By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

Executive Q&A: AI and the Future of the Cookie-Free Web

Xaxis global product lead Jacob Grabczewski talks to Upside about the role of AI and cookies as well as how AI can help marketers navigate the changes ahead.

Since the birth of the web, cookies have been an integral part of the user experience. However, that is about to change, as browser developers take steps to limit the reach of what’s known as third-party cookies -- the tools site owners use to track user activity across the web. Jacob Grabczewski, Xaxis global product lead for Copilot (the company’s proprietary AI platform), talks to Upside about a cookie-free web experience and how marketers are planning to respond.

For Further Reading:

How Ethical AI Is Redefining Data Strategy

From Privacy to AI Implementations, What to Expect This Year

Deep Trouble for Deep Learning: Hidden Technical Debt

Upside: When it comes to data collection, how are cookies being used today?

Jacob Grabczewski: Cookies enable websites to store and retrieve information locally in a browser. For example, cookies can store user language preferences without referencing previous login information or maintaining a formal profile. Any direct exchange of information with a specific site can use this local data, which is known as a first-party cookie. These uses are not going away.

However, there are also what’s known as third-party cookies, which offer a way of sharing information across different sites. This can include multiple properties owned by the same parent company, or partners or vendors who are involved in the experience on a website, including advertising. One such use case is to link passive behavior, such as an ad rendering in the browser while viewing Site A, to actions taken on Site B, such as an e-commerce purchase.

Because this data can be shared across sites and systems, patterns can emerge with enough observations of the content views, ad interactions, and purchases to make educated guesses about the person using the browser. For example, if a browser recognizes a pattern of viewing content about destinations far from its location, combined with reviews of content with tips about credit card rewards, someone with access to that data can deduce that the user of the browser is more likely to be in the market for travel products than either of those visits would suggest independently.

What changes are ahead when it comes to cookies?

Several browser companies have already been blocking third-party cookies, preventing any tracking across sites. Google Chrome, which represents the vast majority of web browsing activity in most markets, has announced that it also expects to block third-party cookies by 2023. Impacts of this will be felt most strongly in the advertising ecosystem. For example, linking ad views and interactions to activity on an e-commerce or brand site will not be possible with the same tools as before, and not with the same specificity. Additionally, it will be more difficult to determine how many times a browser has rendered an ad on different sites, preventing holistic reach and frequency calculations.

How will this change what kind of data is collected and how it’s collected?

The same information will be collected as before: views to a site, displays of an ad, purchases, and so on. What’s changing is that the ability to link this activity to the same browser will not be possible, disrupting calculations about ad effectiveness and limiting the pattern recognition that can take place to make inferences that inform marketing strategies.

What alternative measures or means are being considered or developed?

Several alternatives are being worked on. Browsers are building and testing alternative measurements that limit the amount of data and level of detail made available to third-party cookies by sharing noisy aggregate data or using artificial intelligence (AI) to predict audience interest without sharing underlying activity. Industry initiatives also include creating new standards around privacy-safe sharing of information, requiring explicit consent from individuals who sign in to access content. Companies in different parts of the advertising ecosystem are also developing AI tools to make probabilistic estimates for campaign effectiveness.

What role is AI playing in these alternatives?

AI is at the heart of several alternatives, including the training of predictive models on data from users who have given consent and extrapolating trends where this data is not available, as well as training models on aggregate and anonymized data. It also supports the predictive intelligence behind automated test-and-learn systems that take an active role in measuring outcomes.

What is Copilot doing with regard to such data collection?

Copilot is a Xaxis-built AI technology that optimizes digital media investments towards real business outcomes. It makes thousands of data-driven decisions without collecting data directly by integrating with the platforms our clients use to buy media, ingesting aggregate and event-level data (where available) to train predictive models and create learning loops that optimize buying within and across those platforms on behalf of a client. As measurement shifts toward the probabilistic and aggregate, Copilot is ready to work with the different methodologies each platform implements to continue maximizing effectiveness of our clients’ media buying.

Do you see any obstacles to the future of AI adoption in combating the impact of cookie deprecation?

On the contrary, I expect cookie deprecation to increase the adoption of AI in advertising. When granular data is readily shared, one can draw conclusions with simple math. If organizations can’t rely on having all the data, they will need robust, scalable techniques to make inferences that support decisions with confidence. AI is all about managing uncertainty to confidently arrive at conclusions with a rigor that would otherwise be difficult or impossible to achieve.

[Editor’s note: Jacob Grabczewski leads the product teams behind Copilot, Xaxis’ AI-powered optimization technology. With Copilot, Xaxis offers brands a consistent and powerful optimization layer across their chosen DSPs to deliver on business outcomes. Jacob has almost two decades of experience in digital advertising and technology and has previously held roles in analytics and operations.]



TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.