
Do AI tools cite newer pages more than older ones?
Somewhat, and mostly through updates: a Seer Interactive study of pages cited by ChatGPT, Gemini, and Perplexity found 75% had been updated within the past year, while in a subset only 42% were first published that recently. Ahrefs found AI-cited pages were about 25.7% newer on average than pages in Google's organic results, and ChatGPT showed the strongest lean. Changing a date without changing the page is not what these studies measured.
Key takeaways
- Seer found 75% of pages cited by ChatGPT, Gemini, and Perplexity had been updated within the past year.
- In a Seer subset, 72% of cited pages were fresh by last update but only 42% by first publication.
- Seer's highest share was Gemini and Ahrefs' biggest gap was ChatGPT, so the studies do not agree on one engine.
- Discovered Labs, an AEO vendor, found matching the prompt mattered more than any other page-level factor.
- Change the date only when the page changes, and keep the visible date, schema, and sitemap in agreement.
How much newer are pages that AI tools cite?
Newer by a measurable margin, mostly in when they were last updated. Ahrefs found AI-cited pages about 25.7% newer than Google's organic results, and Seer found 75% of cited pages updated within a year.
Ahrefs looked at 16.975 million cited URLs in a study Ryan Law published on July 28, 2025. Cited URLs averaged 1,064 days since publication, against 1,432 days for organic results. Its text and table differ slightly on some figures.
Seer Interactive's study by Sonny Vasquez, published July 24, 2026, is newer and narrower: 47,097 citations from March to June 2026, across four brands in four categories. It kept pages cited three or more times and could date about two-thirds of them, 7,683 pages. Neither study changed a page and watched what happened next.
Is it the publish date or the update date that counts?
In Seer's data, the update date. In a subset of 4,124 pages, 72% looked fresh by their last update, but only 42% had been published within the past year.
More than a quarter of the pages Seer counted as fresh had first been published over two years ago. They were old by birth, recent by edit, and still cited.
Ahrefs measured days since publication, so quoting 75% beside 25.7% as one finding merges two measures. By page type, Seer found only 45% of cited news and editorial pages updated within a year, against 67% for blogs and guides. We read that as publishing habit: a news story usually stays as filed.
| Measure | Seer Interactive, 2026 | Ahrefs, 2025 |
|---|---|---|
| What it counted | 47,097 citations, 7,683 dated pages | 16.975 million cited URLs |
| Which date | Last update | Days since publication |
| Headline finding | 75% updated within a year | 25.7% fresher than organic results |
Do ChatGPT, Gemini, and Perplexity differ?
Somewhat, and the two studies disagree on which engine leans hardest. Seer's highest share was on Gemini, while Ahrefs found ChatGPT furthest ahead of organic results.
Cited pages updated within a year, per Seer: 78% for Gemini, 73% for ChatGPT, and 65% for Perplexity. Ahrefs reported ChatGPT citing URLs 393 days newer than organic results by one count (references) and 458 days newer by another (citations). Google AI Overviews went the other way, about 16 days older.
One study measured update dates for four brands and the other publish dates for millions of URLs, so we would not crown either engine. Our post on how ChatGPT and AI Overviews pick sources compares them.
Does page age matter more than matching the question?
The match looks like it matters more. Discovered Labs, an AEO services vendor, ranked prompt and content alignment first among the page-level drivers it measured.
Discovered Labs' research on AI citations, last updated August 2026, covers 2 million citations across 10,000 pages and puts that alignment at a standardised coefficient of +0.37. It reports median cited-content age from 5.1 months on Claude to 8.0 months on ChatGPT. That is vendor data, so we read it as one input.
We would fix the match first, because a refreshed page that answers last year's question keeps its problem. Our post on turning customer questions into answers covers finding this year's wording.
None of these studies can answer one question. Pages get updated because someone owns them, and owned pages also tend to be accurate. Is the date doing any of the work, or is it a trace left by the care? We have not found a test that separates the two.
How should you date a page you update?
Change the date when you change the page, show it, and make the structured data agree. A new date on an unchanged page is not what these studies measured.
Google's guidance on publication dates asks for a visible date, matched by datePublished and dateModified in structured data. It says Google weighs several signals to estimate when a page was "published or significantly updated." Our post on schema markup and AI citations covers the markup.
Sitemaps follow the same logic. Barry Schwartz at Search Engine Roundtable reported on July 16, 2026 that Google's Gary Illyes replied on Bluesky that a site with wrong sitemap dates is "probably better off without the lastmods."
Alongside Corvan AI, we run the AI search work for clients, and dating updates this way is part of it. Nobody can promise a citation from a refresh, us included.
Take a pricing page, or a comparison page like the ones in our post on comparison pages. When a number changes, put the new figure in, add a visible "Last updated" line, and let dateModified and the sitemap's lastmod move with it.
Where the facts in this piece come from
- Seer Interactive, study of content recency and AI visibility in 2026 (Sonny Vasquez, July 24, 2026)
- Ahrefs, study of whether AI assistants prefer to cite fresh content (Ryan Law, July 28, 2025)
- Discovered Labs, research on what drives AI citations (AEO services vendor, last updated August 2026) (vendor data)
- Google Search Central, guidance on publication dates (undated)
- Search Engine Roundtable, report on Gary Illyes and incorrect lastmod dates (Barry Schwartz, July 16, 2026)
FAQs about Do AI tools cite newer pages more than older ones
1. Should I update old blog posts for AI search?
Update the ones whose facts or numbers changed, and date the update honestly. Seer found a pattern in cited pages, not proof that updates caused citations.
2. Does changing the date on a page make AI tools cite it?
None of the studies we cite tested that. Google weighs several signals to judge when a page was significantly updated, so a new date on unchanged content has little behind it.
3. How much fresher are AI citations than Google results?
About 25.7% fresher by publish date, in Ahrefs' study of 16.975 million cited URLs: 1,064 days old on average against 1,432.
About Corvan
Corvan makes Corvan AI, AI agents that run your marketing: they plan your campaigns, write every post in your voice, and show you which posts booked meetings. The team also runs go-to-market services: consultation, ad operations, and AI search ranking. Every piece under Resources is written by the team that runs the work, with sources. Meet the team on the team page, or read what it returned for Drive Me Barcelona.
Keep reading

Can you block GPTBot and still show up in ChatGPT?
GPTBot only gathers training data, while OAI-SearchBot decides whether ChatGPT search can show your pages, so you can block the first and still be found.
Read more →
How often do AI answers change which brands they name?
AI answers rarely name the same brands twice, so treat each one as a sample and track how often you appear across hundreds of runs, compared by quarter.
Read more →
Do buyers ask AI assistants longer questions than Google?
Google AI Mode queries average 12 words against 4 for classic search. ChatGPT prompt length depends on whether it searches, and its own searches run 12 words.
Read more →