When I worked in the agency world, three common elements of a measurement framework were: impressions at the top of the funnel, web visits in the middle, and conversions at the bottom.
The rise of AI is breaking this framework.
Traditional search is stronger than ever — AI is not diminishing the importance of traditional search querying for consumers. Google has deftly deployed AI Overviews to head off that risk.
But AI is impacting site visits — search queries are now significantly less likely to result in web visits. From May through July 2025, our team at EDO found a 21% decline in the percentage of consumer search queries that included a link click, a change that roughly coincides with the widescale U.S. launch of Google’s AI Mode in May.
Marketers are understandably concerned that the critical middle of their measurement frameworks is evaporating. But careful measurement of consumer search behavior reveals advertisers have all they need to precisely analyze and optimize for consumer consideration right now.
All it takes is the right search query data and AI models capable of distilling that noisy signal into decision-worthy insights.
For all the talk about “the death of search,” EDO’s unique aggregated behavioral panels actually found an increase in Google search queries during the May-July 2025 period in which click likelihood declined.
Yes, our research shows that traffic to AI websites and apps like ChatGPT has increased by about 3.5x since June 2025. But our findings demonstrate that the overall search pie is growing, meaning that the increase in AI activity is not having a negative impact on Google queries. Instead, consumers are using AI to handle specific kinds of inquiries. Google traditionally has handled simpler brand and product searches and questions, while AI is conditioning consumers to expect more intensive research and answers.
Each query carries the potential to provide a valuable signal of consumer intent — even if the searcher gets all the information they need from an AI overview at the top of the page. That’s because studies have consistently found that when a brand’s share of consumer search in its category increases, market share follows shortly after. No link clicks, no problem.
In the scramble to replace website traffic in their measurement frameworks and analytics, many marketers are making critical errors in the search data they’re using to fill the gap, as well as the analytical tools they’re using to interpret it.
Take paid search. Some advertisers conflate paid search data with the much purer signal of brand consideration and intent that organic search provides. Unfortunately, paid search clicks are just as vulnerable to AI-driven consumer behavior changes as organic search referrals.
Another faulty workaround lies in “verticalized search” data that is solely inclusive of searches that occur on publisher or commerce websites — such as when a consumer searches for a flight on a travel website. In addition to relying on the exact same website traffic that’s in decline, these signals fail to account for queries on Google and AI apps where the overwhelming majority of searches take place.
Simply put, “verticalized search” isn’t real search. EDO has studied its value as a signal of consideration and intent and rejected it as failing our high standards for predictiveness.
Even the best signals of consumer intent can lead advertisers astray if they’re not able to interpret them correctly. That’s why it’s so important to have a measurement model that separates ad-driven consumer behavior from the baseline actions consumers would have taken otherwise. Without the comparison points offered by incrementality modeling, marketers are at grave risk of misattributing regular baseline consumer behavior to the impact of their ads.
While others have spent the past decade chasing the latest hot thing in marketing measurement — from audiences, to attention, to attitudes, to currencies, and beyond — EDO has consistently placed search outcomes at the center of how we measure advertising performance for the industry’s leading brands, agencies, networks, and streamers.
As consumers continue to adopt AI, we’ll continue doing the research, monitoring key trends, and adjusting our modeling when necessary. No matter what changes come down the pike, we’re looking forward to doing what we’ve always done: empowering our clients to identify and optimize the fair value of their media.