Artificial intelligence (AI) continues to be one of – if not the – biggest talking points in digital advertising, particularly as advertisers and publishers continue clawing to find solutions as third-party signals gradually fade away.
But where does AI fit into the future of advertising, is it an ally, and how should businesses approach the implementation of the technology?
Searching for the right answers
As with any emerging technology within a space, there are questions over if it’s really going to do more good than bad for the industry. Any time we head over to our favourite search engines, it seems almost impossible to avoid some form of interaction with the platform’s AI.
Publishers, already having to cope with how changes around identity and targeting are impacting their revenue streams, are seeing AI-driven search taking away their traffic and need to find ways to navigate that.
In this case, the solution lies in fighting fire with fire, according to Debra Fleenor, Founder & President of Adapex.
“We have these two perfect storms brewing out there,” she said, speaking at Prebid Ascent London 2025. “One is the demise of the third-party cookie, and the other is AI changing and shifting search patterns. The solution is the same. All paths lead to Rome. Advertisers and publishers, you have to build a better, stronger, closer, and more valuable relationship with your users.
“So, it’s a foe, because it took away some of your traffic. But it’s a friend too, because it’s AI, ML (machine learning), and LLMs (large language models) that’s going to help you build the tools that are going to help you build a better relationship with your users that keeps them coming back and ingesting your content.”
Arjun Arora, Head of Global Partnerships, Ad-Tech at Sigma Software, doesn’t see the challenge as being one that publishers will struggle to overcome, because they’re “a tough breed [and] have had to evolve time and time again.”
He believes that the “actual value publishers can tap into” will come through them continuing to leverage their first-party data, growing the amount they have at their disposal by “building communities, building subscription models, trying to get the traffic away from just the search engines, trying to get it through email marketing, and getting more audience on the website.”
Ready to embrace the future
Publishers are all at different stages of readiness when it comes to their approaches to the post-third-party cookie world and their AI-related strategies, as highlighted by the panel’s moderator Ronny Linder, CPO, Partner at Relevant Digital.
Bigger publishers, noticing that search is no longer delivering for them in the way it used to, are exploring deals that will help them overcome the issue. Although, these deals aren’t always available to smaller publishers.
“If it’s small publishers, they will have to find or form some consortia. Or they might have to stay close to top AIs – right now, it’s ChatGPT, Google, or Microsoft – to see what their strategies are. But do not neglect the smaller ones as well, because every day you see a story that a new bot has come and it’s doing some wonders,” said Mohsin Pervez, Senior Director, Solutions Engineering & Customer Operations at PubMatic. “All these bots are trying to do different things, but staying close to say Google, OpenAI, or Microsoft is definitely going to help you there.”
However, when it comes to looking at these various technologies, there are a number of things that businesses need to consider, including security, understanding, and skills.
When running a review of the LLMs used by teams internally, Magnite made sure that the right security was in place to “make sure that data we’re inputting into the LLM isn’t being used to train those models, and could potentially resurface somewhere at a later date,” according to Matthew Rooke, Revenue Solutions Lead, DV + EMEA at Magnite. “Working with a secure provider that made sure our data and our clients’ data was protected was really important.”
The second measure to take is the rollout of a companywide AI policy, because “there are some things these tools are to be used for, and others they aren’t to be used for,” Rooke added. This ties in closely with the need for training, “so that they know the positives of these tools, but also things to be wary of.”
Finally, there should be a feedback process to help ensure the technology is delivering value. “Not just rolling it out, but making sure people are using it and it’s doing what they want.”
*Magnite is a client of Bluestripe Communications, owned by Bluestripe Group, publisher of NDA
