In
May and June 2025, the Digital Content Next trade association surveyed 19 of
its member companies, spanning major national newsrooms and global
entertainment brands, to measure what Google’s AI Overviews were actually
doing to their traffic. The result was unambiguous. Across eight weeks,
median year-on-year referral traffic from Google Search was down almost every
week, with losses outpacing gains two-to-one. The overall median decline was
10 per cent. For non-news publishers, the drop was 14 per cent. Even news
publishers, which Google has historically treated with some degree of care,
were down 7 per cent. As DCN’s CEO Jason Kint put it in the
organisation’s August 2025 analysis, these losses are a direct
consequence of Google AI Overviews, not cyclical noise or competition from
social platforms.
The mechanism is straightforward. Google’s AI Overviews generate a
text summary of the answer to a search query at the top of the results page,
synthesising content from publisher websites. Users who find their answer in
the summary have no reason to click through to the source. When AI Overviews
appear in search results, click-through rates fall to approximately 8 per
cent, compared to 15 per cent for equivalent queries without AI summaries.
Google processes billions of searches daily, and AI Overviews now appear for
more than 13 per cent of all queries, a figure that more than doubled in the
first half of 2025 alone. The arithmetic compounds quickly into very large
numbers of lost visits for the publishers whose content is being summarised
without their agreement.
Who Is Being Hit and How Hard
The traffic losses are not distributed evenly. Publishers
producing what is sometimes called “utility content”, the kind that
directly answers factual questions about how to do something, what something
costs, or where something is located, are most exposed. These are precisely
the queries for which AI Overviews are most useful to the user and most
damaging to the publisher. A lifestyle publisher in the UK’s Professional
Publishers Association documented the effect with a single illustrative
example: a query for which their page ranked on the first page of Google
results and impressions remained steady, but the click-through rate fell from
5.1 per cent to 0.6 per cent as AI Overviews captured the query. As Digiday’s
reporting on the DCN data noted, ten to 25 per cent year-on-year
CTR declines are now common even where publishers maintain stable
rankings.
News publishers face a slightly different version of the same
problem. Breaking news and developing stories are less easily reduced to a
static AI-generated summary, which provides some protection. But evergreen
news content, analysis, and explanatory journalism are directly exposed to
the same dynamic that affects non-news publishers. The Reuters Institute’s
annual Digital News Report documents a longer-term shift away from search as
a primary news discovery mechanism, with implications that extend beyond AI
Overviews to fundamental changes in how audiences find journalism
online.
What This Means for You
For anyone who reads news and specialist content online, the
deterioration of the publisher ecosystem is not an abstract commercial
problem. It is the mechanism by which the reporting and expertise that AI
Overviews are summarising gets produced in the first place. Investigative
journalism, specialist subject matter expertise, and local news coverage all
require revenue to sustain the organisations and individuals producing them.
If AI Overviews systematically extract the value of that content without
routing traffic that generates that revenue, the supply of the content being
summarised will diminish over time.
As our analysis of the
hidden costs that AI systems impose on existing ecosystems found,
the indirect effects of AI deployment often become visible only after the
damage has accumulated. The relationship between AI search and publisher
revenue is an economic externality that the current architecture of AI
Overviews does not account for, and the question of whether it should be
accounted for, and by whom, is moving through regulatory discussions in multiple
jurisdictions simultaneously.
What Google Says and Why It Does Not Fully Address the
Problem
Google’s public position on AI Overviews and publisher traffic
involves two main claims: that AI Overviews drive more qualified traffic to
publishers, and that the clicks they generate are higher in intent and
therefore more commercially valuable. Both claims may be partially true.
Traffic driven by users who have already had their initial question answered
is likely to represent more engaged visitors than casual browsers. But the
relevant comparison for publishers is not quality versus quality but the
total revenue yield of the traffic they receive before and after AI
Overviews, and on that measure the DCN data suggests the picture is
negative.
The regulatory dimension is developing separately. As we examined
in our coverage of the
evolving regulatory landscape for AI systems, the EU’s approach to
AI governance includes provisions that could eventually address the
relationship between AI intermediaries and content creators. Australia,
Canada, and the UK are each at different stages of legislative debates about
whether platforms that derive commercial value from news content should
compensate the organisations producing it. AI Overviews add a new dimension
to that debate: the question is no longer just about platform distribution
but about AI systems that synthesise content to eliminate the user’s need to
visit the source.
The Structural Question That Remains Unanswered
The deeper issue beneath the traffic numbers is whether the
current architecture of AI search is compatible with a sustainable ecosystem
of original content production. AI Overviews depend on the existence of
high-quality publisher content to synthesise. If that content is not
economically viable to produce, because the traffic it generates is
insufficient to sustain the organisations producing it, the quality of the
summaries will eventually degrade along with the quality of the sources they
draw on. This is not a hypothetical long-term concern; it is a dynamic that
publishers are already navigating, with real effects on editorial investment,
headcount, and the scope of what they can afford to cover.
The question of who
benefits from AI-driven change and who bears its costs has no
cleaner illustration than the relationship between AI search and the
publishing industry. The users benefit from faster, more convenient answers.
Google benefits from users spending more time within its ecosystem. The
publishers whose expertise and reporting makes those answers possible bear
the costs of decreased traffic without a compensating payment. How that
distributional question is resolved, through regulation, commercial
negotiation, or structural change in how AI search works, will shape the
information environment that AI systems themselves depend on to be
useful.
The counterfactual is difficult to construct but worth attempting.
If AI Overviews did not exist, would publishers be thriving? Almost certainly
not: social media platforms have been reducing news referral traffic for
years, print advertising revenues have been in structural decline for over a
decade, and the shift from desktop to mobile browsing has compressed
attention spans and page views simultaneously. The AI Overviews effect is
real and measurable, but it is arriving on top of a structural crisis that
predates it. That context does not diminish the specific harm, but it shapes
what policy responses are realistic. Compensating publishers for content used
in AI training is one avenue being explored; adjusting how AI Overviews cite
and link to sources is another. Neither is a complete solution, and none
restores the economics of publishing to the pre-internet baseline. What is at
stake is whether the transition leaves any viable economic model for original
content production at the other end.
About the Author
Stuart Kerr is the Technology Correspondent for LiveAIWire. He
writes about artificial intelligence, emerging technology, and the forces
reshaping work, business, and society.