If you judge AI’s impact on news publishers only by referral tags, you’re missing most of the picture. A new Scrunch analysis of millions of searches, AI conversations and page visits finds that just 1.1% of post-chat visits to news sites carry an AI referrer — yet news-related AI conversations still coincide with materially higher follow-through to publisher pages, almost always via other routes.

What the dataset reveals

Scrunch compared individual browsing histories from February to June 2026 to measure how often a visit to a news publisher followed an AI conversation. Readers were 20.5 percentage points more likely to visit a news publisher in the week after a news-related AI conversation than after a non-news AI chat. Scrunch stresses this is a routing correlation, not proof that AI created new demand.

Of the visits that followed an AI chat, roughly three-quarters arrived through direct navigation, about 9% through traditional search, and just 1.1% carried an AI referrer. Scrunch notes an important limitation: referral tags record only the last click before a visit and cannot reconstruct the full sequence of interactions across AI conversations, searches and later page visits.

How AI Overviews change click behavior

The firm also looked at Google’s AI Overviews. Overviews appeared on about one in four news searches during the study period and were most common for utility-style queries — weather, market prices and explainers. Searches that displayed an AI Overview generated publisher clicks roughly 20% of the time; similar searches without an Overview generated publisher clicks about 30% of the time.

When Scrunch compared the same query with and without an AI Overview, the difference in click-through rates shrank to about two percentage points. That suggests much of the raw gap is explained by query mix: Overviews are more likely to appear on queries that historically drive fewer publisher clicks.

AI mentions disproportionately route readers to big names

When an AI assistant named a publisher that the user hadn’t included in the prompt, visits to that publisher rose noticeably — by 10.6 percentage points the next day and by nearly 20 percentage points over the following week versus similar publishers not mentioned. Scrunch frames this as a routing signal rather than definitive causation.

Major publishers dominated AI mentions and the subsequent visits: they accounted for 82% of publishers named by AI assistants and 97% of the follow-through visits tied to those mentions. Mid-sized outlets made up about 17% of mentions but captured a much smaller share of resulting traffic.

What this means for publishers and marketers

The headline figure — 1.1% AI referrers — can mislead if it’s used as the only metric for AI’s impact. The study shows AI visibility often translates into later actions that avoid an AI referrer tag: users may remember a publisher’s name, return directly, or run a traditional search after an AI conversation.

Two practical dynamics matter. First, brand recall drives direct navigation: if a reader remembers a site name after an AI interaction, they’ll often go straight there. Second, mentions inside AI responses function as routing signals that favor established brands over mid-sized publishers.

Practical next steps

  • Don’t rely solely on referral logs. Referral tags undercount AI’s influence; combine them with panel-based or visibility-tracking tools that connect AI-answer exposure to later behavior.
  • Map where AI Overviews appear. Overviews cluster on utility queries with lower publisher click rates; knowing which queries trigger them helps set realistic distribution expectations.
  • Improve brand clarity. Smaller outlets should make their brand and site identity explicit in headlines, metadata and structured data to increase the chance of being surfaced or recalled by readers.

Scrunch used a privacy-safe, opt-in panel and measured page visits, not subscriptions, revenue or time on site. That scope is important: the analysis documents routing and click behavior, not commercial outcomes. For measurement teams, the immediate next step is clear — layer AI-answer visibility tracking onto existing referral and search analytics so you can see both direct and indirect paths from AI conversations to your pages. Watch whether AI mentions continue to concentrate traffic on major outlets; if they do, mid-sized publishers will need clearer branding and discoverability tactics to stay visible.