Articles are dated
What we check
On the sampled pages that are articles, we look for a date the machine can read: a datePublished or dateModified in the page’s structured data, a <time datetime> element, an article:published_time tag, or a <meta name="date">. The score is the share of articles that have one. If no sampled page is an article, we skip the check.
A page counts as an article if your site labels it as one in its structured data. If your site labels none, we fall back on the page itself: a single <article> in the content, or more than 400 words. We also check the two newest posts from your feed. A date counts only if it has a full four-digit year and is no more than a day in the future, so a bare time like 10:00 doesn’t count. The detail names the newest date we found and the page it was on.
Why it matters
An AI assistant weighing two sources wants to know which is newer, and whether advice still holds. An undated article is harder to trust and harder to place: is this from last week or ten years ago?
A date people can see isn’t enough on its own. “Last Tuesday” or a date in an image means nothing to a machine; it needs the date in a form it can read.
How to pass
- Put the publish date (and the last-updated date, if it changes) in each article’s structured data:
<script type="application/ld+json"> { "@context": "https://schema.org", "@type": "BlogPosting", "headline": "How we cut our prices", "datePublished": "2026-10-01", "dateModified": "2026-10-14" } </script> - And show it by the byline, marked up so it can be read:
<time datetime="2026-10-01">October 1, 2026</time>
Most blog platforms do both; check that the theme you use hasn’t removed them.