Selected Papers of #AoIR2022: The 23rd Annual Conference of the Association of Internet Researchers Dublin, Ireland / 2-5 Nov 2022 Suggested Citation (APA): Carah, N., Brown, M.G., Dobson, A., Robards, B.. (2022, November). The Algorithmic Flow of Harmful Industries Advertising on Social Media Platforms. Paper presented at AoIR 2022: The 23rd Annual Conference of the Association of Internet Researchers. Dublin, Ireland: AoIR. Retrieved from http://spir.aoir.org. THE ALGORITHMIC FLOW OF HARMFUL INDUSRIES ADVERTISING ON SOCIAL MEDIA PLATFORMS Nicholas Carah The University of Queensland Maria-Gemma Brown The University of Queensland Amy Dobson Curtin University Brady Robards Monash University Background Seeing an algorithmically-curated flow of advertising as we scroll through our social media feeds is now an ordinary part of our digital lives (Bucher 2017, Carmi 2020, Trott et al. 2021). While researchers have examined both the targeted advertising and promotional brand cultures of social media platforms (Banet-Weiser 2012, Bishop 2018, McGuigan 2019), in this paper we argue for the need to investigate the customised flow of advertising itself. Shifting beyond the targeting of individual ads, focusing on the immersive flow of advertising on social media enables us to consider how advertising constructs and reproduces subject positions along lines of gender, class, sexuality and ethnicity in an era of algorithmic customisation. Advertising shapes our larger public culture but the typical experience of advertising is now confined to our private and algorithmically customised social media feeds. Despite advertisers and digital platforms having detailed portraits of users, digital advertising is difficult to monitor and study. The vast majority of advertising on social media platforms is ‘dark’ -- that is, only visible to the users being targeted. Ads on digital platforms are not published or archived. While there are some transparency tools, like Facebook’s Ad Library, these provide only a partial record of advertising activity. We investigate the customised flow of advertising on social media by analysing collections of ads sent to us by individual users, exploring the patterns that emerge and whether they reflect particular ‘personas’. Our study In this project, partnered with VicHealth, we used a participatory digital method to work with 204 young Australians aged 18 to 25 to collect 5169 examples of alcohol, gambling and fast food advertising from their social media feeds. We focus on alcohol, fast food and gambling advertisers as harmful industries that are extensive and innovative users of social media. They leverage the participatory cultures and algorithmic power of digital platforms like Instagram, Facebook and TikTok to engage with and target young people (Atkinson et al. 2017, Niland et al. 2017). Over a two week period participants took screenshots of ads and sent them to us via SMS. This process was interspersed with SMS chat with the researchers, describing and co-analysing the ads they were seeing and sending. Participants also completed two surveys on their views and perceptions of harmful industries marketing on social media before and after the collection of screenshots, and were invited to download and share with us the ‘ad interests’ data Facebook creates about them in its advertising model. Findings and Analysis In this paper we analyse the collections of advertisements each participant sent us. Advertisers and digital platforms build and optimise custom audiences that target consumers based not only on demographic characteristics like age, gender and location but also on the interests and affinities they share in common with other users. Meta’s ‘lookalike’ audiences, for instance, assemble audiences for advertisers based on shared affinities. Rather than specify particular demographic or behavioural targeting criteria, the advertiser instead iteratively ‘tunes’ an audience based on who engages with their advertising and purchases their products or services. These models become more sophisticated over time as they monitor and learn which consumers engage with what kinds of ads. The ads we see in our feeds do not just reflect the brands, services and products we consume, but also our subject positions, identities, tastes and cultural worlds. To explore the algorithmic flow of advertising on digital platforms we approach the collections of ads each individual participant collected and sent us as products of these ‘lookalike’ audience building tools. We looked for patterns of symbols and themes within each participant’s collection and coded them inductively. We brought together collections that had shared visual and thematic similarities, and then associated them with information about participants including their age, gender and consumption of alcohol, fast food and gambling products. The patterns across participants’ collections illustrate how social media platforms’ advertising models ‘learn’ to reflect and reproduce the identities and subject positions of participants. For instance, many young women in our study had collections of ads for alcohol and bars that featured the colour pink, botanicals and flowers, picnic rugs and soft furnishings. The collections of images sent to us by young participants help to illuminate how the advertising models of social media don’t just target us with advertising, they immerse us in curated algorithmic flows of consumer and brand culture that project markedly classed, gendered and racialised sets of cultural tastes, aesthetics, sensibilities and ‘vibes’. Conclusions and implications Our novel method and analysis demonstrates how individual ad campaigns accumulate into a larger cultural imaginary in our algorithmically-curated social media feeds. The collections of ads we see on social media are an important object of study because they reveal not just the symbolic content and targeting patterns of particular ads, but also because they illustrate how advertising on social media algorithmically-curates an immersive cultural experience. Our study illustrates how social media continues the larger social role advertising plays in the construction and maintenance of consumer subjectivities. The advertising model reflects the creativity and fluidity of identity and expression that we associate with the participatory cultures of social media. We need to conceptualise advertising on social media not only using concepts of ‘targeting’ that imply the precise identification of our characteristics, but instead as a complex feedback loop between the refinement of ad content and themes, the data-driven optimization of audiences, and our reflexive and fluid identities, interests and aesthetic sensibilities. We conclude our paper by arguing that researchers need to focus not just on the analysis of ads, the targeting of ads, and the participatory promotional and brand cultures of social media. We also need to conceptualise and study the algorithmic flow of advertising on social media because this is the basis of our everyday engagement with advertising. The paradox of our present moment is that while we all share the experience of advertising-laden algorithmic feeds, each feed is customised and private. We need to develop novel methods and conceptual frameworks for investigating advertising as a cultural technology that doesn’t just target us as individuals but algorithmically constructs and curates the larger cultural narratives and imaginaries we inhabit and draw on to fashion our identities. References Atkinson, A. M., Ross-Houle, K. M., Begley, E., & Sumnall, H. (2017). An exploration of alcohol advertising on social networking sites: an analysis of content, interactions and young people’s perspectives. Addiction Research & Theory, 25(2), 91-102. Banet-Weiser, S. (2012). Authentic™. New York University Press. Bishop, S. (2018). Anxiety, panic and self-optimization: Inequalities and the YouTube algorithm. Convergence, 24(1), 69-84. Bucher, T. (2017). The algorithmic imaginary: exploring the ordinary affects of Facebook algorithms. Information, communication & society, 20(1), 30-44. Carmi, E. (2020). Rhythmedia: A study of Facebook immune system. Theory, Culture & Society, 37(5), 119-138. McGuigan, L. (2019). Automating the audience commodity: The unacknowledged ancestry of programmatic advertising. New Media & Society, 21(11-12), 2366-2385. Niland, P., McCreanor, T., Lyons, A. C., & Griffin, C. (2017). Alcohol marketing on social media: young adults engage with alcohol marketing on facebook. Addiction Research & Theory, 25(4), 273-284. Trott, V., Li, N., Fordyce, R., & Andrejevic, M. (2021). Shedding light on ‘dark’ ads. Continuum, 35(5), 761-774.