Schema Markup for AI Search

Technical

Last updated September 17, 2026

Schema markup is structured data you add to a page so a machine can read its facts without inferring them from prose. It has been useful for search for a decade. For AI search it does something slightly different, and it is worth being precise about what.

The types that earn their keep

There are hundreds of schema types. Four cover almost all of the useful cases for a product or documentation site.

Organization

Who you are, once, canonically. Name, URL, logo, and the social or reference profiles that describe the same entity elsewhere. This is the anchor for everything in entity SEO, and if you add one thing, add this.

Article

For anything written: headline, author, publish date, modified date. The dates matter more than people expect, because a model handling a time-sensitive question has to choose between sources and will prefer one that states when it was written.

FAQPage

Question and answer pairs. The rule that trips people up: every question and answer must be visible on the page. Marking up a FAQ the reader cannot see is against the guidelines and is the fastest way to lose the enhancement.

Product / Offer

Price, currency, availability. If you sell something, this is where a model gets the number right rather than reading it off a comparison table it half-understood.

The rule that matters most

Schema must describe what is actually on the page. Not what you would like to be true, not a superset, not a rating you have not received.

This is not only a guidelines question. A model that finds structured data contradicting the visible text has learned that your markup is unreliable, and it has no way to tell which half to trust. You are better off with no schema than with schema that disagrees with your own page.

Adding it in WordPress

Most SEO plugins emit Organization and Article schema already, and some emit FAQPage from a block. Check what you have before adding more, because two plugins each emitting an Organization block gives a parser two competing answers.

For anything custom, a small wp_head hook is clearer than a plugin:

add_action('wp_head', function () {
    if (! is_singular('post')) {
        return;
    }

    $schema = [
        '@context'      => 'https://schema.org',
        '@type'         => 'Article',
        'headline'      => get_the_title(),
        'datePublished' => get_the_date('c'),
        'dateModified'  => get_the_modified_date('c'),
        'author'        => [
            '@type' => 'Organization',
            'name'  => get_bloginfo('name'),
        ],
    ];

    echo '<script type="application/ld+json">'
        . wp_json_encode($schema)
        . '</script>';
});

Note what that does not do: it does not hardcode the headline or the dates. Every value is read from the post. A schema block with literals in it is a schema block that will eventually describe a different page than the one it is on.

The knowledgebase you are reading takes this further for FAQs. Rather than accepting a JSON-LD block written by hand, the theme reads the rendered FAQ out of the article and generates the FAQPage schema from it. The two cannot disagree, because one is derived from the other.

Validating it

• Google’s Rich Results Test, for whether an enhancement is eligible.
• Schema.org’s own validator, for whether the markup is structurally valid.
• Your own eyes, for whether it matches the page. No tool checks that, and it is the failure that actually happens.

Where to go next

Entity SEO for WordPress is what Organization schema is in service of.

Did this answer it? Tell us if not and we will fix the article.