Publishers face AI search visibility crisis as summaries cut traffic

AI search has split publisher visibility into two jobs: ranking and extraction. A page can still earn a strong search position and lose the visit if an answer box or summary gives the user everything they need up front, which is why the old SEO playbook is no longer enough.

The new front door is an answer, not a link

The numbers make the shift hard to ignore. McKinsey says about 50% of Google searches already include AI summaries, and it expects that share to climb to more than 75% by 2028. It also says half of consumers polled intentionally seek out AI-powered search engines, while 44% of AI-powered search users already treat them as their primary source of insight, ahead of traditional search at 31%.

That is not a side issue for publishers, it is the new distribution layer. McKinsey’s own analysis says about $750 billion in U.S. revenue will funnel through AI-powered search by 2028, which means the business being routed through this layer is already large enough to shape media strategy. If your content is not easy for a model to interpret, attribute, and summarize, you are not just missing traffic, you are missing part of the market.

The timing matters too. Google’s antitrust win against news publishers in March removed one legal pressure point, but it did not slow the rise of AI-driven search. Publishers are left with fewer ways to influence how content is surfaced, which makes the mechanics of the page itself even more important.

Structure now matters as much as story quality

The highest-impact changes are the ones that make content easier for machines to parse without making it worse for readers. Schema markup belongs near the top of that list, along with clean sectioning, explicit attribution, and pages built around clear answers rather than buried context. If the model can identify the headline, the key fact, the named entity, the date, and the source quickly, it is more likely to credit the work accurately.

That means publishers should think harder about how every page is assembled. Keep the key takeaway high in the story, use specific subheads, separate background from the main point, and make quotes and statistics easy to lift without ambiguity. A page that is organized for humans and structurally obvious to AI systems is far less likely to be flattened into an unattributed blur.

The practical rule is simple: build for discoverability and extractability at the same time. Traditional ranking still matters, but so does whether an AI summary can pull the right sentence, the right number, and the right source name. The goal is no longer just to appear in search results. The goal is to remain visible after the system has rewritten the result into its own answer.

• Put a direct answer near the top of the page, not three scrolls down.
• Use schema on stories, explainers, product pages, and reviews.
• Make attribution unmistakable, with named sources and clean citations in the copy itself.
• Separate utility content, which may be more licensing-friendly, from premium reporting you want protected.
• Review templates for paragraphs that are too generic to be quoted cleanly.

The traffic warning signs are already here

The Reuters Institute for the Study of Journalism says publishers are already worried that AI platforms and chatbots could further reduce traffic flows to websites and apps. Its 2025 Digital News Report says news executives expect search referrals to fall by 43% over the next three years, which is a brutal forecast for any business still leaning on search as a default acquisition channel.

The current traffic picture backs up that fear. Chartbeat-based reporting cited by the Reuters Institute found Google search traffic to publishers declined globally by a third in the year to November 2025. Digiday reported in July 2025 that Google’s AI Overviews reduced click-through rates across more search queries, and that the feature had reached more than 2 billion monthly users, up from 1.5 billion the prior quarter.

The revenue hit is already showing up in the books. Penske Media saw affiliate revenue at its sites fall by more than a third from its peak as AI Overviews expanded. That is the part many publishers still underestimate: this is not just a traffic problem, it is a monetization problem, and affiliate, commerce, and lead-gen businesses feel the squeeze first.

The pressure is also uneven. Axios reported, in work by Sam Bradley, that smaller publishers have been hit especially hard, with sites in the 1,000 to 10,000 daily page view range seeing the most precipitous declines in the AI era. That tracks with what many newsroom operators are already seeing in their own dashboards: the biggest brands can absorb some erosion, while smaller operators lose the margin that kept them alive.

What publishers need to change now

This is the moment to treat AI visibility as an operating problem, not a theoretical one. Danielle Coffey and the News/Media Alliance have argued that Google’s AI features deprive publishers of traffic and revenue, and whether you agree with the politics of that fight or not, the business logic is hard to dismiss. If the answer appears before the click, publishers have to decide which parts of their work are meant to be summarized, which parts should be licensed, and which parts need to remain protected behind stronger distribution controls.

Measurement has to catch up too. Standard search traffic reports are no longer enough because they miss what happens when a summary absorbs the user’s intent before the click. Publishers should be tracking search referral trends, branded search growth, page-level click-through rates, and the gap between impressions and visits across story templates, especially for pages that rank well but still fail to send readers through.

The most useful shift is to stop treating AI visibility as separate from SEO. The same page now has to win in traditional rankings and survive extraction into summaries and answer boxes. The publishers that move first will not just protect search traffic, they will protect the ability to be cited, surfaced, and monetized in a search ecosystem that is already changing shape.