How Old Is the Content Google’s AI Cites? We Dated 243 AI Overview Citations

Everyone in SEO repeats the same advice: AI search loves fresh content. But nobody seemed to have actually dated the pages Google’s AI cites. So we did. On July 28, 2026 we harvested every source cited by Google’s AI Overviews across 30 blogging-niche queries, then tracked down the original publish date and the last-updated date for every page we could.

Key data points from the study:

• 29 of 30 queries (97 percent) triggered an AI Overview, producing 243 citation events across 224 unique pages.
• The median cited page was originally published 12.8 months before being cited. A full 33.3 percent were first published more than two years ago.
• By last-updated date, the median cited page was only 6.2 months old. 69.1 percent had been published or updated within the last 12 months, and 30.2 percent within the last 90 days.
• 46.9 percent of dateable cited pages had been refreshed after publication (updated more than 30 days after their original publish date). Among cited blog articles specifically, that rises to 54.4 percent.
• One in three AI Overview citations (33.3 percent) points to a page with no findable date at all, mostly product and landing pages (48 citations) and videos (22 citations).
• The oldest editorial citation we found was a blog post first published in 2015, still being cited by Google’s AI in 2026 after years of refreshes. The AI also cited a Reddit thread from 2017.

How we ran this study

We ran 30 Google searches in the blogging and SEO niche, split evenly across informational, commercial, and transactional intent, using US settings (gl=us, hl=en) on July 28, 2026. For each query we expanded the full AI Overview and captured every external source it cited, plus the organic top 10 for comparison (that comparison became its own study, linked below).

We then dated every cited page using, in order of preference: schema.org datePublished and dateModified markup, article meta tags, visible byline and “updated on” dates, and platform timestamps (for example, Reddit thread creation times). We could establish at least one date for 66.7 percent of citation events. The remaining third simply have no date anywhere, which turned out to be a finding in itself.

How old is the content Google’s AI actually cites?

Older than the “freshness” narrative suggests. By original publish date, the median AI-cited page was 12.8 months old, and the mean was 24.2 months, dragged up by a long tail of genuinely old pages.

One third (33.3 percent) of dateable AI Overview citations were originally published more than two years before being cited. Google’s AI happily cited a 2015 email marketing guide, a 2016 hosting roundup, and a 2017 Reddit thread, all in July 2026.

If AI search only wanted new content, none of those pages would appear. Publish date is clearly not the filter.

Why “age” is the wrong number: the refresh dividend

The picture flips when you look at the last-updated date instead. Measured by the most recent touch (update date where one exists, publish date otherwise), the median AI-cited page was just 6.2 months old, half its median publish age.

For cited blog articles the gap is even wider: a median original age of 12.4 months versus a median effective age of 5.3 months. And 54.4 percent of cited articles had been meaningfully refreshed since publication.

That gap between the two numbers is the refresh dividend: old pages that keep getting updated keep getting cited. The pages winning AI citations are mostly not this quarter’s new posts. They are one-to-three-year-old assets that someone keeps maintaining. This is the optimistic flip side of the decay debt problem we documented earlier: decay is real, but the escape route is cheap. You do not need a new post. You need a real update.

Which queries cite the oldest content?

Intent changes the math:

• Informational queries cite the freshest editorial content: median publish age 10.7 months, with 58.2 percent of citations published within the last year.
• Commercial (“best X”) queries cite the oldest content: median publish age 18.5 months, and 43.3 percent of citations were first published over two years ago. But these pages are also the most maintained, with 53.7 percent refreshed after publication. Big affiliate roundups age well precisely because their owners keep re-dating and re-verifying them.
• Transactional queries are a different world entirely, dominated by dateless product pages (more below).

A third of AI citations have no date at all

33.3 percent of AI Overview citations pointed to pages where no publish or update date exists anywhere: not in schema markup, not in meta tags, not visible on the page. These were mostly product and landing pages, plus YouTube results harvested without timestamps.

On transactional queries, only 44.3 percent of citations could be dated at all. Google’s AI clearly does not require a date to cite a page, and for commercial intent it leans heavily on pages that live outside the publish-and-decay cycle entirely.

What this means for your blog

Three practical moves fall straight out of the data:

• Set a refresh cadence of 6 to 12 months for your money pages. The median cited page was touched within 6.2 months. If your best posts have not been updated in a year, they are aging out of the citable pool.
• Make your updates visible to machines. Expose dateModified in your article schema and show an honest “Updated” date on the page. A real update that nobody can detect pays no dividend.
• Update, do not republish. The 2015-vintage pages still earning citations kept their URLs and their accumulated authority, and layered fresh data on top. Deleting and republishing resets that clock.

If you want to know whether your own posts are structured to be cited at all, run them through our free AI Citation Grader.

Frequently asked questions

Does Google’s AI prefer new content?

Not by publish date. In our July 2026 study, the median AI Overview citation was originally published 12.8 months earlier, and 33.3 percent were over two years old. What the cited pages share is recent maintenance: the median page had been updated within 6.2 months.

How often should I update posts to stay citable by AI?

The data points to a 6 to 12 month cycle. 69.1 percent of dateable AI Overview citations had been published or updated within 12 months, and 48.1 percent within 6 months. Pages untouched for more than two years made up under 15 percent of citations.

Does changing the date without changing the content work?

We measured meaningful refreshes, and our definition (an update more than 30 days after publication) cannot distinguish honest updates from date-bumping. But Google has stated repeatedly that it detects unchanged content with new dates, and the consistently cited pages in our corpus showed substantive revisions: new data, new tools, new pricing. Treat the update date as a receipt for real work, not a growth hack.

How was content age measured in this study?

For each cited page we recorded the original publish date and the last-updated date from schema.org markup, article meta tags, visible on-page dates, or platform timestamps, then measured both ages against the harvest date of July 28, 2026. Effective age uses the update date where one exists and the publish date otherwise. Full dataset summary below.

Data collected July 28, 2026. Original research by the Blogging Titan team.

Related research from our AI Overview corpus: the dateless third: a third of AI citations have no date at all, 70.4 percent of AI Overview citations do not rank in the organic top 10, and how many sources AI Overviews actually cite.

@misc{bloggingtitan2026aiocontentage,
  title  = {How Old Is the Content Google's AI Cites? Age Analysis of 243 AI Overview Citations},
  author = {{Blogging Titan}},
  year   = {2026},
  url    = {https://bloggingtitan.com/blog-seo/ai-overview-content-age-study/},
  note   = {Original first-party dataset, CC BY 4.0}
}

Written by

Blogging Titan Team

Blogging Titan is an independent team of bloggers documenting what actually grows a blog in the AI search era. We have been building, ranking, and monetizing WordPress sites since 2017, and every guide on this site is based on strategies and tools we have tested ourselves. Want a second pair of eyes on your blog? Request a free blog audit or start with the 2026 playbook.