AI Answers Are Taking the Click. What Should Brands Measure Now?
When Google shows an AI summary, people click far less. Here is what that means for brand visibility, and what to track instead of traffic alone.
When an AI summary sits on top of the results, people click less: about half as often, in the best public data available. For brands, the consequence is that being named and cited inside the answer now matters as much as ranking below it. Traffic is no longer a complete measure of search visibility, so the scorecard has to change.
How much do AI summaries cut clicks?
The clearest independent measurement comes from Pew Research Center. It tracked the real browsing of 900 US adults during March 2025, covering 68,879 Google searches. On visits where an AI summary appeared, users clicked a traditional search result 8% of the time. Without a summary, they clicked 15% of the time. Links cited inside the summary were clicked on roughly 1% of visits. Users were also more likely to end their browsing session after seeing a summary: 26% of the time, against 16% without one.
Google disputes readings like this and points to growing search usage. Both things can be true. More searching does not mean more clicks per search, and the clicks are what most content programmes were built to capture.
Is this getting bigger or smaller?
Bigger. In March 2025, Pew found that 18% of searches in its panel produced an AI summary. Since then Google has pushed further into conversational search. On its October 28, 2026 earnings call, Alphabet said queries in AI Mode run three times longer on average than traditional searches, and that one in five AI Mode searches involves voice, images or documents. Longer, more specific questions are exactly the ones an AI answer can resolve without sending the user anywhere.
The same call showed the commercial side: Google is now placing ads inside AI Overviews and AI Mode in more markets. The answer box has become the primary real estate, for paid and earned alike.
What does “visibility” mean when nobody clicks?
It means being part of the answer. If a buyer asks an assistant which tools, brands or approaches fit their problem, the brands that are named have been seen, even if nobody visits a website. The ones left out have lost the moment entirely.
This is closer to how brand advertising has always worked than to how search marketing has. It rewards being the obvious, frequently cited name in a category, which is a brand problem as much as an SEO one.
What should replace traffic in the scorecard?
Keep traffic, but stop treating it as the whole story. A practical set:
- AI answer presence. For the 20–50 questions your buyers actually ask, check regularly whether your brand is named, and whether your pages are cited. Do it across Google AI Mode, ChatGPT, Perplexity and Copilot, since each draws on different sources.
- Share of voice in answers. Of the brands named for those questions, how often is it you? This is the AI-era version of share of search.
- Branded search and direct traffic. People who meet your name in an answer often come back by searching for it. Rising branded demand alongside flat organic traffic is a sign the answers are working for you.
- Conversion per visit. Visitors who do click through from an AI answer have usually done their research. Fewer visits with a higher conversion rate can be a better outcome than the old volume.
- AI crawler access. Check your server logs or your CDN for visits from GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Googlebot. If they cannot read your pages, you cannot be cited.
How do you get cited more often?
There is no switch, but the patterns are consistent with how retrieval systems pick passages:
- Answer first. Open each page with a direct, two- or three-sentence answer to the question its title implies.
- Write quotable paragraphs. Self-contained passages of 40–80 words, each making one point, are easy for an AI system to lift and attribute.
- Put the number, the source and the year in the same sentence. A figure that can be cited without its surrounding context is far more likely to be cited.
- Be the primary source when you can. Original data, named examples and clear definitions are what answer engines need and cannot generate themselves.
- Keep pages fresh and dated. Visible “published” and “updated” dates help systems choose the current version of a fact.
None of this conflicts with Google’s own position. In February 2023 Google said it rewards original, high-quality content “however it is produced”, and that the problem is content made primarily to manipulate rankings. Writing for citation is writing for clarity.
What should you do this quarter?
Pick the questions that matter commercially, and record which brands the main assistants name today. That is your baseline. Then rewrite your most important pages answer-first, add sourced figures and FAQs, and re-check monthly. Report AI answer presence next to traffic, so that a fall in clicks is not mistaken for a fall in relevance.
Frequently asked questions
- Do AI Overviews reduce clicks to websites?
- Pew Research Center's March 2025 browsing panel found that Google users clicked a traditional result on 8% of visits when an AI summary appeared, against 15% when none appeared. Clicks on the links inside the summary itself happened on about 1% of visits.
- What should marketers measure if organic traffic falls?
- Track whether your brand is named and cited in AI answers for the questions your buyers ask, alongside branded search volume, direct traffic and conversions. Traffic alone now understates visibility, because many answers are read without a click.
- Does Google penalise AI-written content?
- No. Google said in February 2023 that it rewards high-quality content however it is produced, and that using automation to manipulate rankings is what breaks its spam policies.
Sources
- Google users are less likely to click on links when an AI summary appears in the results, Pew Research Center (2025)
- Google AI Overviews are killing the web, Pew study shows (again), The Register (2025)
- Google expands AI Search ads globally as core ad growth slows, TechCrunch (2026)
- Google Search's guidance about AI-generated content, Google Search Central (2023)
Agentic Checkout Meets Reality: What Retailers Should Do Now
OpenAI pulled back from checkout inside ChatGPT while Google pushes its Universal Commerce Protocol. Product data, not checkout, is what to fix first.
ChatGPT Ads Reach 89 Countries. Do They Actually Sell?
OpenAI's ad business grew fast in 2026, but the first independent study finds paid ChatGPT visits earn far less than paid search. How to test it.
The EU's AI Labelling Rules Are Live. What Must Brands Label?
Article 50 of the EU AI Act has applied since 2 August 2026. Deepfakes need a label; most edited marketing copy does not. A practical guide for brand teams.