After a year of Google’s AI Overviews, marketers consider tweaking their paid search strategies

Overviews has been credited with cratering search traffic to publisher and brand sites since it was launched in the U.S. on May 14 last year. Clickthrough rates for publisher Mail Online fell over 56% as a result of the Overviews appearing in search results, for example. Meanwhile, 80% of consumers resolve 40% of their searches without going further than the search results page, according to research published in December by consultancy Bain.

Marketers worry that if organic traffic’s being impacted, the search traffic they try to attract with paid media spend might also be in danger.

Are AI Overviews affecting paid search performance?

“While it’s not yet widespread in the U.K., we’re seeing slight drops in impression share and CTR on generic upper-funnel terms. It’s a trend we’re watching closely for both our clients and the industry in general,” said Billy Bojku, head of performance media at VML U.K. (Overviews have been visible in Britain since August 2024).

According to the nine agency search specialists who spoke with Digiday, paid search performance has been mostly sheltered from the impact Overviews and AI search usage has had on organic search traffic.

That’s because the types of keyword searches Overviews appear against are often ones that brands don’t spend much advertising against.

“AIOs are all concentrated in that early funnel informational section. Paid search is primarily in the mid to end funnel,” said Katie Tweedy, director of SEO and content marketing at media agency Collective Measures.

Because the Overviews typically appear for users who are earlier in the purchasing process, the clients that have seen an impact on traffic haven’t necessarily seen a hit to sales conversions, said Charlie Marchant, CEO of agency Exposure Ninja.

“The general trend that we’ve seen with Overviews is a lot of our clients getting less clicks … [but] we haven’t seen a drop in conversions,” Marchant told Digiday.

Are clients doing anything about it?

Even with those caveats, the changes to organic search behavior do have implications for paid search performance. Search agencies are looking to develop alternative strategies in response. “We’re not waiting around,” said Jason Hartley, head of search and shopping at PMG.

In the event that an Overview mentions a brand by name, some marketers fear that paid ads running on the same page might not be the best use of their budget, said Christopher Liversidge, CEO of performance agency QueryClick.

Concerned about “cannibalizing” their own organic traffic, some of QueryClick’s clients have redeployed that budget elsewhere, Liversidge said. Others have decided to increase spending on brand keywords, the better to close down customers already considering a purchase (and head off competitors that might run ads against their brand keyword in the hope of hijacking a customer journey). “That’s when they’re most valuable [and] you don’t want to lose them,” said Liversidge, though he acknowledged that a consensus on the best path hasn’t emerged.

At VML, Bojku said clients were opting for a similar tactic. “There’s been a noticeable shift towards protecting brand terms, with some increasing brand budgets by up to 20% to maintain visibility when AI-generated answers feature competitor links or marketplaces,” he said in an email. “It’s less about spending less, and more about spending smarter as the search engine results page (SERP) evolves.”

At performance agency Gain, head of paid search José de Carvalho said U.S. clients had observed a 15% to 20% decrease in CTR since Overviews were launched last year. In response, many increased the paid spend set against “broad” keywords, de Carvalho said — for example, searches for “summer dresses.” He said such spending had increased 10%, and now accounts for 55% of the £30 million ($39.8 million) in paid search ads the agency runs annually on Google.

Clients are also eyeing up the paid opportunities that exist within zero-click search environments, few though they may be. According to Digitas svp and head of search Corey Kahn, Digitas has consulted with several clients about devoting a portion of their paid spend to Perplexity’s paid ad inventory, though such conversations hadn’t yet advanced to the test stage.

“We’ve had some conversations with some partners around what that looks like,” Kahn said. “The scalability there is still not quite [there].”

AIO will continue to grow and evolve

Such responses might be heard more, as zero-click search becomes more common and as AI Overviews are deployed across more of Google’s search real estate.

So far, Google has only deployed AI Overviews across a limited portion of searches. But that proportion has increased since the tech giant first unveiled the feature; AI Overviews appeared in 13% of all U.S. desktop searches in March, up from 6.5% in January.

Media agency execs anticipate it to increase further as the months roll on, while maintaining a close watch on Google’s AI Mode beta, a trial the company is running that shows users an entirely AI-generated summary in response to search queries rather than the familiar catalog of blue links. There’s also the integration of paid ads into Overviews to consider. Although Google currently places some ads inside Overviews, it doesn’t provide advertisers with the option to buy that space yet. If and when it does, media execs will be paying attention.

“We expect Google to incorporate more paid ads into the AIO in time, which will then have more of an effect on results — which we’re looking out for,” said Laura Williams, PPC manager at performance agency North.

Questions over the future reliability of their paid search investments are likely to mount. In response, some agencies have sought to develop tools or services offering clients a window into the inner workings of Gemini or the LLMs powering ChatGPT and Perplexity’s search engines — and then advise clients on what they should do next.

Exposure Ninja, for example, has developed its staffers’ findings about generative AI into an “AI search practice” it’s been able to offer as an additional service to clients. Performance agency Roast provides analysis of AI search as an additional service layer for clients, using third-party tools like Accuranker.

Queryclick, for another example, has a sister SaaS business called Corvidae, which Liversidge said provides clients with attribution modeling, helping them predict the proportion of traffic or conversions that had come from an Overview or AI search. Similarly, Jellyfish offers a proprietary SaaS tool that provides clients an estimate of their “share of model” within a given LLM.

Such services and tools hint at the future of both paid search and SEO, practices which are already shifting in response to AI. Zero-click search usage, Digitas’ Kahn said, demands “a total shift from how many brands have built their entire organization over the last 20 years.”