Cross-platform sensitivity and algorithmic adaptability: How transnational creators navigate algorithms across Chinese and US-based platforms

Abstract

Social media content creators and influencers increasingly use multiple platforms to mitigate the risk of (in)visibility in the volatile algorithmic environment. This article focuses on transnational creators’ algorithmic knowledge and practices across Chinese and US-based platforms. Drawing on in-depth interviews, participant-led walkthroughs and online observation with transnational creators, this research finds that creators learn each platform’s algorithmic preferences through ‘cross-platform sensitivity’, which also informs their practices of ‘algorithmic adaptability’ in the ever-changing platform environment. Creators’ cross-platform sensitivity and algorithmic adaptability show forms of everyday resistance to survive and cope with algorithmic power in the precarious multiplatform environment.

Keywords

Algorithmic knowledge algorithms Chinese platforms creators influencers social media entertainment visibility YouTube

Introduction

Algorithms are architectures of online visibility (Bucher, 2012). To make their content visible across social media platforms, content creators and influencers adopt various strategies and tactics to optimise content and boost engagement rates and metrics. Their experiences are shaped by the promise and precarity of visibility, because of the volatile changes in markets, uncertainty and competition in industries, and the constant updates of platform features and algorithms (Duffy et al., 2021). Many creators know the importance of adopting multiple platforms to avoid being overly dependent on a single platform (Glatt, 2022). Some also expand their online influence across the boundary of the ‘parallel universes’ between the Chinese influencer industry, or the wanghong industry, and its Western counterpart built around US-based platforms (Craig et al., 2021: 7).

This article focuses on transnational content creators’ algorithmic knowledge and practice of making themselves visible across Chinese and US-based platforms.¹ The creators who are native to Chinese platforms and gradually going international are part of the ‘global wanghong’, ‘the regional and global extension of platforms, creator practices, and wanghong culture’ (Craig et al., 2021: 161). An example of a global wanghong creator is Li Ziqi, who broke a record for having a Chinese-language YouTube channel with the most subscribers (Zhan, 2021). In addition, the rise of the Chinese wanghong industry has attracted numerous international creators entering the Chinese-language market. More recently, the YouTube megastar MrBeast joined Bilibili and greeted his Chinese audience in an introductory video, ‘We have an audience basically all over the world except China, so I thought it would be cool to start getting the content over to China’ (MrBeast Official Channel, 2024). Although Li Ziqi and MrBeast’s popularity is phenomenal, they show examples of the cross-platform and cross-border flow of creators’ practices.

This research is significant in adding a cross-platform, cross-border perspective to the literature on algorithmic practice by investigating transnational creators’ knowledge production and practice of navigating algorithmic mechanisms across platforms. These platforms include but are not limited to Bilibili, Weibo, Xiaohongshu (Little Red Book or RED), WeChat, Douyin and ByteDance’s ecosystem, TikTok, YouTube and Instagram. Through mixed methods of in-depth interviews, participant-led walkthroughs and online observation with 16 creators located worldwide, this research examines how these creators perceive Chinese and US-based platforms’ algorithms differently, and what practices they develop to make their content visible. I argue that transnational creators learn each platform’s algorithmic preferences through ‘cross-platform sensitivity’, which also informs their practices of ‘algorithmic adaptability’ in the ever-changing multiplatform environment. Sensitivity and adaptability become transnational creators’ rules of survival in a situation of structural uncertainty imposed by algorithms across platforms.

Users’ algorithmic imaginaries and folk theories

Algorithms are encoded procedures that follow specified calculations for changing input data into a desired output, and they are ‘a key logic governing the flows of information on which we depend’ (Gillespie, 2014: 167). Algorithms are architectures of visibility on platforms, where users have to follow platform logic to avoid the possibility of content or accounts disappearing, or the ‘threat of invisibility’ (Bucher, 2012). Despite algorithms’ automated decisions in sorting, ranking, prioritising or filtering certain content, users have the agency to influence algorithmic outputs and respond to algorithmic power through various practices and actions (Bonini and Treré, 2024). This research draws on ‘everyday lived experiences of algorithms and their affects’ (Beer, 2017: 6), building upon the literature on user practice surrounding algorithmic (in)visibility on social media platforms.

Although platform algorithms have a black-boxed nature, scholars draw on users’ perceptions, experiences and practices to understand algorithmic power and its influences. For example, Bucher (2017) suggests that users encounter and make sense of algorithms through ‘algorithmic imaginary’, referring to ‘ways of thinking about what algorithms are, what they should be, how they function and what these imaginations in turn make possible’ (pp. 39–40). Users’ encounters and experiences with the platform play a generative role in shaping and moulding the algorithm, as it constantly evolves and changes through users’ practices and their data input (Bucher, 2017: 41). Users may not understand the functionality of algorithms, but they are aware of the potential of algorithms to forced connections, hence users develop tactics to intervene and resist the algorithms, attempting to wrest back their control over algorithmic power (Van der Nagel, 2018). Users can acquire ‘folk theories’, the ‘non-authoritative conceptions of the world that develop among non-professionals and circulate informally’, which in turn guide users’ behaviour on social media (Eslami et al., 2016: 2372). Although folk theories might seem intuitive and informal, in some cases, they can frame user practice of resistance to algorithmic change in social media (DeVito et al., 2017). Some scholars also consider algorithmic imaginary related to the line of work on folk theories (e.g. DeVito et al., 2017; Siles et al., 2020). For instance, Siles et al. (2020: 13) combine folk theories and imaginaries of algorithms, showing that users enact their agency through practice and imagination. The studies of imaginaries and folk theories of algorithms provide valuable insights into understanding how individuals make sense and feel about algorithms, and how they enact different ways of power and resistance towards algorithmic systems in the different platform environments.

Creators’ algorithmic knowledge and practice for visibility

The studies of folk theories show how users know about algorithms through operational and propositional facts, locating knowledge in thought; yet there are other forms of knowledge located in practices and actions (Cotter, 2022). The growing research focusing on social media content creators’ algorithmic practice, for example, is built on previous work on algorithmic imaginaries and folk theories, while extending the understanding to knowledge expressed via practice. Creators’ and influencers’ experiences are different from ordinary users’ perceptions of algorithms because of their content production, self-branding and other professionalising and commercialising activities (Burgess and Green, 2009; Khamis et al., 2017). Compared to ordinary users who adopt social media for entertainment and are situated at the receiver end of algorithms (Schellewald, 2022: 3), creative workers are more reliant on social media and highly incentivised to learn about ‘algorithmic skills’, or the knowledge of how particular algorithms work in determining online visibility and how the knowledge could be used to leverage content production and distribution (Klawitter and Hargittai, 2018).

The scholarship on algorithmic practices in creator cultures focuses on the process of algorithmic knowledge production and how it informs platform practices (e.g. Bishop, 2019, 2020), and the tactics to play the algorithmic games (e.g. Cotter, 2019; O’Meara, 2019). Bishop (2019) introduces the notion of ‘algorithmic gossip’, the ‘communally and socially informed theories and strategies’ that can inform creators’ (p. 2589) practices around algorithms. Common practices among creators include algorithmically optimising themselves and creating content that is consistent with a platform’s commercial goals to gain visibility, as the non-commercially compliant videos might be punished through relegation and obscuration (Bishop, 2018).

The tactics to optimise and intervene algorithms to gain visibility is what Cotter (2019) calls ‘playing the visibility game’. Cotter observes that Instagram influencers’ active pursuit of online visibility resembles a game with rules encoded in algorithms. Influencers learn the rules or the logic of Instagram algorithms, and play the game through visibility tactics in line with influencer practice of authenticity and entrepreneurship (Cotter, 2019). Stressing upon the stakes of the visibility game, O’Meara (2019) examines the phenomenon of Instagram influencer ‘engagement pods’, the grassroots practice of agreeing to mutually engage with each other’s accounts and posts to simulate interactivity to game the algorithm. This type of collective, organised algorithm hacking activity shows a form of worker resistance among influencers, and it is a response to the material conditions of platformised cultural production, where algorithms wrest knowledge and control of the labour process (O’Meara, 2019: 1).

Other industry stakeholders also contribute to the algorithmic expertise in guiding creators’ visibility practices, such as the self-branded algorithmic ‘experts’ on YouTube, who teach creators to be compliant with YouTube’s organisational strategies and business models (Bishop, 2020). Algorithmic expertise is also found in livestreaming guilds in China, optimising livestreamers based on the algorithmic parameters that some (usually small) Chinese platforms reveal (Liu et al., 2023: 202).

Algorithmic knowledge and practice in the Chinese context is expressed through ‘liuliang’, or ‘web traffic’, the quantitative user engagement data that is measured in the form of metrics, but liuliang has richer taxonomies and creative collocations and wisdom associated with it (Zou, 2023). Content providers in China choose to ‘play with’ or ‘please’ algorithms, as they are aware that algorithms play a decisive role in allocating liuliang and user attention (Zhang et al., 2021). Creators negotiate a balance between chasing traffic and restrictions imposed by platforms and the state (Chen et al., 2023). In recent years, there has been growing attention on Chinese wanghong industry (Craig et al., 2021) and algorithms’ impacts on gig workers and digital labour (e.g. Sun, 2019); however, creators’ practices surrounding algorithms and web traffic on Chinese platforms remain an underexplored area. Only a handful of research looks at traffic allocation and algorithmic manipulation practices by streamers, influencers and media practitioners on livestreaming and short video platforms (e.g. Lai, 2022; Liu et al., 2023; Su and Kaye, 2023).

Current literature on algorithmic literacy and visibility practice is mainly based on the investigation of a single platform, notably Instagram and YouTube. Despite a rise of scholarly attention in the multiplatform context that creators situated in (e.g. Cunningham and Craig, 2019), it remains less known how creators navigate different algorithms across platforms. Platforms are unique in their styles, grammars and logics of architecture and use, or ‘platform vernaculars’ (Gibbs et al., 2015), which in turn mould each platform’s algorithms differently. Much has been written on creators’ perceptions of algorithms on US-based platforms. Yet, creators’ experience of algorithmic visibility across Chinese-language social media and US-based global platforms remains unknown. Noting the uniqueness and difference across Chinese and US-based platforms, this research brings a comparative angle and investigates creators’ algorithmic practice under the cross-platform, cross-border context, which broadens current literature that mainly builds on a single platform context.

This article builds on the growing literature on algorithmic knowledge production and visibility practice, and situates in the context of globalising creator cultures (Cunningham and Craig, 2021). It explores how creators perceive the rules of algorithms across platforms, how they navigate the cross-border context of Chinese and US-based services and what practices they adopt to manipulate algorithms. To understand creators’ multiplatform algorithmic knowledge and strategies, I conceptualise ‘cross-platform sensitivity’ and ‘algorithmic adaptability’ for capturing the cross-platform and cross-border creator practices surrounding algorithms.

Method

To explore creators’ algorithmic knowledge and practices across platforms, I adopted a mixed method of interviews, participant-led walkthroughs and online observation. I conducted in-depth interviews with 16 transnational creators located worldwide (Australia, China, South Korea, United States, Canada, England, France and Germany), varying in ‘wanghong tiers’² (from below bottom-tier to near top-tier³) and content categories. As part of the interview activity, I took a participant-led cross-platform walkthrough approach (Duguay et al., 2024; Light et al., 2018), working alongside 6 creators, stepping through multiple platforms’ interfaces with them and exploring their everyday interactions with algorithms. I also conducted online observation on participants’ content production and platform practices across Chinese and US-based platforms from June 2022 to April 2024, taking screengrabs and observational notes on their content, posts, online interactions and strategies to boost visibility. The mixed methods aim to explore how creators perceive algorithms, how their content production is impacted by platforms’ algorithmic functionalities and what strategies and tactics they adopt for intervening algorithms across platforms.

I recruited 16 creators through personal networks and social media.⁴ Given the massive scale of global creator cultures, the sample size in this research is not representative, but I seek to provide rich qualitative accounts by investigating these 16 transnational creators’ cross-platform algorithmic practices in depth. The most used platforms among creator participants include Bilibili, Weibo, Xiaohongshu, WeChat, Douyin and ByteDance’s ecosystem, TikTok, YouTube and Instagram. Each participant used over five platforms for content distribution, including at least one Chinese platform and one US-based platform. Semi-structured in-depth interviews with 16 creator participants lasted from 1.5 to 4 hours. Since participants were located worldwide, interviews with 11 creators were conducted online; and interviews with five Australia-based creators were conducted in person. As this research is part of an ongoing multiyear ethnographic project, I also followed up with participants after the initial interviews were taken. Among 16 creators, 6 of them agreed to walk through their platform usage with me during the interview. The length of interviews varied as some creators used over 10 platforms, and the walkthrough activity on these platforms took up a longer time.

The participant-led cross-platform walkthroughs build upon the walkthrough method (Light et al., 2018) and comparative walkthroughs (Duguay et al., 2024). The walkthrough method requires the researcher to directly engage with ‘an app’s interface to examine its technological mechanisms and embedded cultural references to understand how it guides users and shapes their experiences’ (Light et al., 2018: 882). Scholars also adapted this approach to compare several apps in the same category (e.g. Duguay et al., 2024). This approach helps study platforms’ social-technical structure, yet it has limitations in accessing certain platform features that are only available to eligible users. It also faces challenges in analysing algorithmic functionalities and in capturing complex data flows (Duguay and Gold-Apel, 2023).

Hence, I adapted the walkthrough analysis as ‘participant-led cross-platform walkthroughs’ that required collaboration between the researcher and participants and situated in a multiplatform environment. In this way, creators could show me their everyday content distribution and workflow, and walk through each platform’s features and interfaces with me while explaining how they understand and navigate different platforms’ algorithmic functionalities. Walkthroughs with participants were also helpful in paying attention to the features that might be overlooked. For instance, during walkthroughs, I was able to observe and learn how participants used the feature of tags differently across platforms, how they found certain tags and why they included them when uploading a video. This method helped to understand their tactical use of platform affordances to negotiate with algorithmic power.

For the context of this research, I address participants via the term ‘creators’, the ‘commercializing and professionalizing native social media users who generate and circulate original content in close interaction and engagement with their communities on the major social media platforms as well as offline’ (Cunningham and Craig, 2021: 1). I am aware of the nuances between ‘creator’, ‘influencer’ and the Chinese term ‘wanghong’. ‘Influencers’ are a type of Internet celebrities ‘who are vocational, sustained, and highly branded social media stars’ (Abidin, 2018: 71). ‘Wanghong’, literally meaning Internet red, has similar connotations to Internet celebrities, influencers and high-profile creators. Wanghong is also polysemic as it can refer to an economic phenomenon, an entire industry, or a specific aesthetic (Craig et al., 2021). In this article, I use ‘creator’ to address my participants, aiming to acknowledge their creative output of original content and videos. ‘Creator’ is an umbrella term for individuals with different levels of online following, and is used across Chinese and US-based platforms.

Among the creators I recruited, 12 of them are China-born Chinese, and four of them are European Caucasians who are often considered ‘waiguoren’ or ‘laowai’ (Chinese vernaculars for foreigners) in the Chinese wanghong industry. I tried to balance out the number of Chinese and non-Chinese participants. Yet, it was relatively harder to recruit non-Chinese creators as participants, as they are considered minorities compared to the majority of Chinese creators in the wanghong industry. Nevertheless, the 16 subjects located worldwide offer invaluable insights for understanding algorithmic experiences in the global influencer industry. I call my participants ‘transnational creators’ because of their migrant, expat and diasporic identities and experiences expressed in their content production. These creators had experiences moving between China and another country, prompting them to adopt both Chinese-language platforms popular in China and US-based platforms that operate in an international market.

In the following section, I will discuss the empirical findings from thematic analysis of the data collected from mixed methods. To answer the question of creators’ knowledge and visibility practices surrounding algorithms across platforms, the analysis will be divided into two parts, one is about cross-platform sensitivity, which reveals the process of acquiring algorithmic knowledge; and the other is algorithmic adaptability, manifesting creators’ practices in navigating different algorithms across platforms.

Cross-platform sensitivity in knowing the algorithms

For creators, knowing the algorithms is about figuring out what type of content and practice are preferred and encouraged by a platform. It requires creators to develop cross-platform sensitivity, that is, the ability to detect the differences and commonalities among platforms regarding their cultures, trends, user demographics, vernaculars and affordances, and to gain a holistic understanding of the dynamics in the multiplatform landscape and engage in the wider network surrounding platform companies.

Sensitivity in the differences and commonalities between platforms

Creator participants learnt the differences between each platform’s algorithms based on their practice and everyday experience on the platforms they used. By learning through doing, participants understood how platforms functioned differently, and gradually found that the rules of visibility game were different on each platform. For example, Melbourne-based Chinese creator Shawn Xiao told me that the more platforms he used, the better he understood the specialties of recommender algorithms across platforms. Shawn shared his insights:

You need to be quite clever in studying these platforms. [. . .] Like WeChat Video Account’s recommendation algorithms prefer short-form and fast-paced videos, [. . .] similar to Douyin. Xiaohongshu prefers informative content, daily life, short and fast videos, and knowledge-sharing. On Bilibili, [. . .], you need an exaggerated title, [. . .] and the video thumbnail is key. [. . .] Bilibili [. . .] may have a laggy algorithm similar to YouTube. On Weibo, if they don’t recommend you via ‘fensi toutiao’ (Fan Headline),⁵ other people can barely see your videos.

Over time, participants like Shawn developed cross-platform sensitivity and learned the specialties of each platform regarding its user bases, platform features and vernaculars (Gibbs et al., 2015). Through trial and error or learning from other creators, they tried to figure out the type of content that was encouraged by a platform. For participants, knowing algorithms was about understanding what the algorithms ‘want’ and making decisions on whether and how to appease them, or gaining ‘practical knowledge of algorithms’ (Cotter, 2022: 16).

Learning what the algorithms ‘want’ also requires creators on Chinese platforms to understand ‘socio-cultural platform pillarisation’ that segments China’s platform society by class, location, affinity or interests (Craig et al., 2021: 86). For example, the Chinese video-sharing service Bilibili is popular among younger generations in China and has a history focusing on animation, comics and games. It is distinct from the microblogging service Weibo, which attracts a broader range of Chinese users consuming current affairs on the platform. When I interviewed Shanghai-based Chinese creator Charles Park in 2022, he reflected on why his Bilibili only had around 16,000 subscribers, compared to his Weibo account with over a million followers. As a creator who produced cinematographic vlogs (video-blogs), Park said:

It’s because I wanted to add a lot of high-end design styles in my content. But on Bilibili, you need to be down-to-earth and show the bizarreness in your content, as their popular videos are mostly pranks and funny sketches.

As he gradually enhanced cross-platform sensitivity to the pillarisation and uniqueness of platforms, Park tried to balance between being fancy and being grounded in his content style to fit each platform’s algorithmic preference.

Cross-platform sensitivity also means perceiving the commonalities and rapid changes across Chinese and US-based platforms. Some participants noticed that short-form vertical videos had swept across different platform economies. For instance, Toronto-based Chinese creator Gloria Gao discovered that short videos were the ‘passcode to web traffic’ (liuliang mima) of US-based platforms. Gloria said:

Isn’t YouTube always known for its long videos? But now they keep pushing YouTube Shorts, so it’s a trend, and Instagram also keeps pushing Instagram Reels, and everyone is competing with TikTok.

Other participants noticed similar changes happened within Chinese wanghong industry. Despite Bilibili not being known for short-form content, some creators noticed its new functions for circulating short videos. To learn what algorithms want, participants were sensitive to the differences and commonalities between platforms, gaining a holistic understanding of the fickle nature of the broader social media entertainment (SME) (Cunningham and Craig, 2019).

Sensitivity in the networks surrounding platforms

To gain knowledge of algorithmic preferences, one way is through practice rooted in personal experience, and another way is through a more social and relational approach to gain insider access to ‘algorithmic expertise’ (Bishop, 2020; Liu et al., 2023). Participants took a collaborative approach with Chinese platform representatives who shared the insider knowledge and logic of ‘liuliang’ or web traffic. To manipulate web traffic allocated to creators’ posts, one way is through the help of human contacts in platform companies, which requires creators to be sensitive to the social networks surrounding platforms.

Most of my participants had contact with platform employees from at least one Chinese platform company. These employees often call themselves ‘yunying’ or operations staff, who work in creator/community operations in a platform company and actively recruit creators from competitor platform companies. Participants gained access to Chinese platform staff either through in-app direct messaging, or ‘platform poaching’, a cross-platform talent-scouting process of recruiting creators from competitor services by providing support and incentives to entice these creators to join a platform (Meng, in press). For example, Shanghai-based British creator Luke Johnston shared with me how he was poached by Tencent, ‘Someone on Bilibili messages me saying, “Hey, we’ve got this new Tencent video platform, do you wanna join this?”’ Tencent employee then added Luke on WeChat and assisted with account registration and one-on-one support. Although Chinese platforms’ creator recruitments were previously assisted by intermediates and public organisations (Yu, 2022: 137), platform companies have placed more emphasis on contacting creators via their own labour forces in recent years.

Platform representatives assisted creators with establishing a profile on their service, introducing web traffic support plans, and managing WeChat groups for creators to learn the latest events (huodong) and campaigns launched by the platform. For instance, Park shared with me that he was in some groups ran by Weibo officials who routinely announced their latest events in the chat. Park explained, ‘They [platform staff] will say, “Under this topic, we will prioritise good quality content and offer web traffic”’. Weibo was not the only platform that shared insider information with creators on WeChat groups. Participants told me that other Chinese platforms, such as Xigua Video, Bilibili, Xiaohongshu and WeChat Video Accounts, also involved in such practice. These platform staff shared insider knowledge on how to obtain web traffic, asking if creators agree to utilise specific video tags or hashtags, make a video under certain topics (huati) and complete certain tasks (renwu) such as posting a video weekly for over 2 months.

Compared to US-based platforms’ positioning creators as minor stakeholders, Chinese platforms have adopted the creator-first strategy to optimise creators for growth (Craig et al., 2021: 171). One way of incubating creators on a Chinese platform is by offering web traffic support’ (liuliang fuchi), the native concept frequently used by Chinese platforms and connote something extra, a bonus or a gratuity generously given to creators (Zou, 2023: 224). The support plan launched by Chinese platforms suggests that platforms can adjust their algorithms to promote certain content, and web traffic is used as rewards to stimulate further content production (Meng, 2021: 320). The partnership between creators and Chinese platforms shows that algorithms on Chinese social media are more ‘knowable’ and involved with intensive human labour, and participants could find a way to crack the code of web traffic through their insider access.

With the help of platform representatives, some participants enjoyed benefits as the ‘reward of visibility’ (Bucher, 2012), though these incentives were often temporary. Fan Headline (fensi toutiao), for example, is how creators gain visibility on Weibo. Five participants mentioned that they could gain access to Fan Headline or other forms of web traffic support for free via their Weibo contacts, or else creators have to purchase to use these visibility services. This type of ‘official gaming tool’ could temporarily modify its algorithms to gain more visibility for a post, showing how Chinese platforms legitimate gaming visibility and commodify the production of automated connections (Meng, 2021: 321). Here, web traffic is distributed based on the amount of money paid to the platform (Zhang et al., 2021: 66), or on the level of insider access that creators have.

As participants developed cross-platform sensitivity over time, some discovered a way to gain insider access across platforms. German creator Volker said:

The recommendation was just to post your content wherever, you start to have some success. Then people and other platforms will recognise you, and they will invite you to theirs.

As a creator on over 10 Chinese platforms, Volker said that platform employees helped foreign-national creators like him get more familiar with Chinese platforms and kick-start an account. But he also said, ‘It’s definitely not the case that if you have somebody giving you fuchi (support), then you definitely will be successful’.

Some creators also questioned the sustainability of the platforms’ support. Two participants shared that they lost connection with their platform contacts, showing the short-lived relationship with platform officials and the precarious nature of the industry. For instance, Melbourne-based Chinese creator Celia Pan shared her experience on Xiaohongshu:

The official [staff] used to interact a lot with creators in the group chat [on WeChat]. They asked creators for their opinions on the topics to promote the following month. [. . .] But now there is no longer that kind of interaction.

Algorithms on Chinese platforms seem more knowable and human labour involved than the ones on US-based platforms. When learning algorithmic knowledge through practice, participants also became more sensitive to social networks surrounding platforms. Some creators gained access to the insider practical knowledge shared by the employees in Chinese platform companies. However, the insider access was temporary, unstable and unevenly distributed. Not all the creators had equal access to these resources. The partnerships between creators and platform officials appear to be mutually beneficial, yet platform–partner relations are inherently asymmetric (Helmond et al., 2019), prompting creators to adopt other strategies to gain online visibility, which I will explain in the below section.

Adaptability in navigating the algorithms

With the algorithmic knowledge at hand, creators choose whether and how to ‘please’ the algorithmic preferences across platforms. In this section, I will describe how creators optimise their content strategies, adapt to different platform features and cross-border ecosystems. Together, these practices and strategies show the importance of algorithmic adaptability in navigating the volatile visibility across platforms and influencer industries.

Adapting content strategies

Cultural production is increasingly contingent on digital platforms (Nieborg and Poell, 2018), which requires content providers to tweak their content to negotiate a platform’s technical affordances, optimise the monetisation of their content and navigate platform governance to engineer cultural and social discoverability, a strategy known as cultural optimisation (Morris et al., 2021: 163). For participants, cultural optimisation not only involves adjusting their content for a single platform, but it also means being agile and adaptable in their content strategies to create optimal conditions for cross-platform visibility.

Some participants distributed the same content across platforms with small tweaks in the video title, thumbnail and description. They tried to figure out the type of content that was unique with personal style and adaptable to different platforms’ algorithmic preferences. As Park mentioned earlier, finding a sweet spot between different content styles. Other participants followed the logic of platform-specific branding (Scolere et al., 2018), tailoring their content to adapt to a particular platform as they believed that could help with the algorithms. For example, Melbourne-based Chinese creator Amy Yang posted short videos under 20 seconds long on TikTok and Instagram Reels. Her longer videos, ranging from 5 to 10 minutes, were shared on Xiaohongshu and Bilibili. Amy said longer videos were about connecting with the audience, whereas short-form content was all about ‘hitting the liuliang distribution mechanisms’. Creators like Amy adapted to different platform environments through a platform-specific approach, shaping their videos to fit into a platform’s cultures, vernaculars, technical affordances and the tastes of ‘imagined audiences’ (Marwick and boyd, 2011).

Cross-platform cultural optimisation also means understanding the cross-cultural tastes of audiences. Luke told me how he adapted his content when distributing videos to Chinese platforms, ‘I put some Chinese memes on Bilibili, rather than YouTube by cutting them out. [. . .] I guess that will make the algorithm better’. Luke used different pop-ups and audio to captivate people’s attention and drive the algorithms across Chinese platforms. I observed that Volker also had Chinese memes in some of the videos shared across Chinese platforms. For these foreign-national creators on Chinese platforms, using memes in their videos was for cultural affinity with Chinese audiences. They used memes and took advantage of the humour and in-joke to generate responses from Chinese viewers and boost engagement rates under the logic of algorithms.

When platforms make constant updates to improve their commercial viability, creators have to adapt their brand subjectivities and practices across platforms and affordances (Arriagada and Ibáñez, 2020). Being adaptable also means swiftly reacting to different forms of platform changes. As some participants noticed the popularity of short videos, they also shifted their focus to making short-form content and hoped that could help boost visibility. Among them, some did witness growth in their social media metrics by posting short videos, showing that adaptability in content strategies was crucial in navigating the changes in algorithmic preferences across platforms.

Adapting platform features

When walking through multiple platform interfaces, participants shared with me how they used platform features and functions to increase online exposure. The practice of tagging was frequently mentioned by participants, as it involved navigating platform features and the surrounding networks with platform employees. Creators adapted their tagging practice to different platforms as the same feature might not be used in the same way. They also held different opinions about the number of hashtags on a platform. For example, Celia said she often added six to ten hashtags in a post shared on Xiaohongshu, as she believed that more hashtags meant more traffic entrances and better relevancy in targeting audiences. Chinese creator Henry Li, however, said that he would not add more than three hashtags in a post, according to the information shared by his platform contact on Weibo. Nevertheless, the practice of tagging revealed these creators’ optimisation tactics to make their content more algorithmically recognisable (Bishop, 2018; Gillespie, 2014).

Some participants performed tactical tagging regardless of the relevancy between their content and the tags. For instance, Henry said he knew that platforms had particular topics (huati) they wanted to promote, but he would not deliberately make a video under a specific topic for the sake of gaining a platform’s reward of traffic. When demonstrating how he distributed a video on Weibo, Henry typed down a hashtag related to music under his post, despite that his chatty video had nothing to do with music. He explained that the hashtag was recommended by his Weibo contacts, ‘They said many celebrities also post photos with that hashtag, but there is no music either. It’s more like an activity promoted by them (Weibo), that’s it’. Participants learnt practical knowledge of traffic allocation through their insider access to platform officials. But they also negotiated and even resisted the algorithmic shaping of their content production by only sticking to what they wanted to create.

There were also circumstances when participants did not add many tags, as they believed the quality of the content was more efficient in driving web traffic. However, with the uncertainty in algorithms, creators still needed optimisation tactics like tagging to potentially boost online visibility, despite that some participants were sceptical about the usefulness of this practice, especially on US-based platforms.

For instance, Amy said, ‘It seems that hashtags are not that important now. [. . .] Even if you didn’t add any hashtags, your video may have the first wave of people ‘like’ it, and your content may be pushed [by the algorithm]’. Amy commented on the practice of tagging, while showing me her frequently-used tags archived in a social media scheduling and management software that she used for content distribution across Instagram, Pinterest and TikTok.

Unlike Chinese platforms that have features and insider access for showing trending hashtags and hot topics, on US-based platforms, participants turned to the help of marketing and search engine optimisation (SEO) tools to find relevant tags. Four participants shared how they adopted SEO tools on YouTube to find relevant tags, but they questioned if tagging could help boost visibility as YouTube’s algorithms evolved all the time. Volker said, ‘I saw a couple of guys who were disregarding all this SEO stuff and tags and everything, but still doing very well from the very beginning. That’s what is a little bit puzzling to me’. Despite that tagging might increase the relevancy of algorithms pushing content to the right audience, some creators cannot tell if tagging would make a significant difference in their online visibility. Tactical tagging reveals creators’ attempts to manipulate algorithms and their adaptability to different platform features. To deal with the rapidly evolving platform features and algorithms, creators must adapt to the continual changes.

Adapting cross-border ecosystems

Many creators cannot avoid feeling anxious about the unpredictable web traffic. The precarious nature of platformised cultural production prompted creators to go beyond the boundary of Chinese wanghong industry and a more international SME built around US-based platforms. The strategy of using multiple platforms is common among creators to mitigate the risks of invisibility due to industry and platform precarities (Duffy et al., 2021). Cross-border platform operations between Chinese and international markets, although not as ubiquitous as cross-platform distribution, have become more popular with the rise of ‘global wanghong’ (Craig et al., 2021).

Cross-border practice showed a level of agency of creators to move around different platform ecosystems. However, it also posed new challenges, as creators’ choices of languages and place of practice might impact their online visibility and the local or global reach of their content (Bidav and Mehta, 2024). Participants who produced videos in Chinese found it difficult to grow on US-based platforms because of linguistic differences. For example, Park stopped managing his YouTube channel and shifted more attention to Chinese platforms, as he felt that YouTube’s algorithm was not effective in recommending Chinese-language content. Park said:

Honestly, I don’t think YouTube is very effective in promoting Chinese-language content. Not everyone can be Li Ziqi, a video posted on YouTube gets millions or tens of millions of views.

What language a creator should use for the content posted on US-based platforms appears to be a cultural choice, yet participants’ experiences show that platforms like YouTube allocate web traffic and visibility through ways deemed inconsistent (Duffy and Meisner, 2023: 300). For many participants, Chinese-language content meant a targeted audience of Chinese-speaking viewers, whereas using English could potentially help to attract a larger pool of English-speaking viewers. Different language choices might lead to different levels of online exposure driven by algorithms. As Melbourne-based Chinese creator Ron suggested, ‘On YouTube, English is, after all, the most widely used language. If you use Chinese, the audience might be much smaller’. The emergence of the globalising creator culture shows the variety of languages in SME (see Cunningham and Craig, 2021). Yet, peripheral creators who produce non-English language content face ‘linguistic precarity’ in their labour practices, as it is more difficult for them to attain a global viewership, pushing them more reliant on their local or regional markets (Bidav and Mehta, 2024: 4).

To overcome the language barrier in cross-border distribution, some participants adopted a multilingual approach when posting on YouTube and Instagram, which shows the effort in adapting to different cultures. Their video titles and descriptions often contained both Chinese and English texts, attempting to attract both Chinese- and English-speaking viewers. These participants’ videos also contained burned-in captions in one or two languages such as Chinese and English.

Despite the efforts of a multilingual approach, it might not be effective in reaching a wider international audience. Having multiple languages could make it confusing to the algorithmic categorisation, as a creator can only pick one language under the ‘chosen language’ feature on YouTube when uploading a video. One example is from Ella and Scott, a Chinese and British pair who constantly shift between speaking Mandarin and English in their videos. When walking through their YouTube interface, Scott explained that he always set the video language in English unless it was only Ella speaking Mandarin. Scott explained why their YouTube videos did not perform well:

Chinese people are much more used to hearing English, and appreciate the English language, whereas English [-speaking] people just don’t care [about] Chinese and they’re like, ‘Ah this is not for me’, switch off.

The action of ‘switching off’ suggests that a video fails to maintain English-speaking viewers’ attention, which in turn restricts the algorithms from pushing the content to more viewers. Some YouTubers were able to reach a global viewership via their innovative practice of translating and subtitling (Lee, 2021). However, for many participants, a multilingual approach not only added to the overall workload but also faced challenges, putting them in a subaltern position for their content getting picked up by the algorithms.

Discussion

In this article, I described how transnational creators developed ‘cross-platform sensitivity’ and ‘algorithmic adaptability’ in navigating the wanghong industry built on Chinese platforms and the international SME built around US-based platforms. Cross-platform sensitivity emphasises the skill to detect the nuanced differences and commonalities between multiple platforms and to engage with a wider network surrounding platform companies to acquire the ‘practical knowledge of algorithms’ (Cotter, 2022). Algorithmic adaptability focuses on the ability to adapt and promptly adjust the self in response to different conditions in a dynamic cross-platform cross-border environment. Here, adaptability shows a form of ‘algorithmic skills’ that creators develop over time to adapt their practices to align with different platforms’ algorithmic preferences (Klawitter and Hargittai, 2018). Cross-platform sensitivity and algorithmic adaptability are both ongoing processes of learning through doing. Sensitivity is grounded in personal experiences and relational practice, which also informs creators’ strategies to react and adapt to different algorithmic preferences. Sensitivity and adaptability become creators’ rules of survival in a situation of structural uncertainty imposed by algorithms across platforms.

In the context of cross-platform transnational creators, the key is being sensitive to algorithmic preferences and adapting to the environments constructed by different platform companies. Adaptation and negotiation are seen as two distinct strategies for navigating platform governance (Poell et al., 2021: 100), but here, I suggest that algorithmic adaptability shows a form of strategic negotiation. Adaptation involves practices in accordance with a platform’s governance frameworks, whereas negotiations refer to the efforts to affect the governance process (Poell et al., 2021: 100–102). ‘Gaming’ the algorithmic systems of platforms, for example, implies negotiation between creators and platforms to resist the algorithmic power (see O’Meara, 2019). Transnational creators’ partnership with Chinese platform representatives revealed a different dynamic of algorithmic gaming. Creators on Chinese platforms could gain traffic support through their insider access, or modify algorithms through the legitimate gaming tools recognised by platforms, temporally bypassing the automated traffic allocation process. Just as creators adapt to algorithmic environment, algorithms react to and learn creators’ online practices, as individuals’ and algorithms’ relationship are always recursive (Bonini and Treré, 2024: 57). Creators’ adaptive strategies of content optimisation, tactics of (dis)engaging platform features and their agency to move across platforms demonstrate their ongoing efforts to negotiate with algorithmic power.

Hence, I suggest that creators’ cross-platform sensitivity and algorithmic adaptability show forms of everyday resistance to survive and cope with, instead of subverting, algorithmic power in the precarious multiplatform environment. Creators’ response to the algorithmic conditions of cultural production is through more subtle, covert and ‘ordinary acts’ (Bonini and Treré, 2024: 26), which aim ‘not directly to overthrow or transform a system of domination but rather to survive’ (Scott, 1987: 301) and to negotiate with better conditions for online visibility. Transnational creators have certain levels of ‘algorithmic agency’ – ‘the reflexive ability of humans to exercise power over the ‘outcome’ of an algorithm’ (Bonini and Treré, 2024: 20). The manifestation of creators’ algorithmic agency includes the mobility in assembling different platforms for optimising conditions for content circulation, temporality intervening algorithms through their partnership with employees at Chinese platform companies. Creators can make decisions on whether or not to ‘please’ a platform’s algorithmic preferences. Yet, as we saw in creator participants’ experiences, not all the algorithmic manipulation actions yielded the desired results. The volatile visibility and changing nature of platforms led to angst among some creators. The level of user agency is subject to the structural power of platforms, and their practices of algorithmic agency are deeply enmeshed with platforms’ institutional changes (Poell et al., 2021). Creators have to play by the rules encoded in each platform’s algorithms as it is the only means of succeeding in the visibility game; however, they might not always play by the spirit of the rules, as they can instrumentalise rules that might conflict with platforms’ values and interests (Cotter, 2019: 908). Ultimately, cross-platform creators have the agency to deprioritise or leave a platform.

Conclusion

This article contributes to the growing literature on algorithmic practice by offering a cross-platform and cross-border perspective to investigate user agency under different platform systems’ algorithmic power. Much of the discussion on algorithmic practice surrounds a single platform context, which neglects the dynamics of the multiplatform environment that social media users are engaging in. By investigating transnational creators’ practices across Chinese and US-based platforms, this research broadens the understanding of users’ algorithmic perception and tactics in an increasingly cross-platform, globalising context. The empirical findings showed that creators learnt cross-platform practical knowledge of algorithms through their practice rooted in personal experience and their relational insider access to Chinese platform representatives. The process of learning what the algorithms ‘want’ across platforms required creators to develop cross-platform sensitivity, or the ability to detect the nuances among platforms regarding their cultures, trends, user demographics, vernaculars and affordances, and to gain a holistic understanding of the dynamics in the multiplatform landscape and engage in the wider network surrounding platform companies. Informed by their algorithmic knowledge and cross-platform sensitivity, creators adapted their content strategies, platform features and cross-border ecosystems to navigate precarity and volatile visibility across Chinese and US-based platforms. Their algorithmic practices revealed the need for algorithmic adaptability in the ever-changing platform environment.

This research has offered a framework of cross-platform sensitivity and algorithmic adaptability to understand algorithmic knowledge and practices in a complex online environment that is multiplatform and cross-border. As this research focuses on perspectives from content creators, the limitation lies in overlooking the roles of multi-channel networks (MCNs), agencies and guilds in assisting creators to navigate algorithms and gain web traffic across platforms. Future studies might investigate how different industry stakeholders and platforms’ governance mechanisms shape creators’ visibility practices and algorithmic resistance.

Footnotes

Acknowledgements

The author would like to thank the three anonymous reviewers for their comments and suggestions. I would also like to thank Bjørn Nansen and Wilfred Yang Wang for their feedback during the writing of this paper.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/ or publication of this article: This work was supported by Melbourne Research Scholarship.

ORCID iD

Ziying Meng

Notes

Author biography

Ziying Meng is a PhD candidate at The University of Melbourne in Australia. Her thesis explores transnational content creators’ cross-platform practices on Chinese and US-based social media services. Her research interests include digital platforms, influencer and creator cultures, Chinese social media, digital ethnography and smart technologies.

References

Abidin

(2018) Internet Celebrity: Understanding Fame Online. Bingley: Emerald Publishing Limited.

Arriagada

Ibáñez

(2020) ‘You need at least one picture daily, if not, you’re dead’: content creators and platform evolution in the social media ecology. Social Media + Society 6: 2056305120944624.

Beer

(2017) The social power of algorithms. Information, Communication & Society 20(1): 1–13.

Bidav

Mehta

(2024) Peripheral creator cultures in India, Ireland, and Turkey. Social Media + Society 10: 20563051241234693.

Bishop

(2018) Anxiety, panic and self-optimization: inequalities and the YouTube algorithm. Convergence 24(1): 69–84.

Bishop

(2019) Managing visibility on YouTube through algorithmic gossip. New Media & Society 21(11–12): 2589–2606.

Bishop

(2020) Algorithmic experts: selling algorithmic lore on YouTube. Social Media + Society 6: 2056305119897323.

Bonini

Treré

(2024) Algorithms of Resistance: The Everyday Fight Against Platform Power. Cambridge, MA: MIT Press.

Bucher

(2012) Want to be on the top? Algorithmic power and the threat of invisibility on Facebook. New Media & Society 14(7): 1164–1180.

10.

Bucher

(2017) The algorithmic imaginary: exploring the ordinary affects of Facebook algorithms. Information, Communication & Society 20(1): 30–44.

11.

Burgess

Green

(2009) The entrepreneurial vlogger: participatory culture beyond the professional-amateur divide. In: Snickars

Vonderau

(eds) The YouTube Reader. Stockholm: National Library of Sweden/Wallflower Press, pp. 89–107.

12.

Chen

Wang

(2023) Serve the traffic: the logic of traffic and its politics on WeChat platform. Social Media + Society 9(3): 20563051231193025.

13.

Cotter

(2019) Playing the visibility game: how digital influencers and algorithms negotiate influence on Instagram. New Media & Society 21(4): 895–913.

14.

Cotter

(2022) Practical knowledge of algorithms: the case of BreadTube. New Media & Society 26(4): 2131–2150.

15.

Craig

Lin

Cunningham

(2021) Wanghong as Social Media Entertainment in China. Cham: Palgrave Macmillan.

16.

Cunningham

Craig

(2019) Social Media Entertainment: The New Intersection of Hollywood and Silicon Valley. New York: NYU Press.

17.

Cunningham

Craig

(2021) Creator Culture: An Introduction to Global Social Media Entertainment. New York: NYU Press.

18.

DeVito

Gergle

Birnholtz

(2017) ‘Algorithms ruin everything’: #RIPTwitter, folk theories, and resistance to algorithmic change in social media. In: Proceedings of the 2017 CHI conference on human factors in computing systems, Denver, CO, 6–11 May, pp. 3163–3174. New York: ACM.

19.

Duffy

Meisner

(2023) Platform governance at the margins: social media creators’ experiences with algorithmic (in)visibility. Media, Culture & Society 45(2): 285–304.

20.

Duffy

Pinch

Sannon

, et al (2021) The nested precarities of creative labor on social media. Social Media + Society 7: 20563051211021368.

21.

Duguay

Gold-Apel

(2023) Stumbling blocks and alternative paths: reconsidering the walkthrough method for analyzing apps. Social Media + Society 9: 20563051231158822.

22.

Duguay

Dietzel

Myles

(2024) The year of the ‘virtual date’: reimagining dating app affordances during the COVID-19 pandemic. New Media & Society 26(3): 1384–1402.

23.

Eslami

Karahalios

Sandvig

, et al (2016) First I ‘like’ it, then I hide it: folk theories of social feeds. In: Proceedings of the 2016 CHI conference on human factors in computing systems, San Jose, CA, 7–12 May, pp. 2371–2382. New York: ACM.

24.

Gibbs

Meese

Arnold

, et al (2015) #Funeral and Instagram: death, social media, and platform vernacular. Information, Communication & Society 18(3): 255–268.

25.

Gillespie

(2014) The relevance of algorithms. In: Gillespie

Boczkowski

Foot

(eds) Media Technologies: Essays on Communication, Materiality, and Society. Cambridge, MA: MIT Press, pp. 167–193.

26.

Glatt

(2022) We’re all told not to put our eggs in one basket: uncertainty, precarity and cross-platform labor in the online video influencer industry. International Journal of Communication 16: 1–19.

27.

Helmond

Nieborg

Van der Vlist

(2019) Facebook’s evolution: development of a platform-as-infrastructure. Internet Histories 3(2): 123–146.

28.

Khamis

Ang

Welling

(2017) Self-branding, ‘micro-celebrity’ and the rise of Social Media Influencers. Celebrity Studies 8(2): 191–208.

29.

Klawitter

Hargittai

(2018) ‘It’s like learning a whole other language’: the role of algorithmic skills in the curation of creative goods. International Journal of Communication 12: 3490–3510.

30.

Lai

(2022) The making of a livestreaming village: algorithmic practices and place-making in North Xiazhu. Chinese Journal of Communication 15(4): 489–511.

31.

Lee

(2021) Translating YouTube vlogs for a global audience: innovative subtitling and community-building. International Journal of Cultural Studies 24(5): 767–790.

32.

Light

Burgess

Duguay

(2018) The walkthrough method: an approach to the study of apps. New Media & Society 20(3): 881–900.

33.

Liu

Tan

CKK

Yang

, et al (2023) Zhibo gonghui: China’s ‘live-streaming guilds’ of manipulation experts. Information, Communication & Society 26(6): 1210–1225.

34.

Marwick

boyd

(2011) I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media & Society 13(1): 114–133.

35.

Meng

(2021) Discursive contestations of algorithms: a case study of recommendation platforms in China. Chinese Journal of Communication 14(3): 313–328.

36.

Meng

(in press) Entrepreneurialism, precarity and self-governance: examining cross-platform creator labour across Chinese and US-based social media economies. In: Fieseler

Goanta

Annabell

, et al (eds) The Hashtag Hustle: Law & Policy Perspectives on Working in the Influencer Economy. Cheltenham: Edward Elgar Publishing, pp. 108–129.

37.

Morris

Prey

Nieborg

(2021) Engineering culture: logics of optimization in music, games, and apps. Review of Communication 21(2): 161–175.

38.

MrBeast Official Channel (2024) Hey, I am MrBeast! I am finally here on Bilibili! Available at: https://www.bilibili.com/video/BV1Nw411E7MW/ (accessed 19 April 2024).

39.

Nieborg

Poell

(2018) The platformization of cultural production: theorizing the contingent cultural commodity. New Media & Society 20(11): 4275–4292.

40.

O’Meara

(2019) Weapons of the chic: Instagram influencer engagement pods as practices of resistance to Instagram platform labor. Social Media + Society 5: 2056305119879671.

41.

Poell

Nieborg

Duffy

(2021) Platforms and Cultural Production. Hoboken, NJ: John Wiley & Sons.

42.

Schellewald

(2022) Theorizing ‘stories about algorithms’ as a mechanism in the formation and maintenance of algorithmic imaginaries. Social Media + Society 8: 20563051221077025.

43.

Scolere

Pruchniewska

Duffy

(2018) Constructing the platform-specific self-brand: the labor of social media promotion. Social Media + Society 4: 2056305118784768.

44.

Scott

(1987) Weapons of the Weak: Everyday Forms of Peasant Resistance. New Haven, CT: Yale University Press.

45.

Siles

Segura-Castillo

Solís

, et al (2020) Folk theories of algorithmic recommendations on Spotify: enacting data assemblages in the global South. Big Data & Society 7: 2053951720923377.

46.

Kaye

(2023) Borderline practices on Douyin/TikTok: content transfer and algorithmic manipulation. Media, Culture & Society 45(8): 1534–1549.

47.

Sun

(2019) Your order, their labor: an exploration of algorithms and laboring on food delivery platforms in China. Chinese Journal of Communication 12(3): 308–323.

48.

Van der Nagel

(2018) ‘Networks that work too well’: intervening in algorithmic connections. Media International Australia 168(1): 81–92.

49.

(2022) Inclusive creativity: Wanghong economy and the Chinese social media entertainment industry. PhD Thesis, Queensland University of Technology, Brisbane, QLD, Australia. Available at: https://eprints.qut.edu.au/235385/ (accessed 7 August 2023).

50.

Zhan

(2021) Li Ziqi breaks YouTube subscribers record for Chinese language channel. Available at: https://www.guinnessworldrecords.com/news/2021/1/li-ziqi-breaks-youtube-subscribers-record-for-chinese-language-channel-647202 (accessed 24 May 2024).

51.

Zhang

Chen

(2021) Traffic media: how algorithmic imaginations and practices change content production. Chinese Journal of Communication 14: 58–74.

52.

Zou

(2023) Provincializing ‘web traffic’: data imaginaries and vernacular construction of liuliang in China. Communication, Culture and Critique 16(4): 221–227.