Zero-Click Search in 2026: What AI Overviews Really Did to Traffic

Where do the zero-click numbers come from?

Most headline figures come from large-scale impression studies — panels and clickstream data covering millions of searches, comparing CTR before and after AI Overviews appeared for the same queries. The methodology is sound, and it is also why the numbers vary so much between studies: the result depends heavily on which queries are in the sample.

Treat any single figure as directional. The pattern is consistent across the research: definitional and informational queries lost the most, commercial and local queries held up, and branded queries improved. The exact percentages will keep moving as the feature changes.

Which queries are actually affected?

Four categories behave differently. Definitional queries — 'what is', 'how does' — lost the most traffic, often by more than half, because the answer is complete on the results page. Comparison and 'best' queries lost a moderate share, because people still want to read reviews before they spend money.

Local and commercial queries held up, and branded queries improved. The pattern is intuitive once you see it: the more a query requires judgement, trust or action, the more likely someone still clicks.

• Definitional ('what is X'): heaviest losses, often 50% or more
• Comparison ('best X', 'X vs Y'): moderate losses, roughly 15–30%
• Local ('X near me'): largely stable
• Branded ('your name'): often improved, up around 18% with an AI Overview

Why do branded queries go up?

When an AI answer cites your domain, it does something an ordinary blue link cannot: it borrows the assistant's credibility. Someone who sees your brand inside a trusted answer arrives pre-convinced, and many of them then search for you directly.

This is the counter-intuitive finding of 2026. The same feature that removes clicks from generic queries adds intent to branded ones — provided you are the source being quoted.

There is a second effect underneath. Being cited repeatedly teaches the assistant which brand belongs to which category, which makes your name more likely to appear in the next answer too. Visibility compounds here in a way rankings never did, because the citation itself becomes an input to future citations.

What should a business website do about it?

Stop measuring success as sessions and start measuring it as citations plus branded search volume. A page cited in AI answers across twenty queries is doing more for pipeline than a page earning ten thousand indifferent visits.

The measurement change is the harder part. If you keep reporting sessions as the headline number, you will conclude that marketing stopped working at exactly the moment it started working differently. Report citations, branded search and qualified enquiries together, and expect the first two to lead the third.

• Track branded search volume monthly — it is the cleanest signal of AI-era visibility
• Add answer-first paragraphs to every page that targets a question
• Publish original data, benchmarks and specific numbers — the most citable material
• Keep entity facts identical across your site, profiles and directories

How do you measure the shift on your own site?

Segment organic traffic by query type rather than looking at the total. In Search Console, compare impressions and clicks for branded, local, comparison and definitional queries separately. A flat total often hides a collapse in one segment and growth in another.

Then add branded search volume and enquiry volume to the same dashboard. If branded searches and enquiries are rising while sessions fall, you are experiencing the shift rather than suffering from it — and you would never know that from a traffic chart alone.

Which content still earns reliable clicks?

Content that requires trust, specificity or action: pricing pages, comparison pages with honest trade-offs, case studies, calculators, templates and anything built on original data. AI can summarise an opinion; it cannot replace a tool, a quote or a decision made with you.

The safest long-term position is to be the source that both search engines and assistants need — not the summary that competes with them.

Pricing pages deserve a special mention. They are among the most resistant to summarisation, because no AI answer can commit to your price on your behalf, and among the highest-intent pages on any commercial site. If your pricing page says 'contact us for a quote', you are giving away the most valuable click you have.

What should you do differently in 2026?

Publish fewer, more specific pages. The generic explainer is the format AI summarises most effectively, so it is also the format that produces the least traffic. A page answering one narrow buying question with real specifics is harder to summarise away and easier to cite.

Add things a summary cannot reproduce: original benchmarks, calculators, templates, real project results, honest trade-off comparisons. Anything requiring your data or your judgement survives the shift intact.

• Answer the core question in the first 60 words of every page
• Publish original numbers and results rather than restated opinion
• Build tools and calculators — they cannot be summarised into a paragraph

Is this the end of SEO?

No, but it is the end of SEO as a traffic-volume game. Rankings still determine who is eligible to be cited, and eligibility is the prerequisite for everything else. What has changed is that ranking first no longer guarantees the click, and ranking third can now produce a citation worth more than a click ever was.

The businesses adapting fastest are the ones that stop treating search as a channel and start treating it as a reputation system. Being the source that answers, cites and corroborates is a slower, more durable position than being the page that briefly ranked.