Publication date and freshness: how AI measures content aging
Why freshness is an AI ranking factor
When an AI model needs to answer a question, it faces a choice: which sources are most relevant and trustworthy? One of the factors influencing this choice is content freshness. An article about best SEO practices from 2019 probably contains outdated information, while an article from 2026 reflects the latest developments. AI models are trained to make this distinction.
Google already introduced freshness as a ranking factor in 2011 with the "Freshness Update". For AI models, freshness has become even more important. Language models that select sources for generating answers prefer recent sources for time-sensitive topics. This means correctly communicating your publication and modification date directly impacts your visibility in AI answers. This is particularly true for the way AI Overviews and SGE select sources.
Freshness is not equally relevant for every type of content. Evergreen content about fundamental concepts does not need to be recent to be valuable. But for content about technology, trends and current developments, a recent date is a strong signal of relevance.
Communicating publication date via structured data
The most reliable way to communicate your publication and modification date to AI models and search engines is via Schema.org structured data in JSON-LD format. The Article schema offers specific properties for datePublished and dateModified.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Publication date and freshness: how AI measures content aging",
"datePublished": "2026-04-22T10:00:00+02:00",
"dateModified": "2026-04-22T14:30:00+02:00",
"author": {
"@type": "Person",
"name": "Reinier Korthals Altes",
"url": "https://aeo-expert.nl/team/reinier"
},
"publisher": {
"@type": "Organization",
"name": "AEO Expert",
"url": "https://aeo-expert.nl",
"logo": {
"@type": "ImageObject",
"url": "https://aeo-expert.nl/images/logo.png"
}
},
"mainEntityOfPage": {
"@type": "WebPage",
"@id": "https://aeo-expert.nl/blog/publication-date-freshness"
}
}
</script>Pay attention to the date format. Schema.org expects the ISO 8601 format: YYYY-MM-DDThh:mm:ssTZD. The timezone designation (TZD) is important: use +02:00 for Dutch summer time or +01:00 for winter time. A date without timezone can be misinterpreted by systems in other time zones.
The difference between datePublished and dateModified
These two dates serve different purposes and both must be set correctly.
- datePublished is the date the article was first published. This date never changes, even if you update the article later.
- dateModified is the date the article was last substantially changed. Only update this for significant changes, not for fixing a typo.
- If dateModified is missing, AI models assume the content has not been updated since publication.
- A dateModified that is significantly more recent than datePublished signals that the content is actively maintained.
Dive deeper: Canonical URLs and duplicate prevention | Schema.org markup for AI | E-E-A-T optimization
HTML meta tags for dates
In addition to JSON-LD structured data, you can also communicate publication dates via HTML meta tags. This provides an extra signal that parsers and crawlers can pick up, even if they do not process the JSON-LD.
<head>
<!-- Dublin Core meta tags -->
<meta name="DC.date.issued" content="2026-04-22" />
<meta name="DC.date.modified" content="2026-04-22" />
<!-- Article meta tags (used by Open Graph) -->
<meta property="article:published_time" content="2026-04-22T10:00:00+02:00" />
<meta property="article:modified_time" content="2026-04-22T14:30:00+02:00" />
<!-- Parsely meta tag (used by some aggregators) -->
<meta name="parsely-pub-date" content="2026-04-22T10:00:00+02:00" />
</head>It is a best practice to communicate the same dates consistently across all channels: JSON-LD, Open Graph meta tags and optionally Dublin Core. Inconsistent dates across these sources can cause confusion for crawlers and AI models.
Freshness and canonical URLs
There is a direct relationship between freshness signals and your canonical URL configuration. When you update an article and adjust the dateModified, this change must occur on the canonical URL. If AI models encounter a non-canonical version with a newer date, this can produce confusing signals. Ensure the canonical version always communicates the most current date.
Visible dates on the page
In addition to machine-readable dates in the source code, it is essential to also visibly display publication and modification dates to visitors. This strengthens trust with human readers and confirms the machine-readable data.
<!-- Visible date with machine-readable time tag -->
<article>
<header>
<h1>Publication date and freshness</h1>
<p class="meta">
Published on
<time datetime="2026-04-22T10:00:00+02:00">April 22, 2026</time>
| Last updated on
<time datetime="2026-04-22T14:30:00+02:00">April 22, 2026</time>
</p>
</header>
</article>The HTML time element
The time element is specifically designed to make dates and times machine-readable within visible content. The datetime attribute contains the machine-readable version in ISO 8601 format, while the text in between shows the human-readable version. AI crawlers and search engines recognize this element and use it as additional confirmation of the date information.
Freshness and E-E-A-T: the connection
Freshness is closely connected to E-E-A-T signals. A website that regularly updates and dates its content demonstrates active maintenance and engagement with the field. This is a strong expertise signal. Conversely, a website full of articles with dates from three years ago suggests the owner no longer actively follows the subject. AI models interpret this as a lower degree of expertise.
Freshness and AI Overviews
In Google AI Overviews, freshness plays a prominent role. When Google generates an AI Overview for a time-sensitive query, it preferentially selects sources with recent dates. If your article has no recognizable date, or if the date is outdated, you risk not being selected as a source, even if your content is strong in substance.
Common mistakes in date implementation
- The date in structured data differs from the visible date on the page. Ensure complete consistency.
- dateModified is automatically updated on every save in the CMS, even when there is no content change. This misleads search engines and AI models.
- Showing only a year instead of a full date. "2026" is much less informative than "April 22, 2026".
- Omitting the publication date entirely out of fear that content will be seen as "old". The absence of a date is a bigger problem than an older date.
- Dates in an unrecognizable format like "22/04/26" instead of ISO 8601 in the structured data.
- Abusing date modification by changing the publication date to today without actually updating the content. Search engines detect this and it can lead to penalties.
Freshness strategy for your content
A thoughtful freshness strategy combines correct technical implementation with content maintenance. The goal is not to manipulate dates, but to keep content current and communicate this honestly. Freshness is one of the many AEO factors that together determine how visible you are in AI-generated answers.
- Schedule regular content audits. Check every quarter whether your most important articles are still up to date.
- Only update dateModified when you make substantial content changes.
- Add a "Last updated" date to the beginning of long-running articles so readers can immediately see how current the information is.
- Remove or redirect outdated content that is no longer relevant instead of leaving it with an old date.
- Combine a content update with a note about what was changed to provide transparency.
Communicating dates through multiple channels
For maximum reliability, communicate the same date through four channels simultaneously. AI models and search engines trust a date more when it is consistent across multiple sources.
<head>
<!-- Channel 1: Schema.org JSON-LD -->
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Article",
"datePublished": "2026-04-22T10:00:00+02:00",
"dateModified": "2026-04-22T14:30:00+02:00"
}
</script>
<!-- Channel 2: Open Graph -->
<meta property="article:published_time" content="2026-04-22T10:00:00+02:00" />
<meta property="article:modified_time" content="2026-04-22T14:30:00+02:00" />
<!-- Channel 3: Dublin Core -->
<meta name="DC.date.issued" content="2026-04-22" />
<meta name="DC.date.modified" content="2026-04-22" />
</head>
<!-- Channel 4: Visible on the page with time element -->
<time datetime="2026-04-22T10:00:00+02:00">April 22, 2026</time>In our AEO scanner, we check whether your page has a recognizable publication date, both in structured data and visibly on the page. A missing or inconsistent date is flagged as an improvement area.
Key takeaways
- Freshness is an active ranking factor for AI models on time-sensitive topics.
- Communicate publication and modification dates via Schema.org JSON-LD, Open Graph meta tags and visibly on the page.
- Ensure complete consistency between all date sources: JSON-LD, OG tags, Dublin Core and visible date.
- Only update dateModified for substantial content changes, not for cosmetic adjustments.
- A thoughtful freshness strategy combines technical implementation with regular content audits.
Frequently asked questions
Should I always show the publication date on my articles?
Yes, for most content showing a publication date is a best practice. It increases reader trust and gives AI models a crucial signal about freshness. The only exception is content that is deliberately timeless, such as an "About us" page. But even there, a "Last updated" date is valuable.
How often should I update my content to stay "fresh"?
This depends on the type of content. Technical guides about rapidly changing topics (like AI tools) deserve a quarterly review. Fundamental how-to articles can be checked annually. It is not about frequency but relevance: only update when something has actually changed. Artificial updates without content changes are counterproductive.
What if my CMS automatically updates dateModified on every save?
This is a common problem. If your CMS updates dateModified on every save, including cosmetic changes like fixing a typo, AI models lose trust in your date signals. The solution is to decouple dateModified from the CMS "updated_at" timestamp and use a separate field that is only manually updated for substantial changes.
Does freshness weigh equally for all types of queries?
No. For time-sensitive queries ("best AI tools 2026"), freshness is very important. For evergreen queries ("what is HTTP"), content quality weighs more heavily than freshness. AI models are smart enough to make this distinction. Focus your freshness strategy on content that is inherently time-sensitive.
Can a publication date that is too old exclude my content from AI answers?
An old publication date does not automatically exclude you, but significantly reduces your chances for time-sensitive topics. An article about "HSTS configuration" from 2019 is technically still largely correct, but AI models prefer a recent article offering the same information. For timeless topics the impact is smaller, but a recent dateModified always helps.
Content without a date is like a news story without a headline: it lacks the context that determines how relevant the information is.
How does your website score on AI readiness?
Get your AEO score within 30 seconds and discover what you can improve.
