- News site keywords trigger AI Overviews at only 10.5% — far below the 20.9% rate for informational sites and the 43.6% peak seen in Science, according to Ahrefs’ analysis of 146 million SERPs.
- The protection is structural, not permanent: Google’s Query Deserves Freshness (QDF) signal and news content’s time-sensitivity appear to suppress AI Overview generation for breaking queries.
- Non-news publishers face the steeper threat — informational, evergreen content sits squarely in the category AI Overviews are designed to answer.
- Non-news brands saw a median -14% year-over-year decline in Google referral traffic versus -7% for news brands, per Digiday, suggesting the trigger-rate disparity is already showing up in traffic data.
What Happened
Ahrefs published a large-scale study examining what triggers Google AI Overviews, drawing on data from 146 million search engine results pages. The study found that AI Overviews appear on roughly 20.5% of all queries — but that figure masks extreme variation across content categories.
Science keywords trigger AI Overviews 43.6% of the time. Health follows at 43.0%. Pets and Animals sit at 36.8%, and People and Society at 35.3%. At the bottom: Shopping at 3.2%, Real Estate at 5.8%, and News at roughly 10.5% when measured against informational site baselines.
The study, which examined 86 keyword traits, found that 99.9% of all keywords triggering AI Overviews carry informational search intent. Shopping, transactional, and time-sensitive queries are largely bypassed.
Why It Matters
The trigger-rate gap has direct revenue consequences. When an AI Overview appears, click-through rates to publisher websites fall sharply. Ahrefs’ own click data shows AI Overviews reduce clicks by as much as 34–58% depending on query type. Users who encountered an AI Overview clicked a traditional search result 8% of the time, compared with 15% for users who saw standard results, according to Digiday.
A lower trigger rate for news therefore translates into less direct click suppression from AI Overviews — even if news publishers face separate headwinds from zero-click search trends and algorithm changes. Non-news informational publishers have no such buffer.
The disparity is already visible in aggregate traffic numbers. Digital Content Next surveyed roughly 40 member companies across May and June 2025 and found the median year-over-year Google referral decline was -7% for news brands and -14% for non-news brands. Google AI Overviews was cited as a primary driver.
Technical Details
Two structural factors appear to suppress AI Overview generation for news queries.
The first is Google’s Query Deserves Freshness (QDF) signal. When QDF is active — as it typically is on breaking or rapidly updating topics — AI Overviews are suppressed in favor of a standard Top Stories carousel. The Gemini model that powers AI Overviews is not trained on real-time data, making it structurally unsuited to answer questions where the answer changes by the hour. As Barry Adams noted at SEO for Google News, QDF and AI Overviews appear to be mutually exclusive for a given query.
The second factor is query length and format. Ahrefs found that AI Overviews appear on only 9.5% of single-word queries but jump to 46.4% of queries with seven or more words. News searches tend to be shorter, proper-noun-heavy, and event-specific — patterns that map poorly onto the long-tail, question-format queries that most reliably trigger AI Overviews. Sixty-six percent of all AI-Overview-triggering keywords are classified as questions.
Science and Health queries, by contrast, are often long, evergreen, and question-shaped — “what causes inflammation”, “how does mRNA work” — which is precisely the format AI Overviews are optimised to answer.
Who’s Affected
Publishers producing evergreen reference content — health information sites, educational platforms, how-to guides, and science explainers — are absorbing the heaviest impact. These categories sit at the top of the trigger-rate distribution and have limited ability to pivot toward the breaking-news formats that appear insulated.
News publishers are not unaffected. Organic traffic to the top 50 US news sites fell from approximately 2.3 billion monthly visits in mid-2024 to under 1.7 billion by May 2025, a loss exceeding 600 million monthly visits in under a year, according to The Digital Bloom’s 2025 Organic Traffic Crisis report. CNN suffered year-over-year traffic declines between 27% and 38%. Those losses are driven primarily by broader zero-click search behaviour and algorithm shifts rather than AI Overviews specifically — but the distinction matters for strategy.
The NPR report on news publishers quoted several editors describing conditions as an extinction-level event, though the data suggests the mechanism differs by publisher type. Shopping sites are statistically least affected by AI Overviews at 3.2%, but face different pressures from AI-powered comparison and product recommendation features.
What’s Next
The current insulation for news is tied to technical constraints that are narrowing. As Google trains AI Overviews on more recent data and integrates real-time retrieval, the QDF exclusion may shrink. NewzDash’s 2025 study found AI Overviews already appearing on some news-adjacent queries where freshness was less critical, suggesting the boundary is not fixed.
Breaking news traffic across Google platforms more than doubled — rising 103% — from late 2024 through early 2026, with Google Discover growing 30% year-over-year and becoming a significant referral source for news publishers, per ALM Corp’s analysis. Publishers leaning into original, time-stamped reporting appear to be capturing that growth.
For informational publishers, the adjustment window is shorter. The concrete steps that emerge from the data:
- Audit content by trigger vulnerability. Map existing content against the high-trigger categories (Science, Health, Pets, People and Society). Evergreen explainers in those verticals face the highest AI Overview competition and lowest marginal CTR.
- Invest in original data and primary reporting. AI Overviews draw on existing indexed content. Content that contains original research, exclusive quotes, or proprietary datasets is harder to summarize and more likely to be cited rather than replaced.
- Treat Google Discover as a primary channel. Discover is growing in referral share and operates on recency and engagement signals that differ from standard organic search, giving publishers a path that bypasses AI Overview suppression entirely.
- Diversify traffic sources. Email newsletters, direct traffic, and social referrals do not carry AI Overview exposure. Publishers reporting the least disruption tend to have the most diversified referral mix.
- Monitor trigger rates at the query level. Tools including Otterly.ai and Ahrefs now surface AI Overview presence by keyword, enabling publishers to identify exactly which landing pages face direct competition from AI-generated summaries.
