TL;DR
- AI engines use Article schema as a trust and recency signal — not a rich-result trigger like FAQ or Recipe schema, so the bar for "correct" is different and less forgiving of shortcuts.
datePublishedanddateModifiedmust be real, consistent, and machine-parseable (ISO 8601) — and they must match what's visible on the page, or crawlers discount the whole block.authorshould be aPersonorOrganizationobject with anurlorsameAs, never a bare string — an unlinked name is just text, not an entity.publisherneeds alogoas anImageObject, andmainEntityOfPageshould point back to the canonical URL — both are commonly skipped and both quietly break validation.- A minimal valid block is below. Copy it, swap the values, validate it, and stop guessing.
Article schema is the JSON-LD block that tells search and AI crawlers what a piece of content is, who wrote it, when it was published, and who owns it. For GEO purposes it matters less for rich snippets and more because LLM-aligned crawlers (and the retrieval layers behind ChatGPT, Perplexity, and Google's AI Overviews) use structured metadata to judge freshness and authorship when deciding what to cite. A page with ambiguous or contradictory dates and an anonymous-looking author is an easy pass for a model weighing two similar sources.
Why Article schema matters more for AI than for Google rankings
Traditional SEO treats Article schema as optional polish — Google rarely shows a special rich result for it outside of AMP-era carousels. But AI answer engines behave differently. When a retrieval system pulls candidate pages for a query like "what changed in the latest iOS update," it needs a fast, structured way to answer two questions: is this current, and is this credible. Visible on-page text can answer both, but it's slow and ambiguous to parse at scale. Structured dates and a named, linkable author answer both instantly.
This is also why Article schema interacts with freshness signals more than most people realize — see the related discussion on stale content and sitewide citation decay. The schema doesn't make content fresh, but it's the fastest way for a crawler to confirm the freshness claim your content makes.
The three mistakes that get you ignored
1. Missing or contradicting dates. The most common failure is a datePublished that doesn't exist, or a dateModified that's older than datePublished, or a schema date that doesn't match the human-readable date printed on the page. Any of these creates a contradiction a parser can detect automatically, and once one field looks fabricated or careless, crawlers have no reason to trust the rest of the block. Dates must be in ISO 8601 format (2025-03-14T09:00:00-04:00), and dateModified should only change when the content actually changes — not on every CMS republish.
2. Author as a plain string. Writing "author": "Jane Doe" is technically valid JSON-LD but semantically useless. It's a label, not an entity. To be treated as a credible byline, author needs to be a Person type object with a name, a url pointing to an author bio page, and ideally a sameAs array linking to LinkedIn, X, or an ORCID/Google Scholar profile. This is the same entity-resolution logic covered in sameAs vs knowsAbout — an AI system can't verify a string, but it can verify a linked entity with a consistent identity across the web. If your CMS defaults to string authors, that's a one-line fix with outsized payoff.
3. Publisher without a logo, and a missing mainEntityOfPage. publisher is required on Article/NewsArticle schema per Schema.org guidance, and it needs a nested logo as an ImageObject with explicit url, width, and height. Many implementations include the publisher name but drop the logo object entirely, which fails Google's structured data guidelines and leaves the entity half-defined. Separately, mainEntityOfPage should be set to the canonical URL of the page as an @id. Skipping it makes the article ambiguous about which URL is authoritative when the same content is syndicated or appears on multiple paths — a problem that compounds with the canonicalization issues discussed in canonical tags and duplicate content.
A minimal valid block
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Article schema done right",
"datePublished": "2025-03-14T09:00:00-04:00",
"dateModified": "2025-03-14T09:00:00-04:00",
"author": {
"@type": "Person",
"name": "Jane Doe",
"url": "https://example.com/author/jane-doe",
"sameAs": ["https://www.linkedin.com/in/janedoe"]
},
"publisher": {
"@type": "Organization",
"name": "Example Co",
"logo": {
"@type": "ImageObject",
"url": "https://example.com/logo.png",
"width": 600,
"height": 60
}
},
"mainEntityOfPage": {
"@type": "WebPage",
"@id": "https://example.com/article-schema-ai"
}
}
This covers the fields that actually get checked: dates, a resolvable author entity, a complete publisher object, and an unambiguous canonical target. Everything beyond this — image, articleSection, wordCount — is additive, not load-bearing.
How to validate and maintain it
Run every template through Google's Rich Results Test and the Schema.org validator before shipping, but don't stop there — spot-check rendered HTML, not just template source, since CMS plugins frequently inject a second, conflicting Article block. Set dateModified to update automatically on substantive content edits (not on every save), and audit author pages quarterly to confirm the sameAs links still resolve. If you publish under multiple bylines or a house account, keep one canonical Person or Organization object per author and reuse it verbatim across every article — inconsistent spelling or URLs across pages recreates the entity-ambiguity problem schema is supposed to solve. This same consistency discipline matters at the organization level too, as covered in Organization schema and the entity backbone.
FAQ
Does Article schema directly cause an AI engine to cite my page?
No — it's a supporting signal, not a ranking factor on its own. It helps a retrieval system quickly confirm authorship and freshness once your content is already a candidate, but it won't compensate for thin or outdated content.
Should I use Article or NewsArticle?
Use NewsArticle for time-sensitive journalism and Article (or BlogPosting) for evergreen or opinion content; the fields required (dates, author, publisher) are nearly identical, so the choice mainly affects how news-specific features interpret the content.
How often should dateModified actually change?
Only when you materially update facts, data, or structure in the piece — not on every CMS autosave or minor typo fix, since inflating dateModified without real changes is itself a pattern crawlers can detect and discount.
