Australian Ad Observatory: Investigating mobile and dynamic advertising via computational and participatory approaches

Advertising remains the dominant model for supporting commercial media platforms, and continues to pioneer strategies of data driven customisation and targeting. Advertisers are at the forefront of experimenting with automated digital media across recommendation, targeting, synthetic and augmented content, logistics and retail. Revenue from advertising funds the digital media platforms that in turn invest in engineering automated models that curate, augment and synthesise our media experience. 

Phase one of the Ad Observatory pioneered a way to observe the targeting of social media advertising across populations of users. We generated the largest known collection of targeted ads that people encounter on Facebook in Australia – 328,107 unique ads from 1909 participants –and built world-first research infrastructure that involved citizens in doing so. The project led to significant findings and impact, and new ways of approaching the study of automated advertising, not only in terms of individually targeted, discrete ads, but as ongoing sequences of ads that are ‘tuned’ to work in tandem with people’s identities and daily rhythms.

Responding to significant recent and ongoing developments in automated advertising (including Generative AI), the Phase 2 Ad Observatory will develop approaches for studying contemporary media and information environments, where there are no longer either shared flows of content, nor stable texts.

It will conduct participatory research with diverse groups of Australians to provide visibility into the targeting of harmful products to particular groups. It will explore experiences of advertising and understand its cultural impact, combining citizen science with data collection.

This project will significantly advance our conceptual understanding of automated advertising, playing a crucial role in documenting the emergence of this new form of advertising and enabling industry, civil society and government to respond to the challenges it will create to observability and accountability.

Objectives:

• Develop research infrastructure and methods to observe dark, ephemeral, and automatically generated ad content and sequences, using a prototype Mobile Ad Observatory Toolkit to collect digital ads across platforms; 
• Conduct participatory research with diverse groups of Australians to explore their experiences of advertising and understand its cultural impact, combining citizen science with data collection; 
• Develop tools for automatically identifying defined categories of advertising (such as political ads and ads for harmful products and services); 
• Provide visibility into the targeting of political advertising during the next Federal election in Australia by recruiting undecided voters in marginal and swinging electorates; 
• Develop a detailed account of competing advertising explanatory models: those offered by the platforms, those offered by users, and those informed from our observations; 
• Conceptualise and develop models for simulating automated advertising at individual and cultural levels, creating tools for users to compare ad sequences and question the automated processes behind them;  
• Examine implications of automated advertising, including the use of generative AI, for current and contemplated platform governance policies, legal tools, and regulatory frameworks; and
• Collaborate with partners to build a public interest network focused on digital advertising, promoting observability, advocacy, and influencing the development of responsible advertising practices.