The Dateless Third: A Third of What Google’s AI Cites Has No Publication Date

Every piece of AI-search advice tells you to keep your content fresh and show your update dates. We wanted to know what share of AI-cited content even participates in that system. So on July 28, 2026 we harvested every source cited by Google’s AI Overviews across 30 blogging and SEO queries, then tried to establish a publish or update date for all 224 unique pages. A third of them defeated us, and the reason turned out to be more interesting than the failure.

Key data points:

• 33.3 percent of AI Overview citations (81 of 243) have no findable date, versus 29.1 percent of organic top-10 results for the same queries.
• The dateless pages are not articles. Of the 81 undated citations, 48 were product or landing pages and 22 were videos. Only 6 were editorial articles.
• Just 4.4 percent of cited blog articles lack a date. Publishers overwhelmingly do date their content. Software companies and platforms do not.
• Transactional queries are a dateless world: 55.7 percent of citations there carry no date, and 36 of those 44 are product pages. Informational queries: 20.8 percent. Commercial: 23.9 percent.
• Google’s AI cites video roughly seven times more than the organic top 10 does (9.1 percent of AI citations vs 1.2 percent of organic results), and almost none of it is dated in a way a crawler can read.
• On 6 of 29 queries, more than half the AI Overview’s sources were undated. For “buy a domain name for a blog”, all 7 citations were dateless. Five of those six queries were transactional.

How we ran this study

This analysis shares its corpus with our content age study of AI Overview citations. We ran 30 blogging-niche Google searches (10 informational, 10 commercial, 10 transactional, US settings) on July 28, 2026, expanded each AI Overview in full, and captured every external source it cited plus the organic top 10 for the same query. We then attempted to date every page from schema.org datePublished and dateModified markup, article meta tags, visible on-page dates, and platform timestamps. A page counts as dateless only when all four methods came up empty.

Who exactly is undated?

The composition matters more than the headline number. Breaking the 81 undated citations down by page type:

• Product and landing pages: 48. Registrar pages, hosting plans, SaaS homepages, pricing pages, marketplace category pages. These almost never carry a publish date, because conceptually they do not have one.
• Videos: 22. YouTube results surfaced in the AI Overview without a crawler-readable timestamp attached to the citation.
• Editorial articles: 6. Out of 137 cited articles total. A 4.4 percent failure rate.
• Q and A pages: 5. Quora and similar.

So the dateless third is not evidence that Google’s AI is indifferent to freshness. It is evidence that a third of the citation pool sits outside the publish-and-update model entirely. Those pages are not stale; they have no age to be stale relative to.

Why this changes the competitive picture

Most freshness advice implicitly assumes your competition is other blog posts, ageing at the same rate you are, so a recent update date moves you up the queue. On transactional queries that assumption collapses. When 55.7 percent of what the AI cites is undated product pages, your real competitor is a registrar’s pricing page that will never look stale and never needs refreshing. No refresh cadence wins that fight, because you are not playing the same game.

This is the flip side of the staleness penalty we measured earlier: rot gets you dropped, but the escape from rot only helps in the part of the pool where dates exist. On informational queries, where only 20.8 percent of citations are undated and articles dominate, maintenance is the whole game. On transactional queries it barely registers.

The video gap nobody is optimising for

One number surprised us most. Video made up 9.1 percent of AI Overview citations but only 1.2 percent of organic top-10 results across the same 30 queries. Google’s AI reaches for video roughly seven times more readily than its own organic ranking does.

If you write and never film, this is a channel where the AI is actively looking for sources and most text publishers are not present. It is a bigger structural gap than any date field.

What this means for your blog

• Segment your freshness effort by intent. Refresh cadence pays on informational and commercial content, where dated articles are the competition. On transactional terms, put that effort into a genuinely better page instead, because your rivals there are ageless by design.
• Date your content anyway. Only 4.4 percent of cited articles lack a date, which means dating is table stakes among publishers, not an advantage. Missing it puts you in a category of one.
• Treat video as an uncontested citation surface. A seven-to-one over-representation versus organic is the widest gap in this dataset.
• Stop benchmarking yourself against the whole citation list. On a transactional query, most of the sources are not competing on the axes you can control. Benchmark against the dated editorial subset.

To check whether a specific post is structured to be quotable by an answer engine at all, run it through our free AI Citation Grader.

Frequently asked questions

What share of AI Overview citations have no publication date?

33.3 percent, based on 243 citations across 30 queries harvested July 28, 2026. The comparable figure for organic top-10 results on the same queries is 29.1 percent. Undated pages are overwhelmingly product and landing pages (48) and videos (22) rather than articles (6).

Does that mean Google’s AI ignores content freshness?

No. It means a third of the citation pool has no date to evaluate. Within the dated portion, freshness signals still operate, and our separate research found a staleness penalty at the two-year-plus end. The dateless third simply sits outside that system.

Why are transactional queries so different?

Because the AI cites commercial destinations there rather than editorial coverage. On transactional queries 55.7 percent of citations were undated, and 36 of those 44 were product pages. Registrars, hosts, and SaaS pricing pages do not carry publish dates.

Should I add a date to my product and landing pages?

There is no evidence in our data that it helps, and the uncited majority of product pages look identical to the cited ones on this axis. We would not treat it as a lever. For editorial pages a date is standard practice, and 95.6 percent of cited articles have one.

How big was the sample and can I reuse the data?

30 queries, 29 with an AI Overview, 243 citation events across 224 unique pages, plus 254 organic top-10 results, harvested July 28, 2026 with US settings. Yes: the dataset summary below is published under CC BY 4.0, free to cite with attribution.

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

Related research from the same corpus: how old the content Google’s AI cites really is, who actually gets cited: brands vs blogs vs forums, and the citation-rank disconnect.

@misc{bloggingtitan2026datelessthird,
  title  = {The Dateless Third: A Third of What Google's AI Cites Has No Publication Date},
  author = {{Blogging Titan}},
  year   = {2026},
  url    = {https://bloggingtitan.com/blog-seo/ai-overview-dateless-citations-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.