by Tim Geenen, Co-Founder and CEO, Rayn
With addressability already under 50% on the open web, and only set to fall further when Google deprecates third-party cookies in 2024, publishers are faced with the growing challenge of ensuring their users are being served relevant content without sacrificing the user experience.
Even before third-party cookies are deprecated, the vast majority of consumers want more control over their data, and many are taking up the option to opt out of data sharing with publishers, decreasing the importance of historical interactions.
For advertisers, there’s an expectation that they can still reach audiences in the same way, if not better than before, as they continue to chase ROI. Publishers are under pressure to provide the data and solutions that make this possible.
This has led to publishers experimenting with deterministic identifiers and probabilistic audiences in search of accountability and measurement. However, especially in Europe, there’s still a lot of uncertainty on which solutions will ultimately work for both advertisers and publishers, at scale.
As an alternative, publishers should instead be prioritising context and moments over previous user actions in the cookieless future. And this approach should be driven by five key pillars: content, context, cohorts, confirmation, and collaboration.
Virtual data
Through the power of synthetic data, machine learning, generative artificial intelligence (AI), and natural language processing (NLP), publishers are able to enhance the digital experiences of their users, and provide advertisers with an avenue to reach consumers without having to worry about jeopardising user privacy.
AI can be used to transform content into context, and be combined with synthetic behavioural data to form audience cohorts. These cohorts are ripe for collaboration, and their effectiveness and accuracy can be confirmed by consumer interactions.
This approach enables personalisation and, in turn, boosts publisher monetisation opportunities, ensuring that media companies, brands, agencies, and commerce platforms can all continue to thrive in the cookieless future.
Importantly, the AI models behind this strategy can only be as good as the data being supplied to them. However, with the collection of customer data becoming more difficult and ethically complicated, there are questions over where this data is going to come from.
This is where synthetic data comes to the fore. Put simply, the generation of artificial data.
Synthetic data emulates real-world data without having any real-world personal information within it. Using first-party data as a starting point, the sensitive, personal elements can be removed and used to establish a correlation between data points. This first-party data could come from a variety of sources, including CRM, behavioural analytics, current trends, voice of the customer, surveys, and more.
The privacy-compliant ‘real’ data can then train generative models to produce synthetic data. And this mimicked data can be used to train the models going forward. This doesn’t just guarantee that consumers’ personal information remains safe, it also promises more sophisticated datasets, and a better understanding of audiences.
Creating a digital twin
An example of how effective the relationship between synthetic data and AI can be seen in digital twins of customers (DToC). But what does that mean?
A digital twin is a way of replicating an element of the real world virtually, and using it to increase efficiency. The concept is something that is often used for industrial machinery and buildings.
In the world of advertising, a DToC is a digital representation of consumers that takes into account both online and physical interactions to predict customer behaviour by simulating their experience with synthetic audiences. What’s learned can then be used to inform best practices for the real world ‘twins’.
The first-party data used helps to create a cohort of digital twins, which are then fed near real-time data through generative AI, so the DToC is constantly being kept up-to-date without ever having to ingest personal information.
With DToC, there’s no longer any need to wait until products are released to know how consumers respond, nor is there to fire out multiple tests of ad campaigns. Instead, the AI predictive models can use the synthetic data to anticipate buying patterns, and how certain audiences are likely to respond to a campaign.
This means publishers and advertisers can be proactive in the ways they are engaging with consumers, and predict how they will respond to any content or ads they’re served before they’re served them. Furthermore, DToC opens the door for the entire customer journey to be mapped out, and opportunities identified in advance, maximising the entire customer experience.
Marketers and publishers are faced with many challenges within the current digital advertising ecosystem, but they should be exploring how synthetic data and DToC can help to overcome some of these obstacles. A focus on the present – through context and moments – is the key to not just continuing to serve relevant content to consumers, but also to ensure their privacy is being respected.
Rayn is a client of Bluestripe Group, publishers of New Digital Age.
