A large hopper heaped with crushed stone above a single teaspoon of the same aggregate on a steel floor.

Zero-click hit 68% while AI Mode is 0.34% of searches

Clickstream data for January to April 2026 shows the click loss is coming from AI Overviews, not AI Mode, so most strategy work is aimed at the wrong surface.

Zero-click searches reached 68.01% of US Google searches between January and April 2026, up from 60.45% in 2024. Over the same window, AI Mode accounted for 0.34% of searches.

Read those two numbers together and the strategic picture inverts. Almost every AI search plan being written right now is aimed at AI Mode, the surface with a third of one percent of query volume, while the actual damage is coming from AI Overviews, which sit on ordinary search results pages that your existing SEO already targets.

What the data says and what it does not

The figures come from SparkToro's analysis of Similarweb clickstream data, US desktop and mobile web. Two caveats worth stating before you put them in a board pack.

It excludes searches made inside the Google mobile app, which is a significant share of mobile searching. And it is US data, so Australian behaviour may differ, though there is no reason to expect it differs enormously.

The direction is the part to trust. Searches producing at least one click fell 9.51 percentage points across two years, a relative decline of 22.9%. That trend line is consistent across every independent measurement.

The supporting evidence

Ahrefs measured a 58% CTR reduction for the top-ranking organic result when an AI Overview is present. Chartbeat, across more than 2,500 publisher sites, measured global Google search traffic down 33% year on year, with smaller sites losing around 60% of search referrals over two years against 22% for large publishers.

That last split is the one to sit with. The traffic loss is not evenly distributed. Large, established, well-known sites are absorbing it. Smaller sites are being hollowed out, which means brand strength is now a ranking-adjacent asset in a way it was not five years ago.

What Google actually published

In May 2026 Google shipped a guide called Optimising for generative AI features. It is short, it is plain, and it contradicts several things being sold as AI search services right now.

On llms.txt files: Google Search does not use them. They neither help nor harm.

On structured data: not required for generative AI search.

On content chunking: there is no requirement to break your content into tiny pieces.

On manufacturing brand mentions across the web: not as helpful as claimed.

The eligibility gate Google does name is simple. The page must be indexed and eligible to appear with a snippet. If you have applied nosnippet or a restrictive max-snippet directive anywhere, you have removed yourself from AI features on those pages. Check that first, because it is a five-minute audit that occasionally finds a catastrophe.

Google also names the content type it does not want, using its own example: commodity content, something like 7 Tips for First-Time Homebuyers. That format is still the default output of most content programmes.

The one genuinely good piece of news

AI-referred traffic converts far better than search traffic. Semrush puts AI-referred visitors at around 4.4 times the organic conversion rate. Seer's client data shows ChatGPT-referred traffic converting near 16% against roughly 1.8% for Google organic. ChatGPT referral volume grew 206% year on year from January 2025 to January 2026.

And it is still around 1% or less of most sites' sessions. Both facts are true. A tiny stream of extremely qualified visitors is worth having and is not yet worth restructuring your business around.

What to change

Change the KPI before you change the tactics. Sessions from organic search will keep falling and that is not necessarily a failure. Track impressions, citation presence, branded search volume and assisted conversions alongside sessions. A page that is cited in ten thousand AI Overviews and gets fewer clicks may be doing more work than it did before.

Segment your reporting by query class. Informational queries are where the clicks vanished. Transactional and navigational queries have held up far better. If you report one blended organic number, you are averaging a collapse and a plateau and learning nothing.

Audit snippet directives. Every nosnippet, every data-nosnippet attribute, every max-snippet value. Any of them can remove a page from consideration.

Stop producing the listicle. Google named the format. Publishing more of it is publishing into the category the ranking system was explicitly built to discount.

Deprioritise AI Mode-specific work. Not zero, because 0.34% will grow. But if AI Mode gets more attention in your plan than AI Overviews, your plan is aimed roughly a hundred times off target.

The traffic did not disappear. It got rationed, and the ration goes to whoever gets cited.

Written by David Eid. Published .