2020; Iordache et al., 2023; Lobato et al., 2024). Therefore, content prominence embodies a form of interface-based circulation power that refers to the ability to steer attention and shape cultural visibility. What is promoted within the interface often mirrors the strategic or commercial goals driving the organisation (Hesmondhalgh & Lotz, 2020). In the case of PSM, such prominence may also reflect the realisation of public service objectives.

Prominence is thus a strategic curatorial act, shaped by both human and algorithmic decision-making. It has measurable implications: Content that is more prominently positioned is more likely to be discovered, and ultimately more likely to be viewed (Hesmondhalgh & Lotz, 2020), creating a feedback loop where content’s popularity reinforces its visibility and viewership (Webster, 2014).

Media diversity has long been recognised as a key democratic principle ensuring that citizens have access to a broad range of sources and viewpoints to make balanced and well-considered decisions (Hartley et al., 2020; Helberger et al., 2016). For PSM, diversity is therefore not only an outcome of market plurality, but a core institutional responsibility tied to citizenship, inclusion, and cultural representation. However, the meaning of diversity has evolved alongside technological and institutional change. Early approaches treated diversity as a supply-side property, that is, the range of content types and sources made available (McQuail, 1992). Napoli (1999, 2011) challenged this perspective by arguing that true diversity should be assessed in terms of exposure, conceptualising it as the extent to which audiences consume a diverse array of content. A similar distinction has also been made, albeit using different terminology, between supplied and consumed diversity (Moreau & Peltier, 2004). This distinction is important because, despite the expansion of choice brought about by the digital era, audience attention often concentrates on familiar or preferred content (Helberger, 2012; Hindman, 2007; Napoli, 2011), producing a gap between supply-side and demand-side diversity. More recently, the emergence of interface-driven curated media environments has further refined the audience perspective. In digital platforms, the content to which users are exposed is no longer equivalent to the content they ultimately choose to consume. For example, in the context of VoD services, where content prominence plays a crucial role in determining discoverability, it becomes necessary to distinguish between the content to which audiences are exposed through the VoD interface and the content they actually watch – that is, consumed diversity. Recent approaches therefore call for distinguishing between exposure diversity, referring to the diversity of content presented on the interface, and consumption diversity, referring to the diversity of content with which users actually engage (Loecherbach et al., 2020; Mattis et al., 2024). Likewise, research on PSM VoD services has conceptualised exposure as the presentation of content within VoD interfaces, emphasising that the diversity of exposed content constitutes an important object of analysis in its own right (Bengesser & Sørensen, 2024; Sørensen, 2025). Bengesser and colleagues (2024) have further argued that publishing strategies mediate the entire circulation chain from catalogue presence to interface visibility and ultimately to consumption.

Building on this interface-oriented understanding of diversity, the article distinguishes three interlinked stages of diversity: available diversity, referring to the range of content available in the catalogue; exposure diversity, referring to the subset made visible through the interface (i.e., content prominence); and consumed diversity, referring to the content audiences watch.

While demand-side diversity is often conceptualised and operationalised at the level of individual users and their media repertoires (e.g., Helberger et al., 2016; Moe et al., 2021), in this study I examine consumed diversity at the aggregate audience level, corresponding to what Napoli (1999) termed horizontal diversity. I examine how editorial curation and audience choice together redistribute audience attention across content characteristics. This provides a perspective for assessing the public value of curated PSM VoD services. Rather than asking how diverse the content consumed by individual users is, I examine, at the aggregate audience level, whether the diversity available in the catalogue is preserved, amplified, or reduced as content moves through interface prominence to audience consumption.

Further, Stirling (2007) distinguished diversity across the aspects of variety (number of options), balance (distribution among them), and disparity (degree of difference). The assessment of diversity fostered by PSM should extend beyond variety, encompassing motivated balance between content types and sources (Ibrus et al., 2023). Therefore, in this article, the focus is placed on assessing variety and balance in order to evaluate PSM as both an enabler and a curator of diversity. In this context, content prominence should be recognised as a pivotal tool enabling the realisation of these strategic objectives.

By combining insights on affordances, prominence, and diversity, this framework introduces the concept of diversity transformation – the process through which content diversity is reshaped across the stages of availability, exposure, and consumption. At each stage, content is subject to a different selection mechanism: It is first selected for the catalogue, then editorial curation determines which titles receive prominence on the interface, while audience choices determine which titles are ultimately consumed. Within this framework, prominence is recognised as both a technical affordance and a strategic editorial intervention, positioning exposure diversity as the outcome of a complex interaction between catalogue composition, platform logics, interface design, curation, and user behaviour. The approach is particularly relevant for evaluating whether PSM VoD services fulfil their public mandate, as it enables operationalisation through empirical indicators such as diversity measures, prominence scores, and viewership metrics. It also makes visible how public service intentions are translated through platform-based visibility structures.

However, it should be noted that this approach risks overemphasising the role of the VoD platform interface and its affordances in shaping viewing behaviour. Recent research has highlighted that users are not simply passive recipients of prominence or platform design and may, in fact, be blind to or indifferent to the affordances presented. Instead, their consumption patterns are frequently shaped by “default” ways (Johnson et al., 2023) or ritual step-by-step processes (Iordache et al., 2023), which points back to Nagy and Neff (2015), who argued that the use of affordances depends on users’ interpretations, intentions, and the contexts in which interaction takes place. The impact of external influences also needs to be highlighted. Lüders and Sundet (2021) pointed out the role of peer recommendations. Similarly, Iordache and colleagues (2023) recognised external “nudges”, referring to off-platform promotions and reviews, and Johnson and colleagues (2023) emphasised the role of paratexts and word of mouth.

Transformation of PSM

The ongoing transformation of European PSM, driven by platformisation and the rise of on-demand media culture, entails a gradual shift from broadcast scheduling to digital publishing. Publishing practices have become the key instruments through which PSM realise their public objectives (Bruun & Lassen, 2023; Kelly & Sørensen, 2021), and VoD services are becoming the strategic entry points to PSM content (Bruun, 2023; Iordache & Raats, 2023). However, across comparative research, PSM strategies have been found to move between adaptation and innovation. D’Arma and colleagues (2021) and Donders (2019) have shown that although PSM organisations have rhetorically embraced platformisation, most continue to repurpose linear content for VoD, maintaining organisational silos and linear key performance indicators. Their online expansion remains largely reactive, driven by legitimacy concerns and constrained budgets rather than by coherent digital-first strategies. Further, building on Napoli (1998), D’Arma and colleagues (2021) described two responses to competitive threats: differentiation, emphasising domestic and culturally specific programming, and mimicry, through the adoption of commercial VoD features and practices. These strategies may coexist to meet audience expectations shaped by global streamers while safeguarding cultural distinctiveness and editorial accountability.

The transformation has also redefined professional roles and workflows. Bruun (2023) and Bruun and Lassen (2023) have shown how scheduling work has evolved into new forms of digital publishing, merging linear and on-demand operations through trans-programming models and interfaces acting as new “prime spaces” where editorial priorities are negotiated daily. Ultimately, PSM VoD publishing strategies illustrate a balancing act between strategic objectives, editorial autonomy, and audience expectations – all in an effort to sustain relevance, diversity, and universality.

Case of ERR

ERR is in transition towards becoming a digitally advanced broadcaster; however, it has not reached the stage of VoD-centric, online-only publishing, and data-integrated editorial practices observed in Nordic or Western European PSM organisations (Bruun, 2023; D’Arma et al., 2021; Iordache & Raats, 2023). ERR is witnessing shifts in audience preferences: Linear television remains strong but is declining, while the viewership of VoD is increasing. ERR’s VoD service Jupiter, launched in 2020, includes content from all ERR television channels and a limited selection of online-only content. The shift from catch-up service to a more advanced digital presence came relatively late. Despite this, Jupiter has become the most-used domestic VoD service in Estonia. Its popularity rivals Netflix, the sole global service with a considerable market share (Kantar Emor, 2025).

The design of Jupiter is relatively simple, beginning with a hero banner, followed by thematic rows that present titles using vertical posters (see Figure 1). The organising principle of content is thematic and genre-based, rather than channel-based. In addition to the homepage, content is displayed on dedicated presentation pages that group together series, television shows, and films. The structure of the thematic rows is fairly stable. Their number, focus, and order remained largely consistent during the studied period.

The content is presented in many different positions (across 76 thematic rows, comprising 2,800 positions in total), which suggests that a large and diverse range of content is being shown. Jupiter’s personalisation is limited to functional features such as Continue Watching and Watchlist, with no personalised recommendations or layouts. As only around 5 per cent of users engage with these features, their effect on the prominence indicator is marginal. All in all, this configuration enables analysis of how content prominence structures exposure diversity and how it relates to viewership and consumed diversity.

Figure 1

Diversity transformation across metadata categories

The following analyses examine what this article conceptualises as diversity transformation: the process through which the diversity changes across the stages of availability, exposure, and consumption. By comparing these three stages across metadata categories, the analysis identifies in which categories, in which direction, and to what extent these changes occur.

First, the distributions of content types across different diversity stages are analysed. As shown in Figure 3, the catalogue contains a broad range of content types, with a particular emphasis on culture and arts, and factual (news, life, infotainment, informative and documentary) programming. However, a clear shift in priorities becomes visible in the prominence structure, that is, in the editorial decisions. For example, while series account for only 7 per cent of titles in the catalogue, they make up 21 per cent of the total prominence. At the same time, the visibility of culture and arts content (9% of the total prominence) is smaller compared to its share in the catalogue (16%). Viewers’ preferences, however, leave little doubt: Scripted content accounts for more than half of all viewership, divided evenly between series (27%) and feature films (27%). Compared to their share in the catalogue, infotainment and news also attract a notably higher share of viewership.

Figure 3

Figure 4 demonstrates that in the catalogue, 65 per cent of titles have a domestic origin, whereas for prominence the figure decreases to 55 per cent and comprises less than half (46%) of viewership.

Similar to content types, it should be noted that the shift away from domestic content towards other origins begins at the stage of prominence and continues in the same direction with viewership figures. However, it is worth noting that in terms of origin, audience preferences are ultimately more diverse than the catalogue itself. This is due to the fact that content types and origins are interwoven, as scripted content is predominantly acquired from abroad, while factual and culture and arts content tends to be locally produced.

Figure 4

Figure 5 shows a strong concentration toward content produced in recent years across all diversity stages. However, the share of titles from the past three years varies significantly. Such content makes up 31 per cent of the catalogue, 47 per cent of the prominence, and 54 per cent of all viewership. Despite the concentration of viewership toward newer content, it is noteworthy that over 45 per cent of the total viewership was devoted to content older than three years. In comparison, a previous study on ERR showed that for linear television, this figure reached only 22 per cent, even though the linear programming and the VoD catalogue were similar in terms of production years (Lobanov & Ibrus, 2026).

Figure 5

To further illustrate these tendencies of diversity transformation, it is useful to examine the type of content most commonly promoted in the hero position on the homepage. The hero position is used almost exclusively to promote scripted content: 48 per cent of the prominence is allocated to series and 47 per cent to feature films, while all other content types individually account for no more than 1 per cent of the total prominence. In terms of origin, EU productions dominate (28%), followed by Estonian productions (18%), co-productions (15%), and titles from the UK and North America (14% each). Hero prominence is also concentrated on recent productions, with titles from 2020–2025 accounting for 62 per cent of the total prominence and productions from 2010–2019 for a further 25 per cent, whereas content produced before 2000 receives only marginal prominence. All in all, the contrast between the catalogue and the hero position makes the diversity transformation particularly visible.

Predictors of viewership

To examine the relative associations of prominence, broadcasting, premieres, and duration of availability with viewership, two multiple linear regression models were estimated (see Table 1).

The dependent variable was the number of streaming visitors per title. Model 1 included prominence score, broadcast score, premiere score, and days available. Model 2 additionally included production year, content type, and origin as control variables, allowing the associations between the main predictors and viewership to be examined while accounting for the metadata categories. Model 1 explains 62.5 per cent of the variance in viewership. Prominence (β = 0.50) and premiere score (β = 0.49) are the strongest positive predictors. Broadcast score (β = −0.09) shows a small negative coefficient, while duration of availability is also weakly and negatively associated with viewership (β = −0.05).

Adding the metadata categories in Model 2 increased the explained variance from 62.5 per cent to 64 per cent (ΔR² = .015, ΔF(18, 3978) = 9.13, p < .001). Although the improvement is statistically significant, its magnitude is modest. More importantly, the coefficients for prominence (β = 0.47) and premieres (β = 0.51) remain largely unchanged. Broadcast score and duration of availability retain small negative coefficients (β = −0.10 and β = −0.03, respectively). Among the control variables, country of origin and content type are statistically significant as categorical predictor groups, whereas production year is not. Model 2 thus shows that the strong associations of prominence and premieres with viewership cannot be explained simply by differences in content type, country of origin, or production year.

The negative coefficients for broadcasting and duration of availability require some explanation. In the bivariate analysis, broadcast score was positively correlated with viewership and prominence (both r = 0.59), and particularly strongly with premieres (r = 0.79). The change in the broadcast coefficient when prominence and premieres are included simultaneously suggests overlapping explanatory information among the predictors and a possible suppression pattern. The negative broadcast coefficient should therefore not be interpreted as evidence that broadcasting reduces VoD viewing. Also, these associations may reflect overlapping editorial and scheduling logics, whereby content that has been recently broadcast or premiered often receives increased prominence on VoD.

Similarly, duration of availability has a weak positive bivariate correlation with viewership (r = 0.06), but its coefficient becomes slightly negative when prominence, broadcasting, premieres, and content characteristics are taken into account (β = −0.03 in Model 2). One possible explanation lies in the composition of the catalogue: Domestic productions often remain permanently available, whereas acquired content may be available for shorter periods while receiving greater editorial prominence and attracting more viewers. However, the small negative coefficient indicates that duration of availability has little independent explanatory value in the model, rather than suggesting that longer availability reduces viewership.

Taken together, the regression results identify prominence and premieres as the strongest predictors of viewership in the models examined.

Variations across temporal scales and metadata categories

Although the regression models identify the strongest predictors across the catalogue as a whole, the descriptive measures and bivariate correlations reveal substantial variations between content groups and time periods.

First, because the data were collected daily over the course of a year, the analysis allows prominence and viewership to be examined at both daily and annual levels. The correlation between prominence and viewership is moderate in the daily data (r = 0.57) and stronger in the annual data (r = 0.69). This difference is consistent with the cumulative nature of VoD viewing, whereby titles can receive prominence and attract viewers at different points during the observation period. An immediate effect of prominence would imply that viewers watch highlighted content on the same day it becomes visible to them. In practice, however, audience responses may also be delayed. Content made highly visible may be remembered but not immediately viewed; users may save it to their watchlists for later viewing, or they may encounter it at a point when its visibility has decreased but it remains available in sections relevant to them. However, because the two correlations are calculated at different levels of aggregation, their difference does not establish that long-term effects are stronger than immediate effects. Repeated exposure, memory effects, and delayed discovery are plausible mechanisms, but they cannot be directly identified from these correlations.

Duration of availability, by contrast, has only a very weak annual correlation with viewership (r = 0.06), differing from the findings of Thurman and colleagues (2023). Thus, a longer observation period gives more titles the opportunity to attract at least some viewership, but keeping an individual title available for longer does not necessarily mean that it will attract more viewers, particularly in a catalogue where a substantial share of older content remains permanently available but may receive little or no prominence. Taken together, these findings reveal a stronger association between prominence and viewership in annual than in daily data, alongside a very weak association between duration of availability and viewership.

Second, prominence and its association with viewing differ markedly across content types. Scripted series receive the highest average prominence score (174), followed by feature films (143), news (108), and informative content (106), compared with an overall average of 105 among presented titles. However, the correlation between prominence and viewership is above average for slightly different content types. For example, culture and arts content receives significantly less visibility (85) than average, yet the visibility it does receive is strongly correlated with viewership (r = 0.93). Scripted series receive the highest level of prominence, and this visibility is strongly correlated with viewership as well (r = 0.81). In contrast, feature films also receive high prominence, but the correlation between prominence and viewership is considerably weaker (r = 0.50). A different pattern emerges for music content, which receives little prominence (68), and its association with viewership is likewise weaker than for most other content types (r = 0.45). These patterns indicate that prominence does not translate into audience attention in the same way across content types. Editorial priorities and audience preferences sometimes align, but at other times diverge.

Strong correlations can be observed between viewership and broadcast score (r = 0.59), and between viewership and premiere score (r = 0.68) as well. News, infotainment, informative, and culture and arts content show stronger than average correlations with both indicators (r = 0.70–0.89), suggesting that audiences for these content types are highly responsive to what has recently aired or premiered, a pattern that is understandable given their topical and time-sensitive nature. By contrast, scripted series and feature films display much weaker associations (r = 0.42–0.56), implying that their on-demand audiences rely less on linear television cues and are guided by different viewing logics. In scripted content, the viewing value tends to endure longer, making its connection to current events or novelty less critical.

Third, differences in prominence are also evident across countries of origin. Acquired foreign content, available for a limited time, receives much more prominence than domestic productions (70) that usually remain permanently accessible. Content from North America and the UK, in particular, receives especially high prominence (228 and 151, respectively). The audience seems to respond positively to this higher visibility, as across most foreign groups the correlations between prominence and viewership are strong (r = 0.71–0.85). Another aspect of this trend is that the vast majority (89–97%) of foreign content in the catalogue is presented, while only 60 per cent of domestic content is highlighted over the course of a year.

Regarding broadcast and premiere scores, content from closer regions (Estonia, the EU, and the Nordics) shows stronger correlations with viewership, suggesting that audiences rely more on novelty when engaging with local and culturally proximate content. In contrast, content from more distant origins exhibits weaker correlations, indicating that novelty and broadcast exposure play a much smaller role in driving viewership for these regions.

These variations provide an important qualification to the aggregate regression findings. Although prominence and premieres emerge as the strongest predictors across the catalogue, the extent of visibility and the strength of their bivariate associations with viewership differ considerably across content types and countries of origin. The findings thus highlight the role of editorial decisions in determining which content becomes visible, while also revealing how the relationship between prominence and audience attention varies across content groups.