Schema markup for AI answers: what still matters

What is schema markup and why does it still matter?

Schema markup is structured data, usually written as JSON-LD in a script tag, that states facts about a page in a vocabulary search engines share: this is an article, written by this person, published on this date, about this organisation. It removes guesswork about entities, which is useful for search features, for knowledge panels and for any system that reads your HTML.

What it is not, on current evidence, is a citation lever. A 2026 Ahrefs matched-control test summarised by Geoptimizer found that adding schema moved AI citations by +2.4% in AI Mode and +2.2% in ChatGPT, both indistinguishable from zero, and −4.6% in AI Overviews. Google’s May 2026 guide says no special schema is required for its generative AI features.

So keep schema for the jobs it does well: making authorship, dates and entities unambiguous, supporting the search features that still use it, and keeping your site’s facts consistent across templates. Measure AI visibility with other tools, because adding markup will not show up there.

What changed for FAQ schema in 2026?

Google had limited FAQ rich results to authoritative government and health sites in 2023. On 7 May 2026 the FAQ rich result stopped appearing in Google Search altogether, the deprecation was documented on 8 May, and the supporting documentation was removed in June.

FAQPage is still a valid Schema.org type and Google says unused structured data does not cause problems. Keep it if the questions and answers are visible on the page and genuinely useful, because a clear question-and-answer block is still easy for any reader, human or machine, to quote. Just do not expect a rich result from it.

The lesson from the FAQ change is broader than one type. Rich results come and go as search engines change their pages, so markup added only to win a visual feature has a short life. Markup that describes your content accurately keeps its value whatever the results page looks like next year.

Which schema types should a marketing site use?

Use the few types that describe your real content accurately. More types do not mean more visibility; wrong or invisible claims can count against you under Google’s structured data guidelines.

How do you write Person and ProfilePage markup?

Google supports ProfilePage structured data for pages about a creator or person. The required property is mainEntity, a Person or Organization, with a name; recommended properties include sameAs links to the person’s other profiles, an image and a description. For a portfolio or author page, this is the clearest way to tie the page, the person and their LinkedIn or company profiles into one entity.

Inside the Person object, add jobTitle, url and a short list of knowsAbout topics that match the content on the site. Keep every value consistent with the visible page and with your other profiles; consistency across sources is what lets engines connect the dots.

What does good Article markup look like on a blog post?

An Article or BlogPosting block should repeat what the reader already sees: the headline exactly as it appears in the H1, the author as a Person with a url pointing to their profile page, datePublished and dateModified in ISO format, the main image, and the publisher as an Organization or the same Person for a personal site. Add a BreadcrumbList so the post’s place in the site is explicit.

Generate the block from the same data that renders the page rather than writing it by hand. When the title or date changes in one place and not the other, the markup starts making claims the page does not support, which is the most common way structured data goes wrong on content sites.

• headline: the visible H1, word for word.
• author: a Person with name and url.
• datePublished and dateModified: ISO dates that match the page.
• image: the hero image URL, at a useful size.
• mainEntityOfPage: the canonical URL.

How do you test structured data before it ships?

Test every template, not every page. Run one example of each page type through a schema validator and Google’s Rich Results Test, fix any errors, then check a sample of live pages after each release. Watch Search Console’s enhancement reports for new errors, especially after template changes.

Also check that the JSON parses at all. A single stray quote in a title can break the whole block, so escape values properly or build the object in code and serialise it, instead of concatenating strings in a template.

What mistakes should you avoid?

The free schema generator on this site writes valid JSON-LD for Article, Person, Organization and FAQPage from a short form, so you can paste it into a template and validate it in one pass.

• Marking up content that is not visible on the page.
• Copying one block of FAQ markup onto many pages.
• Inventing ratings, reviews or credentials; fabricated author details now sit under Google’s deception policies.
• Letting dates go stale: update dateModified only when the content changes.
• Shipping invalid JSON; test every template with a validator before it goes live.