Abstract
This study challenges dominant assumptions that user autonomy on streaming platforms is largely overridden by algorithmic systems. Based on a national survey (N = 1,501), we identify three distinct modes of discoverability, browsing, searching, and researching, that reflect how users discover content within and beyond platform structures. Browsing involves undirected exploration through curated menus and algorithmic recommendations; searching is a directed strategy shaped by catalog availability and interface design; and researching relies on external inputs such as social recommendations, media coverage, or third-party tools. These user practices demonstrate that discoverability is not a fixed platform outcome, but a contingent process negotiated between user behavior, platform affordances, and social context. By combining information-seeking theory with critical media research on discoverability, we propose a contingent discoverability model that offers a more nuanced account of how content becomes visible, and how users exercise agency within structured systems of guidance.
Keywords
Introduction
The accessibility of audiovisual content has reached unprecedented levels with the rise of streaming platforms. Subscription video-on-demand (SVOD) services have transformed viewing habits and content consumption, offering flexibility that surpasses traditional television (Evens et al. 2024). As platforms have become the main gateways to entertainment, understanding how users discover content is crucial, especially as domestic media industries face global competition for local audiences (Doyle 2023). Although platforms collect vast user data, their reliance on selective, non-standard metrics obscures what counts as “trending” or “popular” (Wayne 2022) raising concerns about cultural preservation and the sustainability of regional media (Sundet and Lüders 2023). These issues highlight the need for policies to enhance the discoverability of regional content, addressing an urgent demand for balance in an increasingly transnational market (Øfsti 2023).
Research on streaming media has largely examined how platforms affect the discoverability of content (McKelvey and Hunt 2019), through recommendation algorithms (Pajkovic 2022), personalized thumbnails (Eklund 2022), and curated lists (Thurman et al. 2024), thereby shaping how users interact with content (Evens et al. 2024). This work raises questions about user agency, understood here as the capability to act otherwise (Giddens 1984). Although users experience choice, platforms constrain options, embedding discoverability in power dynamics tied to economic and cultural structures (Lüders and Sundet 2022; Pilipets 2019; Van Esler 2021).
Less attention has been paid to how users actively exercise agency in their exploration. Few studies have empirically mapped the range of behaviors within streaming platforms, particularly how users balance platform-driven recommendations with self-directed search. To address this gap, we ask: How do interactions between users and streaming platforms, through features such as recommendation algorithms, search functions, and curated lists, mediate varied approaches to content discovery?
Our study contributes by developing a model that integrates user agency and platform structure, clarifying the conditions under which users actively direct their choices versus when they are more passively guided by algorithms or interface influence. Analyzing diverse user interactions with streaming interfaces thus advances a richer understanding of discoverability as a dynamic process shaped by both intentional user behaviors and platform affordances. Drawing on survey data from Norway, we identify three distinct user modes of discoverability, browsing, searching, and researching, that together form the empirical foundation for the contingent discoverability model developed in this study.
Previous Research and Theory
Following Giddens (1984), we approach agency as the capability to act otherwise, though always exercised within structural constraints. In streaming contexts, this agency is contingent: interface limits undermine promises of user control (Van Esler 2021), while choice architectures such as prominence shape decision-making (Lobato et al. 2024; Thurman et al. 2024).
Recent work highlights prominence as a more concrete mechanism, referring to the positioning and highlighting of certain content or services within catalogs and interfaces, decisions that strongly shape audience choices (Johnson et al. 2024; Lobato et al. 2024). This emphasis resonates with earlier discussions of visibility as a modality of power in digital media (Bucher 2012), where what is foregrounded or obscured reflects underlying platform logics. In contrast, discoverability concerns the likelihood that users come across content, actively or serendipitously, and thus extends beyond prominence to include the institutional, technological, and cultural conditions that govern access and exposure (Iordache et al. 2025). McKelvey and Hunt draw on Fred Turner’s concept of surrounds to describe how “platforms arrange choices on or between screens” and define discoverability as the interplay between these surrounds and vectors, “the pathways users take to find content” (McKelvey and Hunt 2019, 1–2). Their framework highlights how discoverability is co-produced by platform architectures and algorithmic logics that shape user movement and engagement.
Building on these insights, our research focuses on the behavioral side of this interplay. Rather than adopting McKelvey and Hunt’s terminology directly, we use their framework as a conceptual backdrop to analyze how user agency operates within these platform-structured environments. We develop a layered view of discoverability that integrates user agency with platform influence, examining four dynamics: algorithmic steering, gatekeeping mechanisms, and cataloging, and social recommendations.
Algorithmic Steering: Guiding Choice Through Design and Data
Algorithmic steering is visible in the ways platforms both design interfaces and control information about what counts as popular. Although streaming platforms promote extensive user choice, they often employ selective data releases, such as Netflix’s tendency to publicize only certain viewership metrics, to reinforce platform-defined popularity rather than actual viewer interest (Wayne 2022). This controlled narrative and algorithmic design guides user behavior within set boundaries (Evens et al. 2024; McKelvey and Hunt 2019; Pajkovic 2022; Van Esler 2021). For instance, Eklund (2022) highlights Netflix’s personalized thumbnails as part of an algorithmic culture, designed to capture attention through customized visuals that subtly direct users toward specific titles. Pajkovic (2022) similarly critiques Netflix’s recommendation system, describing its use of circular and economic logic to influence taste formation by reinforcing existing viewing patterns, which limits the variety of content users encounter. This aligns with McKelvey and Hunt’s (2019) concept of discoverability as media power, where platform design subtly guides users through “vectors” or pathways, directing choices to align with corporate interests. Extending this argument, Evens et al. (2024), note that “technological affordances” like navigability and interactivity are presented as user-friendly but are carefully structured to maximize user retention within the platform. This is conceptualized as choice architecture by Thurman et al. (2024), and emphasizes how content positioning, particularly on the opening interface rows (such as “Top Picks” and “Trending”), functions as an influence mechanism, significantly predicting what users choose to watch. Van Esler (2021) further critiques how Netflix’s interface design prioritizes platform-owned content, creating an illusion of user-driven discovery while reinforcing brand engagement, and Cox (2018) describes the Netflix interface as a form of «flow» with an «industrial control» comparable to that of linear television. These studies collectively suggest that user autonomy is often constrained by algorithmic steering, which positions platforms not as neutral content providers but as active agents in shaping cultural consumption.
Gatekeeping: Filtering Access to Catalogs and Interfaces
Gatekeeping is a critical mechanism through which streaming platforms control content availability and discoverability. Colbjørnsen et al. (2021) describe several levels, including legal, technological, and financial, that determine content presence in streaming catalogs and platform interfaces. Each level is described as a potential stumbling block for availability and discoverability. Platforms not only curate which content appears prominently but also determine how content is categorized and filtered. Prominence, or the prioritized placement of content, differs from broader discoverability, which, as Mazzoli (2020) notes, involves multiple gatekeeping roles that shape content circulation. This gatekeeping shapes users’ perceived options, often guiding them toward content that aligns with the platform’s economic goals. A strong predictor of content viewership is where and how long something is positioned on the interface, as Thurman et al. (2024) show. Bucher (2012) similarly emphasizes the architectural organization of visibility, where platforms control what is seen or unseen, compelling providers to adapt to platform standards. Van Esler (2021) points out that platforms often prioritize algorithmic recommendations over traditional search and browse functions, which typically offer greater user control. Similarly, Johnson (2019) observes that customizable filters, common in library searches, are rarely available on online video platforms. This restricted access reinforces the notion of platform power, positioning streaming platforms as gatekeepers that not only influence but limit content diversity through their design and categorization choices.
Cataloging: Organizing Libraries and Shaping Exposure
Catalogs structure vast content libraries to facilitate discovery. Helberger et al. (2015) and Van Esler (2021) draw an analogy between streaming catalogs and libraries, suggesting they offer users the freedom to browse. However, this browsing freedom is mediated by platform-specific recommendation systems, which actively shape user engagement with content. Recommendations vary widely in form and intent, from personalized suggestions (“My List”) to algorithm-driven lists (“Top 10”) and system-generated categories. McKelvey and Hunt (2019) argue that such recommendation systems create individualized content flows, leading to a feedback loop that reinforces established viewing preferences rather than promoting genuine content diversity. Eklund (2022) adds that user interaction data is continuously fed back into recommendation algorithms, creating a self-reinforcing loop that adjusts suggestions based on prior behavior. Extending this perspective, Rodríguez Ortega (2024) introduces the idea of networked memory, where streaming platforms create a personalized spectator-interface that prioritizes individual consumption histories, subtly distancing users from shared cultural experiences in favor of more isolated, algorithmically shaped encounters. This approach restricts users to content types they are already familiar with, limiting their exposure to new genres or diverse voices. Pajkovic (2022) critiques this algorithmic “taste-making,” noting that Netflix’s interface uses popularity indicators and trending metrics to further prioritize specific content. Thus, while streaming catalogs appear vast and unrestricted, platform design often limits actual discoverability, encouraging users to remain within familiar genres and recommendations. Bengesser et al. (2024) find that while there were between 657 and 935 British titles in the Netflix catalogs in Denmark, Germany, the Netherlands, and Italy respectively, in any one of these “only about 200 titles could be reasonably found (. . .), while only about 50 titles are promoted” (Bengesser et al. 2024, 17).
Social Recommendations: Discovery Beyond the Platform
In contrast to platform-driven recommendations, users also rely on social recommendations from family, friends, and external sources, such as social media or third-party streaming search engines like JustWatch and PlayPilot. These external recommendations can provide alternative discovery paths, introducing users to diverse content that may not be featured prominently within platform interfaces. Social media and news media also act as discovery vectors, offering suggestions outside the platform’s ecosystem, which can counterbalance algorithmic preferences. Rodríguez Ortega (2024) expands on this by noting that platforms shape cultural preferences indirectly through user engagement with socially endorsed content, which is often integrated into the platform’s recommendation systems. This fusion of social recommendations and platform design adds a layer of complexity, as users navigate between platform-curated lists and socially suggested content. Although social recommendations provide users with diverse perspectives, they also indicate the extent to which platforms retain control by incorporating external recommendations into their algorithmic feedback loops.
Summing up, the literature suggests that while streaming platforms offer an extensive range of content, the discoverability of this content is moderated by platform power dynamics that balance user autonomy with algorithmic and design-driven control. Algorithmic steering, gatekeeping mechanisms, and cataloging structures enable platforms to influence viewing behaviors by giving prominence to specific content while limiting discoverability through genuine exploration. Social and external recommendations offer a counterbalance but are often reabsorbed into platform algorithms, creating a hybrid model of discoverability where user agency is present yet constrained. Still, empirical evidence of real-world audiences, and how their interactions with platforms shape discoverability is missing in the current literature. Further research is needed to develop a comprehensive model that integrates user agency and platform influence, clarifying under which conditions users actively explore content versus when they are guided by algorithmic design and interface features.
Analytic Design: Four Modes of Information-Seeking
The understanding of media usage has evolved from early user-centric views of audiences as rational and goal-oriented (Blumler and Katz 1974) to media-centered perspectives that highlight media’s role in shaping information needs and use (Rubin and Windahl 1986). While these historical perspectives laid the groundwork for thinking about discoverability, they must be adapted to contemporary digital environments, including streaming services.
Here, information science provides a bridge: it expands media-oriented approaches by addressing the infrastructures, technologies, and user practices that underpin information seeking (Savolainen 2022). Although recent work stresses sociotechnical notions of infrastructures and their implicit power structures (Borgman 2015; Haider and Sundin 2019), the field’s cognitive roots-focused on mental models, user needs, and search strategies-remain useful for analyzing discoverability (Belkin 1990; Wilson 1999). These perspectives help explain how audiences navigate streaming services where the line between information and entertainment is often blurred (Savolainen 2022).
We apply information seeking theory here because it provides conceptual tools for distinguishing between different modes of user behavior. Bates (2002) four modes are particularly useful for analyzing discoverability on streaming platforms, as they capture both intentional searches and more exploratory, serendipitous forms of engagement (detailed in Figure 1). The active and passive dimensions denote varying levels of user engagement in the information acquisition process. Directed seeking involves specific targeting of desired information, whereas undirected seeking entails users exposing themselves to information in a more exploratory way, sometimes by chance rather than deliberate search.
Four modes of information-seeking, adapted from Bates (2002).
These spectra form four modes of information engagement. Being aware represents how much of our knowledge is acquired through undirected and passive behaviors, characterized by a state of simply being aware of information in our surroundings, as exemplified by the “news finds me perception” (Gil de Zúñiga et al. 2017). Monitoring is also a passive mode but directed, where we maintain back-of-the-mind alertness for things that interest us. This aligns with the idea that individuals allocate their limited attention resources strategically, prioritizing information that is relevant or appealing to them while navigating a landscape of abundant stimuli. By monitoring, we do not actively seek information, nor do we simp take in information by being aware, but we gather what we encounter that helps answer questions and satisfy seeking needs that we might have. Browsing is a counterpart to monitoring, where we actively expose ourselves to information, but without a specific goal or to satisfy a particular need. Browsing may lead us into unexpected encounters that we may or may not follow up on. Searching is an active behavior directed toward specific content and known items, for example when we search for a movie that we know we want to watch.
These four modes are not straightforward or mutually exclusive, and they may be entered simultaneously or in a sequence. The active search strategy can for instance be nuanced by a focused and non-focused mindset (Li et al. 2019), where the focused searcher formulates exact queries targeting specific titles on a streaming platform, and the non-focused searcher queries genres and addresses broader characteristics of the content universe. Passive monitoring may lead us to formulate a directed search and so forth.
Building on these modes of engagement and specification of targets, our research aims to further develop this model by integrating user agency and platform influence. We propose a comprehensive approach that explores when and how users actively explore content versus when they are subtly guided by platform features like recommendation algorithms, search functions, and curated lists. This model will bridge a theoretical gap in streaming media research, offering insights into the conditions that encourage user-driven discovery and those that align with platform-guided interaction. Through this nuanced perspective, we aim to advance the discourse on streaming discoverability, capturing the dynamic interplay between user intentions and the affordances of streaming interfaces.
Method and Data
Our data derive from a national web survey (November 2022) on attitudes towards and use of cultural goods in Norway (movies, streaming, books, museums, and media). Although single-country, the case is analytically significant: Norway is not an EU member but, through the EEA, implements much EU media and competition regulation (Syvertsen et al. 2014), giving audiences access to the same global streaming infrastructures as EU users. Consumption is further shaped by external structural conditions, a small language-specific market with high purchasing power, strong public service institutions, and near-universal broadband (OECD 2024), alongside this regulatory alignment (Bengesser 2024). Together, these conditions underpin record streaming levels: nearly 80 percent of households subscribe to at least one service, with growth across Netflix, TV2 Play, Prime Video, and others (Wordbank 2025). This hybrid context combines Western European traits, global platform access, and high digital adoption. It thus illustrates how national and global dynamics interact, making Norway useful for understanding broader European streaming and discovery patterns while highlighting possible variation from larger EU markets (Hallin and Mancini 2004; Syvertsen et al. 2014).
The sample (N = 1,501) is 51 percent male and 49 percent female. Median yearly household income is 600,000 to 700,000 NOK ($60,000–70,000), representative of the national median income (Statistics Norway 2025). Respondents are geographically dispersed with this age distribution: 15 to 29 (22.6%), 30 to 39 (16.6%), 40 to 49 (15.8%), 50 to 59 (16.1%), 60+ (28.8%).
The survey section analyzed for this research covers twelve questions on interaction with streaming services’ interface. Some of the questions were about the use of scrolling, searching, and browsing for content, while others were about the use of personalized recommendation features (such as “for you” and “your list”) or the platform’s recommendation system, tips from family and friends or recommendations from external sources (news media, streaming guides, editorial reviews, social media). Other questions of potential relevance include background demographic data.
The data were weighted by gender, age, area of residence, and education to correct for web panel deviation. Analysis proceeds in three steps. First, descriptive patterns are identified and aligned with Bates (2002) model. Second, cluster analysis addresses how platforms’ dual roles, actively guiding and passively cataloging content shapes discovery. Third, a principal component analysis with Oblimin rotation was conducted on the platform-use items. As for the use of weighted or unweighted data, no substantial differences in the regression model estimated were found. As the results from the unweighted analysis will lead to more efficient reporting and minimizing of the estimated standard errors is preferred, the unweighted analysis is reported (Winship and Radbill 1994).
Results
Applying Bates (2002) information-seeking model (Figure 1), we first describe users’ content discovery behaviors along four theoretical dimensions of activity and direction. This provides the groundwork for the subsequent factor analysis, which identifies three main empirical modes of discoverability.
Descriptive Findings
Of the sample (N = 1,501), 79 percent of respondents used online streaming to watch TV series and films. Searching represents an active, directed mode where users seek specific content. This is reflected in our data, with 88 percent of respondents often (39%) or sometimes (44%) using the search function to find content. Likewise, 76 percent of respondents explore categories or genres, 28 percent often and 48 percent sometimes, as part of this directed search. External streaming search engines such as JustWatch and PlayPilot further exemplify an active, targeted discovery strategy, with 21 percent of respondents using these tools, though only 5 percent use them frequently.
Browsing, an active, undirected mode, describes users exploring content without a specific goal. Here, 55 percent of respondents reported always (13%) or often (42%) browsing by scrolling through titles. Recommendation features, including “Top 10 lists” and “newly added content,” also support this behavior, with 26 percent of respondents often and 37 percent sometimes engaging with these options. Additionally, 64 percent rely on algorithm-based recommendations (23% often, 41% sometimes) as part of their exploratory browsing.
Monitoring is a passive, directed mode where users stay alert to relevant content without actively searching. In our findings, 64 percent of respondents engage with recommendation algorithms based on their viewing history (23% often, 41% sometimes), passively tracking content aligned with past preferences. Similarly, 63 percent use “My List” and “My Content” features to save titles for later (28% often, 35% sometimes), maintaining awareness of specific content without immediate engagement. Social navigation through recommendations from family and friends also supports monitoring, with 84 percent of respondents finding content this way, 31 percent doing so often and 53 percent sometimes.
Awareness represents a passive, undirected mode where users are exposed to information without seeking or targeting content. From the user perspective, social navigation through recommendations from family and friends can be described as passive awareness, with 84 percent of respondents finding content this way, 31 percent doing so often and 53 percent sometimes. In addition to social media (65%) and editorial news media (54%) recommendations occasionally inform users of new content. However, neither of these fully capture the passive, undirected exposure that defines passive awareness, as these sources often prompt some degree of directed attention or intentional follow-up.
Summing up this part of the analysis, most respondents actively decide to search with specific goals, while about half engage in browsing behaviors, scrolling through titles without a clear target. Platforms enable discovery through recommendations, with personalized suggestions (e.g., “because you watched. . .”) being slightly more popular than general options (e.g., “new releases”), and category exploration emerging as the most common behavior. However, Bates (2002) conceptualization of passive discovery does not align clearly with our results. When platforms do not facilitate discovery and users do not take deliberate action, our findings reveal significant levels of social navigation. Recommendations from family and friends, social media, and traditional news media are influential sources, while streaming search engines play a more limited role. These findings suggest that Bates’ notion of passive awareness insufficiently accounts for the social and contextual factors that shape content discovery in practice, highlighting a limitation in applying the model to our data.
Findings from Factor Analysis
Our descriptive findings reveal distinct patterns among streaming users that extend beyond Bates (2002) original framework, particularly regarding active and passive engagement, and the role of platform-driven and social recommendations. To validate and refine these behavioral dimensions, we conducted a factor analysis to identify the core components underlying content discovery behaviors on streaming platforms. This analysis distilled and redefined key information-seeking dimensions, providing the empirical basis for a new model that more accurately reflects how streaming platforms shape user discovery experiences. Building on the descriptive analysis, we grouped respondents according to these patterns to address the research question of how streaming platforms’ dual roles, actively guiding and passively cataloging content, affect users’ discovery behaviors.
The survey included items for all constructs, and respondents indicated how frequently they used each interface function on a 5-point scale, from 1 (always) to 5 (never). We retained factors strong enough to explain the data (with an eigenvalue of 1 or higher) and with at least two items meeting the 60-40 rule, showing clear groupings. The items that loaded onto each factor were then summed and averaged. Together, these factors explained 39.36 percent of the total variation in responses after adjustment. Reliability for all constructs, measured by Cronbach’s alpha, was well above the acceptable level of .7, indicating consistency. Table 1 presents the initial results, revealing three main modes of discoverability based on user behavior: browsers, searchers, and researchers.
Factor Analysis.
The discoverability mode of browsing corresponds to an interface-dependent and largely undirected form of discovery. It allows the platforms to take an active attention-steering role through personalized recommendations based on the users’ past behavior and preferences such as “for you” and “your lists.” In total 26.4 percent of the respondents fall under this mode. Although the user actively engages with the interface, navigating through various menus, recommendations, and categories, the experience is to varying degrees undirected and thus significantly shaped by the platform’s design, interface, and algorithms. These systems influence what content is presented and how it is categorized. Thus, while users may feel they are making autonomous choices, all their interactions are within the confines of the platform and are heavily mediated by the platform’s structure and recommendation algorithms, indicating high platform constraint on user agency.
The second mode of discoverability commences with 38.4 percent of users relying on their directed searching to find what to watch in the streaming services’ catalog of content. The search mode represents a catalog-dependent, directed behavior within the platform. These users predominantly use this as their means to find content, but if we include those who use this “often” rather than “always” the mode grows to 72.2 percent of the respondents. While the use of search suggests high user agency, the search function is a part of the platform’s structure which both enable and constrain users. However, as these users also rely highly on recommendations of friends and family, and their discoverability is not wholly within the platform. At the same time, the browsers and the searchers are both likely to simply start browsing on the front page of the platform.
The third mode of discoverability primarily relies on researching sources outside the streaming services, such as news media and streaming search engines like PlayPilot or JustWatch, to determine what they will watch. It combines content-dependent and context-dependent discovery, extending beyond the platform. Those who utilize this mode constitute 7.3 percent of the respondents in the survey. For the most part, their behaviors are based on these external recommendations. Some of these behaviors such as using third-party services like Justwatch etc. to find content are more active while others, such as receiving recommendations from social media are likely to be more passive.
Summing up, the results identify three distinct modes of discoverability: browsing, searching, and researching. Browsing is interface dependent and largely undirected, shaped by the platform’s design, recommendations, and algorithms. Searching is catalog dependent and directed, reflecting users’ active search within the boundaries of the platform structure. Researching extends beyond the platform, combining content-dependent and context-dependent discovery through external sources such as news media, streaming search engines, and social media.
Discussion
This section develops the contingent discoverability model based on three user modes of discoverability: browsers, searchers, and researchers. The model conceptualizes discoverability as a negotiated process between user agency and platform influence. In contrast to earlier platform-centered or governance approaches (Iordache et al. 2025; Mazzoli 2020; McKelvey and Hunt 2019), the model highlights how discoverability is contingent on user behavior, platform algorithms, and contextual factors that both enable and constrain user agency.
Negotiating Discoverability Within and Beyond Platforms
The contingent discoverability model specifies two intersecting dimensions: whether user behavior is directed or undirected, and whether discovery occurs within or outside the platform (Figure 2). This approach incorporates an information-seeking perspective to highlight user agency, recognizing that discovery also involves active negotiation and not just navigation of platform structures, as McKelvey and Hunt’s (2019) concept of vectors suggests.
Contingent discoverability.
Within the platform, discoverability unfolds through catalog-dependent and interface-dependent negotiations between user agency and algorithmic control. Directed behaviors, such as searching for specific titles or navigating catalog menus, illustrate how users act within the constraints of what is available. Undirected behaviors, such as browsing through interfaces and algorithmic recommendations, are likewise guided by the structures and affordances of the platform. Both modes of discovery intersect with algorithmic processes that promote certain content over others, as platforms integrate socially endorsed and commercially prioritized material into their systems (Wayne 2022). Discoverability within platforms therefore remains a negotiated process rather than a user-driven one, continuously shaped by the boundaries of platform power (Mazzoli 2020). This aligns with Lobato et al. (2024), who show that interface-level prominence on smart TVs similarly constrains user agency through preinstalled apps and default placements.
Beyond the platform, discoverability is shaped by content-dependent and context-dependent processes, and users acquire knowledge from a wider social and media context (Iordache et al. 2025). Our survey findings show that these external pathways influence content discovery through both directed actions, such as searching for reviews of new releases, and more undirected exposure, such as receiving unsolicited recommendations from friends and family.
More importantly, the analysis shows that platform catalogs and interfaces not only enable and constrain user agency but are themselves influenced by content and context. For users who know what they are looking for, discoverability within platforms becomes largely catalog-dependent. For those who are uncertain, discoverability becomes interface-dependent, guided by browsing and recommendation functions. Users who primarily rely on external sources engage in content-dependent discovery, where attributes such as genre, quality, and creative personnel determine discoverability in news media, third-party applications and other channels. Finally, users who are not actively searching may still encounter content through contextual cues, such as social media or word of mouth, making their discovery context-dependent.
The contingent discoverability model thus captures the complexity of real user behavior, showing how discoverability is simultaneously enabled and constrained by user agency, platform design, content characteristics, and social context.
Discoverability in Practice: Browsers, Searchers, and Researchers
We have positioned the observed discovery behaviors along the four interrelated conditions of discoverability-interface, catalog, content, and context-to illustrate how these factors jointly shape user behavior rather than determine it absolutely (Figure 3).
Typology of discoverability pathways on streaming platforms.
The browsing mode represents interface-dependent discovery. Whether content is presented through algorithmic recommendations or editorial lists, user agency remains limited by the constraining structure of the interface. User needs are served only insofar as they align with the platform’s commercial objectives (Cox 2018). In line with Evens et al. (2024), we found that personalized recommendations are not commonly used, possibly because algorithms promote generic content regardless of user interests or viewing history (Lüders and Sundet 2022). As Van Esler (2021) and McKelvey and Hunt (2019) note, recommendations are shaped by platform business goals, privileging originals and blockbusters. Our findings show high user awareness of such mechanisms and recognition of platform strategies.
The searching mode is fundamentally enabled by the platform catalogs. While traditional television relies on a scheduled zapping experience dictated by broadcasters, streaming platforms introduce a user-centric circular flow (Evens et al. 2024). Viewers have the freedom to navigate extensive content libraries guided by algorithmic interfaces. While the use of search functions on platforms is more directed and thus increases user agency, the constraining effects of these structures are most evident in the composition of catalogs, the placement of search functions, and the results returned. Many platforms, notably Netflix, guide users toward related available titles when searching for content not in the catalog. Few, if any, SVOD platforms offer the kind of filtering options found in traditional library searches (Johnson 2019). Still, the high share of users employing a searching mode suggests that catalog presence matters, even for content with low interface prominence. Moreover, the correlation between searching and recommendations from friends and family indicates that searching is also shaped by contextual factors outside platform control.
Users relying on external sources for content discovery represent the researcher mode, which operates beyond the platform and depends on both context and content. The main drivers of discoverability here are traditional news media, social media, and third-party services. Recommendations from one’s social network are far more popular than automated suggestions, indicating that human curation helps address content overload and the dilemma of choice. This pattern also suggests that trust in information sources plays an important role in discovery. At the same time, the researcher mode is strongly influenced by the attributes of the content itself—such as quality, novelty, or creative personnel—which attract reviews and media attention and thereby increase discoverability. Platforms can only influence these users indirectly through marketing and catalog design.
Navigating Control and Prominence
Several researchers question the extent of user autonomy on streaming services, asserting that structural forces foster only a perception of control. As Lüders and Sundet (2022, 338) argue, “Certain programs are made more visible than others, search functions are downplayed in favour of pre-organized catalogues, and while recommendation algorithms depend on patterns of use, viewers cannot determine the criteria informing how algorithms work.” Similarly, Van Esler (2021, 736) critiques Netflix for interfaces that “reduce the agency of the viewer” by curating library titles rather than enabling personalized exploration. These critiques highlight how platforms prioritize economic goals through algorithms and catalog structures, often constraining genuine user choice.
Our findings, however, challenge the severity of these critiques. While algorithms influence user behavior, they also enable users to navigate libraries effectively. As Evens et al. (2024) note, technological affordances like navigability and interactivity provide users with practical tools to explore content, even if these tools are designed with retention and profitability in mind. Users relied on these pathways to expand options and challenge platform-curated constraints. The three distinct modes of discoverability could also be read as analogous to vectors in McKelvey and Hunt’s (2019) framework. Whereas their analysis centers on how platforms arrange and guide discovery through surrounds and vectors, our model emphasizes the varying ways users navigate and respond within these structured environments.
This interplay between platform control and user agency is evident in the two dominant modes of discoverability we identified: searching and browsing. Both searchers and browsers often begin their journeys on the platform’s front page, suggesting that platform influence and user autonomy coexist. Users acknowledge that platforms utilize their data to recommend content, but our findings show they value these features for their convenience and relevance. Feedback loops, while criticized for reinforcing narrow viewing habits (Lüders and Sundet 2022; McKelvey and Hunt 2019), were seen by users as pragmatic tools for tailoring recommendations to their preferences. In addition to personalized recommendations, more general market-based systems, such as popularity metrics and trending content, are appreciated for their role in simplifying content discovery. This dynamic demonstrates that user autonomy and platform influence are not necessarily antagonistic.
The contingent discoverability model highlights how streaming platforms and third parties jointly shape content prominence. Gatekeeping plays a central role in determining whether content appears in streaming catalogs and interfaces, operating at legal, technological, and financial levels (Colbjørnsen et al. 2021). This process creates significant barriers for content providers, where platforms prioritize content based on their economic interests (Wayne 2022). Our findings suggest that while gatekeeping limits diversity, users actively navigate these constraints. Many rely on external discovery paths, such as social media recommendations, news outlets, or third-party tools like PlayPilot, that expand their viewing options beyond platform algorithms. These external pathways align with Rodríguez Ortega’s (2024) concept of networked memory, where socially endorsed content becomes integrated into platform algorithms, creating a feedback loop that blends internal and external discoverability mechanisms. This integration illustrates that while platforms maintain control, users actively leverage external recommendations to access more diverse content.
At the same time, platforms also exert influence outside of their interfaces, for example in the form of advertising and the presence of dedicated buttons on remote controls or pre-installed smart-TV apps (Lobato et al. 2024). Nor are external recommendations free from platform influence. Platforms often absorb socially endorsed content into their algorithmic systems, reinforcing a prominence hierarchy shaped by both external inputs and internal mechanisms. For instance, as Colbjørnsen et al. (2021) note, social media advertising can generate awareness for independent producers, driving initial viewership that may result in favorable algorithmic treatment. Our findings demonstrate that users value these blended pathways as a way to navigate the abundance of content and mitigate platform-imposed constraints.
This dynamic interplay between platform design, user agency, and external efforts underscores the complexities of discoverability. While platforms undeniably restrict content diversity through practices like algorithmic taste-making (Pajkovic 2022) and feedback loops (McKelvey and Hunt 2019), external recommendations help counterbalance these biases. Users in our study relied on these pathways not only to expand their options but also to challenge the constraints of platform-curated recommendations. Ultimately, the contingent discoverability model reveals that discoverability is shaped by multiple layers of influence, where platform structures, external efforts, and cultural contexts interact in complex ways. This highlights the need for greater transparency in how platforms manage prominence and underscores the value of third-party tools and external recommendations in promoting a more equitable and diverse content ecosystem.
Conclusion
A broad field of media researchers (Evens et al. 2024; McKelvey and Hunt 2019; Pajkovic 2022; Thurman et al. 2024; Van Esler 2021) claim that the features of streaming platform interfaces significantly constrain viewer agency. Building on Bates (2002) model of information-seeking, this study extends the framework by integrating insights from algorithmic governance (Bucher 2012), choice architecture (Thurman et al. 2024), and media dependency theory (Rubin and Windahl 1986) to illustrate how user agency and platform influence interact. We find that both users and platforms have influence, and that this is not always in opposition. In addition, users have ways of discovering content that are not directly under the influence of the platforms. While robust, the findings are limited by their cultural context, reliance on self-reported data, and the rapid evolution of streaming platforms. Future research should explore diverse markets and long-term trends to further refine the contingent discoverability model and its applications.
The contingent discoverability model highlights that discoverability is not passive, but shaped by a dynamic interplay of platform design, user behavior, market conditions, and social context. While content prominence within interfaces is not the only factor that governs discoverability, the model shows how platforms influence discovery. Importantly, by emphasizing user agency, our model does not reproduce platform narratives of user choice but situates agency as contingent and negotiated within technological and institutional constraints. By repositioning discoverability as a negotiated process rather than a fixed outcome of platform control, the model contributes to a more dynamic understanding of how users and platforms co-produce access and attention in streaming environments.
Finally, while this article does not directly engage with ongoing European policy debates on prominence and discoverability, the contingent discoverability model supports calls for greater transparency and enforceable definitions of prominence, alongside policy measures that strengthen catalog diversity and promote genuine audience access.
Footnotes
Acknowledgements
We used ChatGPT (OpenAI, GPT-5) for language editing, grammar checking, and improving textual flow. The tool was not used for generating ideas, analyses, or interpretations.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was in part carried out in the research project EUVoD (2021-24) funded by Aarhus University’s Independent Research Fund (AUFF).
Ethical Considerations
Not applicable.
Consent to Participate
Not applicable.
Consent for Publication
Not applicable.
Data Availability
Not applicable.