Is AI search cutting your website traffic, and what should you track instead?

Last updated October 14, 2026. Written by the Third Coast AI team.

AI search is cutting clicks on informational searches, and the measurements agree on the size: Pew Research found Google users clicked a traditional result in 8% of visits when an AI summary appeared against 15% when it did not, Ahrefs measured a 34.5% lower click-through rate for the top-ranking page when an AI Overview was present, and both found that a large share of searches now end on the results page, which means sessions are no longer a fair measure of whether your marketing is working and the thing to track instead is whether the engines name and cite you when a buyer asks, and what the fewer visitors who do arrive go on to do. Commercial searches were slower to get AI answers than informational ones, but that gap is closing.

This page sets out the evidence on clicks, explains why the loss is uneven by query type, and gives the set of measures we report to owners in place of sessions. It is written for someone looking at a traffic graph that has gone the wrong way and wondering what to make of it.

How large is the click decline?

Pew Research Center's July 2025 study tracked the browsing of 900 US adults through March 2025 and found that users who encountered an AI summary clicked on a traditional search result in 8% of all visits, while those who did not encounter one clicked nearly twice as often, in 15% of visits. Users ended their browsing session entirely after 26% of pages with an AI summary against 16% of pages with only traditional results, and they clicked a link inside the summary itself in just 1% of visits, which disposes of the idea that being cited in the summary brings the click back. Ahrefs' 300,000-keyword click study from April 2025 compared 150,000 keywords with an AI Overview against 150,000 informational keywords without one and found the presence of an AI Overview correlated with a 34.5% lower average click-through rate for the top-ranking page, with the average desktop click-through for position one on those keywords falling from 0.073 in March 2024 to 0.026 in March 2025.

The presence numbers are moving too. BrightEdge's tracking puts AI Overviews on nearly half of Google queries, and Semrush measured a 71% rise on commercial-intent searches between November 2025 and April 2026, so the category of search that once seemed protected, the buyer looking for a vendor, is now getting a written answer more often than not. Pew's study predates most of that expansion.

Why is the loss uneven?

Because an AI answer resolves some questions completely and others not at all. A searcher asking how long to boil an egg is finished when the summary says twelve minutes, and the recipe site loses a visit it was never going to monetize well. A buyer asking who should install a commercial HVAC system in Ottawa County gets a list of three companies and a set of criteria, and then has to pick up the phone, visit a site or ask a follow-up, so the click still happens but it happens later and to fewer companies. Ahrefs' study found 99.2% of keywords that trigger AI Overviews are informational, which is why its comparison was built on informational keywords; the commercial side is where the click survives, and it is also where being one of the three names matters most.

Our own measure of that is simpler. Glenn Gabe's panel of sites still shows AI search sending under 1% of visits to most of them, while Semrush's AI search traffic study logs a 4.4 times higher conversion rate for AI-search visitors than for classic organic traffic. Fewer visits, better ones, and a recommendation that happened before the visit.

Which metrics should replace sessions?

The replacement is a small set, and the first two are the ones most companies have never measured. Semrush's 2026 AI Visibility Index reported that 45% of marketing leaders cannot accurately measure their brand's visibility in AI answers and only 9% have tools that track every relevant metric, which matches what we find on first calls.

• Named rate. The share of runs, across a fixed bank of 60 to 120 buyer prompts, in which each engine names your company, reported by engine and by prompt cluster and run more than once per prompt. This is the number that replaces rank.
• Citation rate. The share of runs in which one of your pages is cited, and which page, so that you know which content the engines trust.
• Accuracy. What the engines say about your pricing, services, locations and people against what is true, logged from fact-check prompts and re-checked quarterly.
• Share against competitors. Who else is named for the same prompts, since a prompt where you are absent is lost to someone specific.
• AI referrals and what they do. Sessions from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com in your analytics, small in count, tracked for conversion rate rather than volume, plus Bing Webmaster Tools' free AI Performance report for Copilot.
• Branded search and direct traffic. People who were told your name by an engine often arrive by typing it, so a rise in branded search alongside a fall in informational organic is the signature of AI visibility working.
• Lead quality. Source and outcome of each inbound lead, since the 4.4 times conversion difference shows up here first.

Should you still invest in informational content?

Yes, with a different purpose. Informational pages used to earn visits and the visits earned leads; now the same pages earn citations and the citations earn a place in the recommendation, which is a longer chain with a smaller click at the end. The pages worth writing are the ones that answer the hidden searches the engines run behind a buyer's question, because those are retrieved during a commercial decision rather than during homework, and our content structure guide describes the shape they need. The pages not worth writing are the generic explainers that an AI summary resolves on the results page, unless they carry a decision factor the engines will quote with your name attached.

How should you present this to an owner?

Show the two graphs side by side: organic sessions falling, and the named rate across the prompt bank rising or flat, with the fact-check log underneath. A falling traffic graph with a rising named rate is a business being recommended more and visited less, which is the pattern the click studies predict, and the lead log will confirm it within two quarters. A falling traffic graph with a flat named rate is a problem, and it is one the audit is built to diagnose. Either way, the reporting guide describes the monthly report we would put in front of the leadership team.

Frequently asked questions