Friend or Foe? AI’s Transformative Impact on Websites and Contextual Targeting

By Geoffroy Martin, CEO at Ogury

As advertisers look forward towards a post-cookie landscape, many ad tech companies see contextual targeting as an easy-to-deploy solution to help fix the addressability problem. However, significant challenges are casting doubt on the sustainability of this model, particularly the rise of artificial intelligence (AI).

Towards an Increasing Scarcity of Web Pages…

Conversations around AI in digital publishing have, so far, mostly revolved around AI-generated content. But as highlighted by the scandal involving Sports Illustrated’s publisher – which saw revelations that some of the publisher’s articles had been written by fake, AI-generated contributors – its usage is being met with criticism and resistance.

In reality, the use of artificial intelligence links more directly to the way we consume online content. If an AI tool can provide an answer to any question, about any topic, why would we need to visit a dedicated website?

Soon, you won’t need to be a ChatGPT enthusiast to experience this change. Google’s AI Overviews (formerly Search Generative Experience) has just been deployed in the US, leveraging Google’s AI to respond directly to search queries by generating text or multimedia answers, much like a chat. This represents a significant threat to websites’ organic traffic, and consequently, their survival.

Despite this looming issue, the loss of the cookie is the main concern for publishers currently – even major news outlets. On cookieless browsers like Safari, publishers have experienced cost per mille (CPM) drops of up to 60% for several years now. Initial tests on Chrome show losses of at least 30%. Concerns raised by the UK Competition and Markets Authority over Google’s Privacy Sandbox solution have caused further issues, ultimately leading to the tech company once again postponing its cookieless roll-out to all Chrome users. In light of these threats, some publishers are considering drastic measures, such as reducing their sites’ portfolio or downsizing their teams. 

…and a Surge in Prices

The disappearance of cookies is also on the lips of every advertiser. While agencies, and even more so brands, have long avoided the issue—with some continuing to do so thanks to the delay granted by Google—many of them have recognized the urgent need for cookieless advertising. In this context, nothing seems simpler, in appearance at least, than… contextual targeting.

In some ways, contextual marks the grand return of marketing ‘common sense’ – targeting audiences by placing ads in contextually relevant locations. Any adtech company today is capable of leveraging contextual and semantic targeting with, at best, an overlay of machine learning.

A typical example of such targeting would be: you are browsing the cycling section of ESPN, and you’re shown an ad for an electric bike. Admittedly, such a targeting method is quite basic and limited – people have multiple interests and may not necessarily visit sites related to all of them.

The current craze for contextual targeting is fueled by the fact that cookies haven’t yet disappeared from Chrome. When we enter the post-cookie world (by early 2025), every brand will want to purchase ad space on the same sites. This means over-saturated inventories, higher CPMs, decreased performance, and ultimately, lower ROI.

The Age of Zero-Party Data

If the amount of contextually relevant sites for a campaign is limited, the number of available sites, and therefore pages, could decrease. As the 2025 deadline approaches, brands and agencies need truly sustainable solutions if they are to see long term success.

One method that is emerging as a front runner is zero-party data. Voluntarily shared by consumers in a fully transparent and consented exchange of information, zero-party data is traditionally difficult and costly to attain. It provides richer and more precise insights into consumer habits than third-party data, including preference data, purchase intentions, personal context, or how they wish to be recognized by brands. Combining these insights with traditional targeting data—such as semantic details or bid request data – enables brands to unlock previously unavailable information.

The ability to understand and qualify site audiences beyond their browsing context is all the more important as the open web gains momentum. Over the past several years, it has been capturing market share in walled gardens, as advertisers recognize the need to diversify their data sources. Moreover, the potential disappearance of pages and sites should remind us of the existence of a whole other environment, in which relevant advertising can be delivered without relying on contextual targeting: applications. Indeed, unaffected by MFAs or the loss of the cookie, in-app provides the perfect place to engage audiences and develop a zero-party data solution.

The promise of contextual targeting as an easy-to-integrate cookieless solution may be appealing to advertisers, but under scrutiny, it remains limited and not sustainable in a world without identifiers. Instead of short-term fixes, advertisers need to invest in solutions that go beyond simple semantic analysis to qualify a site’s audience. As the cookie deadline inches closer and the rise of AI threatens the open web as we know it, marketers must begin to test solutions that will allow them to continue to drive results in the future.

Tags: AI