Interviews, insight & analysis on digital media & marketing

Embracing AI and innovation in a post-cookie world

By Ian Liddicoat, CTO of Adludio 

For many professionals in the advertising and media space, the impending deprecation of cookies feels like a process that has been dragging on for years, and to a certain extent, this sentiment is accurate. Google has now scheduled this transition for early 2025, but given previous experiences, it may be a can that gets kicked further down the road once again. 

For advertisers, the extended preparation period, which includes actions such as the removal of a 1% sample of cookies for Chrome users, has provided ample time to prepare and decide on the best approaches to maintain personalisation in a highly privacy-compliant manner.

Leveraging first-party data and consent 

For many, this has meant ensuring that first-party data for their consumers is accurate and that consent has been sought and stored at an individual level. This is the most obvious response and includes web registrations, email campaigns to register consent, and similar measures. However, for a good number o0f advertisers, this was already happening and, to some extent, will not fully mitigate the absence of cookies in linking behaviour across devices to deliver personalised experiences.

Another common approach has been to contextualise or hyper-contextualise campaigns and utilise a range of external factors, such as detailed location, retail events or weather conditions at the point of delivery, while tracking performance against hold out samples. There is no doubt that these approaches have merit and will continue to form part of the targeting mix in a post-cookie environment.

For some advertisers with rich transactional data, there are further opportunities to rethink their approach to audience segmentation. 

Instead of relying on static segments driven by lifetime value, demographics, and lifestyle data, there are now opportunities to apply machine learning techniques and enable fully dynamic segmentation across the lifetime of a customer relationship without the need to retain cookies. This approach can also include factors such as hashed email addresses and browser fingerprinting while still remaining privacy compliant. Machine learning can also be applied to dynamic attribution across sessions, where content can be ad-served and optimised at or close to real-time.

Innovating creative content and leveraging AI 

Another area in which we will witness significant innovation is the creative content itself, including the use of highly gamified interactions. Again, computer vision and machine learning can be used to optimise the structure of an ad unit for the combination of interactions and creative objects that are most likely to engage a consumer or audience segment. 

These techniques are a significant step beyond traditional dynamic creative methods and open up major opportunities for advertisers and publishers to share unique insights, in both directions, about how the creative structure of a given ad variant performs with a certain publisher and across specific inventory. When combined with brand safety and within brand guidelines, these techniques are particularly powerful and will enable advertisers to steadily increase their creative intelligence repository. This library of attributes should ultimately form part of the brand valuation itself. For each brand, this data uniquely identifies which combinations of creative objects, text, sound, and interaction types will excite and engage consumers across products, markets, and media channels.

Redefining campaign performance measurement 

It is also clear that advertisers have to rethink how campaign performance is measured and focus on robust measures of attention and how these do or do not influence incremental sales. The Interactive Advertising Bureau (IAB) is making great strides in defining accepted measures of attention that will form a key part of the post-cookie advertising environment. Advertisers need to clearly define attention measurement criteria and the relationship between these, creative intelligence, and dynamic segmentation.

Clearly, the dominant theme here is the application of AI. There are simply too many data points for humans and now for traditional statistical models to handle. This is particularly the case for the application of machine learning and computer vision methods to significantly enhance the relevance of the content presented to a consumer. While cookie deprecation is not the only driver of this innovation, one could argue that the successful adoption of such techniques in an integrated manner delivers on the promise of digital marketing that has always been present.