AI Overviews vs AI Mode: What’s Actually Different for Publishers

No study reviewed here directly isolates AI Mode’s traffic impact from AI Overviews’ — but Chartbeat’s network data shows Google Search referrals to publishers falling roughly 33-40% year-over-year through 2025-2026, and an academic study specifically testing an AI Mode-only search experience found it cuts click-through rates and publisher trust further than the current mixed search experience.

Google ships two distinct AI-in-search surfaces that publishers routinely conflate: AI Overviews, the summary box that appears above traditional results for many queries, and AI Mode, a separate conversational search experience that replaces the results page with a chat-style interface. They are not the same feature, they were rolled out on different timelines, and — this is the uncomfortable part for anyone trying to isolate their effects — the public traffic data mostly measures Google Search referrals in aggregate, not AI Mode and AI Overviews separately.

What’s actually different between the two

AI Overviews sit inside the familiar search-results page: a generated summary above the normal blue links, which usually remain visible below it. AI Mode replaces the search-results page itself with a chat interface, where the “page” in the traditional sense — a ranked list of links a user can scan — mostly disappears, and whatever links appear are inline citations inside a conversational answer. As of a Press Gazette report from May 2026, AI Mode was not yet the default search experience, which the piece framed as a reprieve for publishers rather than a reassurance — the direction of travel, not the current state, was the concern.

The traffic numbers, and why they don’t cleanly separate

The most load-bearing data point comes from Chartbeat, via the Reuters Institute: between November 2024 and November 2025, Google Search referral traffic across more than 2,500 sites in Chartbeat’s network fell 33% worldwide and 38% in the US. A later figure, from what’s described as Chartbeat’s 2026 Publisher Playbook, put the year-over-year decline at 40.2% for the July 2025-July 2026 window — nearly double the 21.9% decline measured the year before that, with Google Discover’s decline widening to 34.3% over the same period. Axios-reported Chartbeat data from March 2026 broke the drop down by publisher size: small publishers down 60% over two years, medium publishers down 47%, large publishers down 22%.

Metric Figure Window Source
Google Search referrals, overall -33% worldwide / -38% US Nov 2024 – Nov 2025 Chartbeat via Reuters Institute
Google Search referrals, overall -40.2% YoY (vs -21.9% prior year) Jul 2025 – Jul 2026 Chartbeat 2026 Publisher Playbook (secondary reporting)
Small publishers -60% 2-year window to ~Mar 2026 Chartbeat via Axios
Google Discover referrals -34.3% YoY Jul 2025 – Jul 2026 Chartbeat 2026 Publisher Playbook

None of these figures isolate AI Mode or AI Overviews from each other, or from the broader shift of query volume toward zero-click answers generally — they measure total Google Search referral decline, which AI Mode, AI Overviews, and ordinary featured snippets all contribute to simultaneously. That’s a real gap in the public evidence, and it’s worth stating plainly rather than attributing a precise share of the decline to either feature individually, which would be more confident than the data supports.

The one study that does test AI Mode specifically

The clearest AI-Mode-specific evidence found here is an academic study reported by Search Engine Land: researchers tested an AI-Mode-only search experience (rather than today’s mixed experience where AI Mode is opt-in) and found it substantially reduced click-through rates to publisher sites and reduced user trust in the information Google surfaced, relative to traditional search. That’s an experimental estimate of a counterfactual — what traffic would look like if AI Mode became the default — not a measurement of current real-world AI Mode traffic, which matters because Google has not made AI Mode the default experience as of the most recent reporting reviewed here.

For context on the general zero-click dynamic driving all of this, see our coverage of the AI Overviews CTR collapse and how to measure AI referral traffic across ChatGPT, Perplexity, and Gemini — Google’s two features sit alongside a broader shift where informational, how-to, recipe, and affiliate content is the most exposed category, per multiple publisher-side reports.

What publishers are actually doing about it

A Reuters Institute survey of media leaders found they expect Google referral declines to continue — more than 40% further over the next three years, per one summary of the survey — and most planned to deprioritize traditional Google search optimization in 2026 in favor of direct audience relationships and other channels. Google’s own position, voiced by Search VP Liz Reid, disputes the framing that AI features are weakening the web, pointing to internal click-volume data that the company says tells a different story; outside reporting has continued to note declining publisher-reported traffic in the same period, so this is an active, unresolved dispute between the platform and the sites measuring their own referral logs, not a settled question.

Why “AI search” traffic figures need a feature label, not just a vendor label

Most publisher-side traffic reporting lumps every Google AI surface together under a single banner — “AI is killing our traffic” — which is directionally defensible given the referral numbers above but makes it hard to act on. A newsroom deciding where to invest SEO and content effort benefits from knowing whether a query type is being intercepted by AI Overviews (where the underlying page can still rank and get some click-through) versus AI Mode (where the interaction model itself de-emphasizes clicking through at all). Until a tracker publishes a feature-level breakdown — something none of the sources reviewed here currently do — publishers are left inferring which feature is responsible for a given query-category’s decline from indirect signals like whether AI Mode is enabled for their traffic’s geography and device mix.

FAQ

Is AI Mode responsible for the drop in Google referral traffic?
Unclear from the public data — Chartbeat’s referral-decline figures (33-40% depending on window) measure total Google Search referrals, not AI Mode specifically, and AI Mode has not been the default search experience during the periods those figures cover. The clearest AI-Mode-specific evidence is an experimental study of an AI-Mode-only scenario, not current real-world traffic.

What’s the practical difference between AI Overviews and AI Mode for a publisher’s traffic?
AI Overviews appear above a still-visible results page, so some click-through to ranked links typically survives. AI Mode replaces the results page with a chat interface where links become inline citations inside a generated answer — a structurally different exposure, though no dataset reviewed here cleanly separates the click-through impact of one from the other.

Which publishers are most exposed to this traffic shift?
Reporting consistently names informational, how-to, recipe, and affiliate-content publishers as the most exposed, since those query types are the ones most easily answered directly in a generated summary without a click-through being necessary.

Last updated 2026-11-13. This page is refreshed as benchmarks and scores move.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top