The average age of URLs cited by AI assistants was 1,064 days roughly 2.9 years old. In contrast, URLs appearing in organic Google results averaged 1,432 days old about 3.9 years.
This makes AI-cited content 25.7% fresher on average. AI systems—especially tools like ChatGPT—show a clear recency bias, often prioritizing newer or more recently updated content to deliver up-to-date answers. For example: ChatGPT exhibited the strongest preference for fresh pages among the platforms studied, sometimes citing content hundreds of days newer than what’s typical in Google rankings.
Perplexity and ChatGPT were noted for ordering in-text references from newest to oldest. This trend has been echoed across SEO and marketing discussions in 2025–2026, with implications for content strategy.
Publishers benefit from regular updates, keeping material current to increase the chances of AI citations; even if traditional SEO still favors established, authoritative pages. The full Ahrefs report (published around July 2025) dives deeper into model-specific differences.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
Noting that Google’s own AI overviews and classic organic results tend to favor older content more than third-party AI tools do. Overall, it highlights how AI-powered search and discovery is shifting emphasis toward freshness compared to traditional Google results.
The Ahrefs study on AI citations favoring fresher content; Ahrefs used data from their Brand Radar tool, an AI visibility tracking feature to collect and analyze a large-scale dataset of citations. They extracted 16.975 million cited URLs rounded to 17 million in summaries across multiple platforms.
This included citations from: Third-party AI assistants: ChatGPT, Perplexity, Gemini, Copilot. Google’s own features: AI Overviews (and potentially related tools). They calculated averages for both “time since publication” and “time since last update” to assess overall freshness bias.
Average age of AI-cited URLs: 1,064 days (? 2.9 years). Average age of URLs in organic Google SERPs: 1,432 days (? 3.9 years). Freshness difference: (1,432 – 1,064) / 1,432 ? 25.7% fresher for AI citations.
They segmented results by individual AI platform to highlight differences; ChatGPT showed the strongest recency bias, often citing content hundreds of days newer than Google’s typical results; Perplexity and others also leaned fresh but less extremely.
Google’s AI overviews were an outlier in some cases, sometimes citing slightly older content ?16 days older on average than organic SERPs in related analyses. Citation ordering was noted in some platforms; newer sources often listed first in in-text references.
The study drew from real-world queries and responses captured via Brand Radar, covering a broad range of topics and prompts. It focused on visible citations and URLs referenced in AI-generated answers, not internal retrieval or unshown sources.
Comparisons were apples-to-apples: same underlying content universe, but contrasting what AI tools chose to cite vs. what ranks organically in Google for comparable informational intent. This methodology leverages Ahrefs’ massive crawl/indexing infrastructure and AI-specific tracking to provide empirical evidence of recency bias in generative AI systems.
The full report includes charts, per-platform breakdowns, and implications for content creators—emphasizing that regular updates significantly boost chances of AI visibility, even if traditional rankings favor more established and authoritative pages.
The study has been widely referenced in 2025–2026 SEO discussions as evidence that freshness matters more in AI-driven discovery than in classic search rankings.
Citation Gaps Has Big Implications for SEO in AI Era
LLMs frequently cite URLs from much deeper in Google’s search results or sometimes not ranking highly at all, rather than pulling primarily from the top 10 or even top 20 organic positions.
Analyses including from Ahrefs and others aggregated in SEO reports show that only about 12% of URLs cited by major LLMs such as ChatGPT, Perplexity, and Copilot rank in Google’s top 10 for the relevant queries.
This implies that roughly 88% do not appear in the top 10. More specifically, around 80% or in some characterizations, up to 80–90% of cited URLs by tools like ChatGPT don’t rank in Google’s top 100 at all for the original query.
This figure is cited repeatedly in SEO discussions and summaries of studies, often tracing back to examinations of millions of citations. Traditional Google rankings reward factors like backlinks, keyword optimization, and page authority.
LLMs especially those with web access like ChatGPT Search or Perplexity often prioritize different signals: content structure; clear intros, lists, statistics, recency, entity clarity, first-party authority, or even training data biases.
They may favor deeper pages on authoritative domains, Reddit threads, Wikipedia, or niche sources that don’t dominate classic SERPs. Google’s own AI Overviews formerly SGE/AI Mode show more overlap with traditional rankings; often 60–90% from top results in some studies, but even there, a notable portion comes from outside the top 10.
Brands and publishers are increasingly focusing on Generative Engine Optimization (GEO) tactics: authoring in citable formats, using schema markup, publishing original research on their own domains, and building topical authority beyond just backlinks. The exact “eighty percent” phrasing shows up in LinkedIn posts, blogs, and SEO commentary summarizing these findings.
80% don’t rank in Google’s top 100 at all”, often referencing aggregated data from tools like Semrush, Ahrefs, and others analyzing hundreds of thousands to millions of LLM responses. LLMs aren’t just reranking Google’s top results; they’re using retrieval and synthesis strategies that pull from a much broader and sometimes unexpected pool of the web.
The citation gap—where roughly 80% of URLs cited by major LLMs (like ChatGPT, Perplexity, Gemini, and Copilot) don’t appear in Google’s top 100 results for the query—has profoundly reshaped content marketing strategies in 2026. Traditional success metrics like Google rankings and organic traffic are no longer sufficient on their own, as AI-driven discovery now mediates a growing share of user intent, often delivering synthesized answers with minimal or zero clicks to source sites.
This shift forces marketers to treat LLMs as a primary audience with one in four marketers now viewing them this way, per recent reports. Visibility means earning citations or mentions in AI responses, which can build brand authority, influence buyer journeys, and drive indirect conversions—even without direct traffic.
However, it also exacerbates challenges like reduced referral traffic from traditional search down 15–50% in some analyses and zero-click behaviors. Success is increasingly measured by Share of LLM, brand lift, AI mentions, and downstream influence rather than raw clicks or sessions.
Tools now track “cited in LLMs” alongside traditional analytics, as influence can occur without traffic. GEO focuses on making content citable by AI systems, emphasizing signals like clarity, structure, originality, and trustworthiness over pure keyword rankings.
Many brands now allocate budgets to GEO with predictions of 5x more spend on LLM optimization vs. classic SEO by 2029. Hybrid approaches combine both: SEO for foundational crawlability and rankings, GEO for AI-specific citability. Emphasis on high-quality, authoritative, original content.
LLMs favor E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals amplified for machines: Original research, proprietary data, unique datasets, case studies, and first-hand insights (these create “information gain” competitors can’t replicate).
Human-first storytelling with expert bylines, credentials, quotes, and transparency to counter AI-generated content saturation. Content updated within the last 30 days gets 3.2x more citations especially in ChatGPT, with 76%+ of top-cited pages refreshed recently.
Content Formats and Structure Optimized for Extraction
AI pulls passages, not full pages, so prioritize: Clear hierarchies (H1–H6 headings, bullet lists, tables, step-by-step guides). Self-contained paragraphs or sections (the “Island Test”—each block stands alone). Brand mentions and reputation influence citations 3x more than traditional backlinks in some studies.
Budget and Resource Reallocation
87% of marketers plan to increase content budgets in 2026 to counter AI disruption, focusing on durable assets like long-form reference content, research reports, and authoritative guides. Teams dedicate resources to monitoring LLM citations, reverse-engineering competitors’ sources, and iterating.
In essence, the era of “AI noticing your content” has arrived. Brands that adapt by creating citable, trustworthy, fresh assets—while treating LLMs as a core distribution channel—gain an edge in influence and long-term relevance. Those clinging solely to traditional Google rankings risk invisibility in the AI-mediated web.