Google is still sending you traffic. For now.
Three experts at WAN-IFRA assessed where search is heading, and what publishers should do while there is still something left to protect
Publishers churning out commodity content and living off Google clicks are, in Barry Adams’s phrase, in serious trouble. He was one of three search specialists at WAN-IFRA, and the least diplomatic. Adams, an independent SEO consultant whose clients include the New York Times, came with a blunt read on AI overviews and where the traffic is heading. Clara Soteras Acosta, an SEO and digital strategy consultant and author of the SEO Playbook for Publishers, brought fresh data from Google I/O and a framework for acting on it. David Buttle, founder of DJB Strategies and a former director of platform strategy at the Financial Times, brought supply-side research spanning ten-billion-page referrals from eight major UK publishers. They disagreed on plenty. On one thing they didn’t: publishers treating this as a new crisis haven’t been paying attention to the old one.
Plan for zero
What would you do differently if you knew the Google traffic was going to zero? Soteras Acosta opened by asking publishers to build as if it already was. Not because anyone expects it to, but because it forces the right discipline. Planning for zero, she argued, is the only assumption that forces a publisher to stress-test everything it has been taking for granted. Even a slow, steady single-digit annual decline is enough to threaten a model with nothing else underneath it. The question is not whether to prepare for reduced search traffic, but whether you have already left it too late.
Soteras Acosta was just as careful, though, to push back on the doom-mongering. The “Google search is over” takes that have circulated for the past two years are inaccurate. Google remains, by a large margin, the dominant source of referral traffic for most publishers. ChatGPT, despite its growth, sits at a fraction of Google’s scale. Google isn’t dying, but the exchange rate of content for clicks is deteriorating, and publishers who haven’t built anything else are exposed.
AI overviews: the 40% problem
Adams walked through the mechanics. Trigger an AI overview on a query and the publishers sitting in the top five organic results lose around 40% of the clicks they’d otherwise get. In some cases it’s 10%; in others it reaches 70%. The average sits around 40%.
The partial protection for news publishers is real but narrowing. When Google detects a Top Stories box is relevant, it generally does not show an AI overview alongside it. But the shelf life of that window is shrinking: Adams cited research showing an AI overview appeared for a breaking news story just four hours after publication. For evergreen and informational content, there is no equivalent shelter. Adams was direct: publishers producing commodity informational content and relying on organic clicks are in serious trouble.
“Google has been telling us what it was going to do for fifteen years. What AI has done is accelerate that process. Now Google can actively replace that content with an AI.”
Barry Adams
Discover, social, and the shrinking window
Soteras Acosta’s presentation went beyond AI overviews to address where discovery is happening. Google Discover remains the primary traffic channel for many publishers: volatile, yes, but still significant. Her data from three publishers of different sizes showed Discover outperforming search for certain content categories. The risk is that the Discover feed is no longer just publisher content. In the US, Facebook and Twitter are appearing in Discover positions. In the UK and France, YouTube and Twitter sit at the top. The window through which publishers can reach audiences is closing on multiple fronts simultaneously.
She also flagged a structural shift in how Discover may be ranked going forward. Google signalled in December that social engagement metrics could feed into Discover placement, which means a publisher’s visibility in the feed may increasingly hinge on how its content performs on platforms it doesn’t control. That has direct implications for editorial strategy. Content that does not travel socially may simply not surface. Soteras Acosta was not alarmist about it, but she was clear: publishers need to understand which of their content generates social conversation, and treat that as an editorial input rather than a distribution afterthought.
AI Mode and the 1 billion number
Google claimed one billion monthly active AI Mode users in the figures released around Google I/O. Soteras Acosta presented those numbers; Adams interrogated them. His read was that most of those users clicked “show more” in an AI overview and were taken to AI Mode without intending to be there. Before Google made that switch automatic, deliberate daily AI Mode users numbered around 75 million, a small fraction of Google’s two billion daily users.
What the usage data does show, and what Soteras Acosta emphasised, is that search behaviour is changing regardless of which interface delivers it. More than one in six AI Mode searches are now multimodal, meaning image input alongside text, growing at over 40% month on month. Follow-up queries within a session are growing at the same rate in the US. People aren’t just searching more; they’re searching differently, in longer, chattier back-and-forths; and publishers built for single-keyword, single-click traffic simply aren’t built for that.
ChatGPT: a different problem
Buttle drew a sharper structural line. Google’s whole model rests on a value exchange: it uses publisher content to rank results, and in return sends traffic. That relationship gives publishers leverage, however diminished. ChatGPT has no equivalent obligation. OpenAI has signed a small number of licensing deals, but Buttle’s view was that those are principally about litigation risk management, not a genuine commitment to publisher value. The implication: ChatGPT’s growth is bad for publishers not just because it takes share from Google, but because it does so without any reciprocal traffic obligation at all.
The competition cuts the other way too. The more dominant Google stays, the less reason it has to keep up a meaningful exchange with publishers at all. Soteras Acosta noted that European fragmentation makes a coordinated publisher response harder: the UK’s nascent coalition approach has some precedent, but individual markets are operating under different regulatory frameworks and at different stages of AI overview rollout. France, for now, has had limited exposure to AI overviews. That is a lag, not a long-term protection.
The KPI problem
Soteras Acosta’s most pointed argument was about measurement. Publishers are still evaluating performance in terms of page views and unique users. Those metrics made sense when traffic was the engine of the business. They are the wrong instruments for what publishers need to build now. Engagement, time on site, retention, community conversion: those are the numbers that matter now, not because they’re more virtuous, but because they’re what’s left standing when referral traffic dries up.
She gave two examples of publishers that had reoriented around this. One Argentine economics publisher created a board game that teaches families how household economics works in everyday life, community engagement as a direct editorial product. The New York Times Cooking section, facing the same evergreen traffic erosion as everyone else, shifted its content strategy toward original formats and creator partnerships. Neither is a template. Both illustrate the same underlying logic: the value a publisher creates for its audience must exist independently of whether Google sends anyone to find it.
What GEO means
Adams closed on generative engine optimisation, the consultancy world’s repackaging of SEO for the AI era. His verdict was blunt: GEO is SEO with a 50% markup. The fundamentals have not changed. Authoritative, well-structured, non-commodity content that serves genuine user intent is what ranks in organic search, gets cited in AI overviews, and surfaces in answer engines. Publishers who have been doing that consistently are in reasonable shape. Those who do not have a structural problem that no amount of rebranded optimisation will fix.
Buttle’s framing was complementary. The publishers that will survive are those that have become destinations rather than waypoints: brands with a defined audience, a clear editorial identity, and talent that creates a direct relationship between the publication and its readers. The mechanism for getting there is not a new SEO strategy. It is an editorial and commercial decision about what the publication stands for, and for whom. That decision, Soteras Acosta suggested, is the one publishers can no longer put off.
Barry Adams is an independent SEO and audience growth consultant. Clara Soteras Acosta is an SEO and digital strategy consultant and author of the SEO Playbook for Publishers. David Buttle is founder of DJB Strategies and a former director of platform strategy at the Financial Times. All three spoke at the WAN-IFRA World Congress in Marseille. Clara’s blog post from this session can be accessed here.