SEO is still worth paying for, but the part that pays has moved. The traffic AI is taking is mostly the informational kind: the “what is” and “how does” searches that filled company blogs for a decade. The searches people make when they are looking for a specific company, or getting ready to buy, have held up so far. Budgets that still chase the first kind are paying for a channel Google is closing.
The clicks AI took were the research clicks
The clearest evidence comes from a randomized field experiment by researchers Saharsh Agarwal and Ananya Sen, reported by [Search Engine Journal]( https://www.searchenginejournal.com/google-ai-overviews-study-finds-lost-clicks-weren-lower-quality/581298/) in July. Showing Google’s AI Overviews cut organic clicks by 39.8 percent. The summaries appeared on about 41 percent of queries.
The detail most teams miss is where they appeared. AI Overviews showed up on 53 percent of informational queries, 15 percent of searches for a specific company or site and 6 percent of transactional ones. The click losses sat almost entirely in the informational group. Company-name and buying searches showed no measurable change, though the authors had smaller samples there.
The study also tested Google’s argument that the lost clicks were low-value “bounce clicks.” They were not. Roughly 18 percent of visits ended within 10 seconds whether or not an AI Overview was shown, so the visitors who disappeared were as engaged as the ones who stayed. One honest caveat: the paper is a working draft and has not finished peer review.
A second experiment, from researchers at the University of Pennsylvania and Northeastern University and covered by [Search Engine Journal]( https://www.searchenginejournal.com/research-shows-google-ai-mode-sends-less-clicks-is-a-poor-user-experience/590221/) in September, found people assigned to Google’s AI Mode were substantially less likely to click through to outside websites. They also ran fewer searches.
People are searching more and clicking less
None of this means search is shrinking. On its third-quarter call, Alphabet reported its fastest year-over-year growth in Google search queries in five years, according to [Duane Forrester’s analysis in Search Engine Journal]( https://www.searchenginejournal.com/google-search-queries-hit-a-5-year-high-and-search-revenue-didnt-follow/595485/). Search revenue grew 12 percent to $63.35 billion, down from 19 percent growth in the first quarter, and Google’s chief business officer Philipp Schindler pointed to “lower monetization we were seeing in AI Mode” as one reason.
The business logic is plain. Google is answering more questions on its own page and is now working out how to sell ads inside those answers. The same article notes Google plans to expand ads in AI Overviews and AI Mode. Forrester predicts the classic results page stops being Google’s default between 2028 and 2030. That is his forecast, and we treat it as one, but three straight quarters of slowing search revenue growth point the same way.
Your buyer is briefed before they reach your site
The bigger change is upstream of Google. A [McKinsey survey]( https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/who-wins-when-ai-does-the-shopping-new-evidence-on-agentic-commerce) of about 21,000 consumers in 11 markets found 53 percent had used AI for shopping in the past three months. The firm ran the survey in June and published it on October 28. In the US the figure was 48 percent. Nearly a third of AI shoppers said an AI chat or search tool was the single biggest influence on their most recent purchase.
McKinsey’s advice to brands is to make the AI’s consideration shortlist, which it ties to “machine-readable content and product data, third-party authority, and AI-search optimization.” It also repeats a Shopify figure: AI searches tied to a structured catalog convert at roughly twice the rate of searches working from scraped or outdated data.
That survey covers consumer shopping. Our view is that the same pattern is arriving in services and B2B, where a founder asks an assistant for three firms worth calling before anyone visits a website. That is our inference, not a McKinsey finding.
Where we would put SEO money in 2027
The work that still pays is concentrated on two things: the pages a buyer lands on when they are close to a decision, and the facts an AI tool reads before it recommends anyone.
- Split your search traffic by intent. Separate informational queries from branded and commercial ones in Search Console. Expect the first group to keep falling, and stop letting it set the mood in your monthly report.
- Rebuild the decision pages first. Service pages, case studies, location pages, comparison pages and the page that explains how to start working with you. These are where company-name and buying searches land, the searches the Agarwal and Sen study found holding steady.
- Make your facts readable by machines. Use structured data and keep what you do, who you serve, where you work and how to begin identical across your site and your profiles. McKinsey suggests a simple test: could your systems answer an assistant’s questions accurately right now?
- Earn mentions you do not control. “Third-party authority” in McKinsey’s language means being cited by people other than yourself: press, partners, directories and reviews.
- Keep publishing expertise, for a new reason. Articles that answer real questions are what AI tools read and quote. Judge them by citations and branded search growth. Sessions are the wrong scoreboard.
The tradeoff your report will show
Decision-stage searches have far less volume than informational ones. If you shift budget the way we recommend, total traffic may look worse while inquiries hold or improve. Leadership teams that only watch traffic will read that as failure. Agree on the new scoreboard before you change the plan.
We would stop buying high-volume blog content written to rank for definitions. A ranking report built on informational traffic now measures the part of search AI is absorbing first.
Start this month with the intent split. It takes an afternoon, and it tells you how much of your current SEO spend is pointed at the shrinking part. If the answer is most of it, we would be glad to look at your version of the plan with you.
