Content Freshness — datePublished & dateModified for AI

AI systems actively use publication and modification dates to rank content. A page with an explicit, recent dateModified is preferred over undated content when users ask time-sensitive questions.

TL;DR: Add datePublished and dateModified in your Schema.org JSON-LD and in <meta> tags. Update dateModified every time you make a substantive content change. Use ISO 8601 format (YYYY-MM-DD or full datetime).

Why do AI systems prefer fresh content?

AI retrieval systems — including the indexers behind Perplexity, ChatGPT with browsing, and Google's AI Overviews — use content freshness as a ranking signal for queries where recency matters. When a user asks "What is the current best practice for X?", the system favours pages with a recent dateModified over pages that carry no date at all. Undated content is treated as potentially stale and ranked lower in time-sensitive contexts.

How does datePublished work in Schema.org?

datePublished is a property of the Schema.org CreativeWork type (and its subtypes: Article, BlogPosting, NewsArticle). It tells AI crawlers and search engines when the content was originally created. Once set, this value should never be changed retroactively — doing so is treated as a freshness manipulation signal.

How does dateModified work in Schema.org?

dateModified is a property of the Schema.org CreativeWork type that records the most recent substantive update. Unlike datePublished, this value should be updated every time you revise the content. Trivial changes (fixing a typo, updating CSS) do not warrant a dateModified bump — substantive changes (new section, updated facts, revised statistics) do.

What counts as a substantive update?

A substantive update is one that changes the informational value of the page for a reader. Examples include: adding or replacing a code snippet, updating statistics or prices, adding a new section, revising a recommendation based on new information, or correcting a factual error. Formatting changes, typo fixes, and visual redesigns are not substantive updates.

How to signal updates with HTML meta tags

In addition to JSON-LD, you can provide date signals via standard <meta> tags. These are read by some AI crawlers that do not parse JSON-LD, and by HTTP headers when a full fetch is not performed.

<meta name="article:published_time" content="..."> — Open Graph article published time.
<meta name="article:modified_time" content="..."> — Open Graph article modified time.
<meta http-equiv="last-modified" content="..."> — HTTP last-modified equivalent (less commonly parsed).

The JSON-LD approach is more reliable and should be treated as the primary signal; HTML meta tags provide a useful fallback.

Step-by-step — implement content freshness signals

Add a datePublished value to your Article JSON-LD on every page that has been published.
Add a dateModified value set to the same date as datePublished when first publishing.
Create a process (a CMS field, a deploy script, or a commit hook) that updates dateModified whenever content changes.
Confirm the dates are visible in the rendered HTML — not just in a JavaScript variable.
Validate with Google's Rich Results Test to ensure the dates are parsed correctly.
Optionally add a human-readable "Last updated: [date]" line to the page for user trust (AI systems also parse visible dates in addition to structured data).

Code example — JSON-LD with datePublished and dateModified

<!-- JSON-LD structured data: Article with freshness signals -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your Article Title",
  "datePublished": "2026-04-16",
  "dateModified": "2026-04-16",
  "author": {
    "@type": "Person",
    "name": "Your Name"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Your Organisation",
    "url": "https://example.com"
  },
  "url": "https://example.com/your-article",
  "mainEntityOfPage": "https://example.com/your-article"
}
</script>

<!-- HTML meta tag alternatives (fallback for crawlers that skip JSON-LD) -->
<meta property="article:published_time" content="2026-04-16T00:00:00Z">
<meta property="article:modified_time" content="2026-04-16T00:00:00Z">

<!-- Optional: human-readable date on the page itself -->
<p class="byline">
  Published: <time datetime="2026-04-16">16 April 2026</time> ·
  Updated: <time datetime="2026-04-16">16 April 2026</time>
</p>

Common mistakes to avoid

No dates at all: Undated content is deprioritised for any time-sensitive query — always include at least datePublished.
Backdating datePublished: Setting a publish date earlier than the actual publication is treated as a manipulation signal and can harm ranking.
Bumping dateModified without real changes: AI crawlers compare date signals against sitemap lastmod and HTTP Last-Modified headers — inconsistencies are detected and penalised.
Using non-ISO date formats: Dates like "April 16, 2026" or "16/04/2026" may not be parsed correctly; always use YYYY-MM-DD or YYYY-MM-DDTHH:MM:SSZ.
Dates only in JavaScript: If your dates are injected by JavaScript and the AI crawler does not execute JS, the dates will be invisible. Render them server-side.
Inconsistent signals: If your JSON-LD says 2026-01-01 but your sitemap says 2024-06-15, AI crawlers will distrust both signals.

Official sources

Test your website now — free AI Visibility Score Up to 19 checks in under 30 seconds