Digital platforms claim to offer more transparency than ever through public ad libraries and political-ad reports. However, new Australian research suggests these tools provide only a narrow glimpse of the advertising users actually receive – and fail to show how platforms algorithmically “tune” ads in real time.
A 2024 study by researchers from Monash University and the Australian Ad Observatory argues that modern ad delivery is no longer just about targeting specific demographics. Instead, platforms now shape sequences of ads as users scroll, using real-time behavioural signals to continuously adjust what appears on screen.
The researchers call this process “tuned advertising”, and their data shows it is already widespread across major social media platforms.
Advertising becomes a dynamic “flow”
According to the study, tuned advertising relies on constant optimisation. Algorithms decide not only which ads to show, but when, how often, and in what order. Delivery shifts based on everything from a user’s scrolling speed to their recent clicks.
This means people don’t simply receive isolated ads – they experience personalised streams of ads over time.
To investigate that stream, the research team worked with more than 1,900 volunteer participants. Through a data-donation browser plug-in, the project collected 737,418 ad impressions, representing 328,107 unique ads across platforms.
By analysing these impressions, the team reconstructed “tuned sequences” – the evolving series of ads a user sees as they move through a platform. Rather than random placements, the sequences often reflected algorithmic adjustments based on immediate behaviour.
Researchers say this dynamic structure is largely invisible to the public.
Why ad libraries don’t capture tuned advertising
Major platforms promote their transparency tools as evidence of accountability. Meta and Google both maintain public ad libraries that list political ads, and in some cases commercial ads, along with limited spending or demographic information.
But the Australian researchers found that these libraries fall short in several key ways:
• Ads disappear quickly.
Many tools only list ads while they are running. Once the campaign ends, the ad – and its metadata – vanish.
• Key delivery data is missing.
Most libraries do not reveal how many people saw an ad, how often it was displayed, or how delivery changed over time.
• Libraries show isolated ads.
The tools treat ads as static creative assets. They do not reveal how an ad fits into a longer sequence of related messages delivered to a user.
• Searchability is limited.
Unless a researcher knows the exact advertiser name, some ads are nearly impossible to find.
Researchers argue that without information about ad exposure, not just ad content, it is impossible to understand how platform algorithms shape advertising at scale.
The case for “Ad Observatories”
To fill the gaps, the researchers propose new forms of independent oversight – what they call “ad observatories.” These would combine browser-based data donations, voluntary user participation, and academic analysis to track ads as they actually appear on screens.
The Australian Ad Observatory trial shows what this model can uncover. By capturing ads directly from participants’ feeds, researchers were able to map not only what ads appeared, but in what order and under what conditions.
They argue this approach provides a more accurate representation of how platforms distribute ads – especially from industries such as gambling, alcohol, and political campaigns, where delivery patterns can raise social concerns.
Regulatory stakes are rising
The findings come as governments around the world debate tougher requirements for platform transparency. Many current proposals focus on expanding ad libraries or strengthening political-ad disclosures.
But researchers warn that these measures still rely on outdated assumptions about how digital advertising works.
“Tuned advertising doesn’t show up in a library,” the study notes. “It happens in the flow of use — the part we can’t currently see.”
Without new forms of observability, regulators may be relying on partial or misleading information when assessing issues such as political messaging, harmful-product advertising, targeted misinformation, and algorithmic bias.
A hidden source of power
The study highlights a growing concern in digital-policy circles: that platform advertising has become both more personalised and more opaque.
When algorithms determine which users see which messages — and how often — platforms gain influence over attention, behaviour, and public discourse. And because tuned advertising operates through behavioural signals, not explicit demographic targeting, the effects can be unevenly distributed across different groups.
For researchers, journalists, and policymakers, the absence of durable delivery data makes it difficult to analyse these patterns or assess potential harms.
Experts say meaningful accountability will require platforms to provide:
- Durable, public ad archives that store ads long after campaigns end
- Delivery metadata, including reach, frequency, audience characteristics, and time-based delivery patterns
- Tools that show ad sequences, not only standalone creatives
- Independent auditing access, allowing researchers to examine how algorithms make delivery decisions
Some of these measures would require legislative mandates, while others could be implemented voluntarily. So far, platforms have offered only incremental changes.
A transparency gap that’s only getting bigger
As digital platforms become central to political communication, marketing, and public life, tuned advertising raises new questions about visibility and power.
The Australian study indicates that while platforms promote their transparency tools, the most influential aspects of ad delivery remain out of sight.
For now, researchers say, the public can see what ads exist – but not how they are used. And without access to that delivery data, it remains difficult to answer a simple question:
Are we truly seeing what platforms are doing?