A search result used to mean a visitor. In 2026 it increasingly means a summary — read, answered, and closed without a click. AI Overviews now appear in 58% of Google queries, up from 12% in 2024, and 64.82% of all searches end without anyone landing on a website. If your funnel was sized for last decade's traffic volume, the math underneath it has already changed.
What changed in search this year?
Google's AI Mode pushes the zero-click rate even further, to 93% of its queries. The pattern is consistent across the reporting: fewer raw sessions land on any given site, because the assistant already gave a usable answer. That is not a ranking problem you fix with more content. It is a traffic-shape problem, and it shows up first in your analytics as a flatter top of funnel with the same or better bottom-of-funnel numbers.
Why fewer visitors can mean more revenue
The visitor who clicks through an AI summary already has the background. They are not arriving to learn what your category is; they are arriving to decide between options, or to find the specific proof — a price, a timeline, a named source — that the summary could not give them. That is a higher-intent visitor than the one who used to click every blue link. Zoho's PageSense team reports these post-summary clicks converting 23% better than traditional search visitors. A funnel built for volume treats this visitor the same as a cold one: full explainer, long form, generic CTA. A funnel built for 2026 gets to proof and price faster, because the visitor already did the research your homepage used to do for them.
The traffic is not coming back
Anyone selling you a plugin or a prompt trick that recovers the clicks an AI Overview absorbed is selling the demo, not the fix. There is no putting those sessions back. There is only converting the ones you kept at a higher rate than before.
Where CRO and GEO meet
The same shift that shrinks click volume is why generative engine optimization matters: earning the citation inside an AI Overview is now part of the funnel, not separate from it. A page that is easy for an AI system to summarize accurately is also, usually, a page with a clear thesis and named proof near the top — the same edit a CRO audit would recommend anyway. Both disciplines fail for the same reason: a page written to rank instead of a page written to be believed. Treat GEO and CRO as one funnel with two entry points, not two vendors sending two invoices.
What to change in your CRO program now
None of this calls for a redesign. It calls for five specific, testable changes, in this order:
- Audit your top landing pages assuming the visitor already read a summary of what you do — cut the category explainer, keep the differentiation.
- Move proof — a price, a named source, a specific number — above the fold instead of after three scrolls.
- Cut form fields on your primary CTA; every field is a decision a summary-primed visitor did not come to make.
- Split AI-referral traffic into its own analytics segment instead of lumping it under organic; the conversion rate is a different number and a different story.
- Re-run CTA and headline tests monthly; a test result from a different visitor mix is not evidence for this one.
How do you know if it's working?
Total session count will likely keep drifting down; that is the industry, not your marketing. The number to watch is conversion rate by traffic source, month over month, with AI-referral and organic search reported separately. Set this up as a referrer-based segment: flag sessions whose referrer domain matches known AI assistants (chatgpt.com, perplexity.ai, gemini.google.com) apart from Google's organic listing, since Google does not yet expose an AI Overview click as its own referrer. If AI-referral sessions convert at or above the 23% lift the PageSense data describes, your landing pages are matched to the visitor you are actually getting. If they convert at or below your old organic baseline, the page is still written for a visitor who no longer exists.
Fewer clicks. Each one matters more.
Hamed Azimi
Founder, NexusWave
Hamed Azimi is the founder of NexusWave Technologies, a Vancouver-based AI-native digital agency and product studio.
