Short answer
Google’s first official guide to generative AI search, published 15 May 2026, states that structured data is not required for AI Overviews or AI Mode, and that no special schema markup is needed to appear in them. Schema does not unlock AI citation. It still helps Google understand your pages, which is a different and smaller job.
This contradicts a lot of what is being sold. Claims that a particular schema type makes you several times more likely to be cited are everywhere.
None of those multipliers are confirmed by Google or by any AI vendor. Google’s own guidance says the opposite.
What changed in 2026
| Date | Change |
|---|---|
| 7 May 2026 | FAQ rich results stop appearing in Google Search |
| 15 May 2026 | Google’s AI search guide: no special schema needed for AI features |
| June 2026 | FAQ report and Rich Results Test support removed |
| 15 June 2026 | FAQ structured data documentation removed entirely |
| August 2026 | FAQ support removed from the Search Console API |
The FAQ dropdown had already been restricted to government and health sites in 2023. May 2026 ended it for everyone.
Do I need to remove FAQ schema?
No. Google has said you do not need to remove FAQPage markup, and that structured data which is not being used does not cause problems for Search.
I use it on many pages on this site and have left it in place. It no longer produces a visible result. It does no harm either.
The FAQ content itself still matters. Question headings with direct answers are easy for any system to read. That value was always in the text, not the markup.
What schema still does
It helps Google understand what a page is and who is behind it. That is real, just modest.
- Entity clarity. Organisation and Person markup connect your site, your name and your profiles elsewhere.
- Remaining rich results. Some types still produce visible features, such as product and review markup.
- Breadcrumbs that show your site structure in results.
- Disambiguation. Telling Google which of two similar things you mean.
Keep the schema that does one of these jobs. Do not add schema expecting it to buy AI visibility.
What actually earns AI citation
The same things that earn rankings, because AI features draw on pages Google already ranks.
- Being indexed and ranking for the question in the first place.
- A direct answer in the first sentence under a clear heading.
- Specific, accurate information. A figure nobody else publishes is quotable.
- Content in the HTML, not loaded later by a script.
- Topical depth across connected pages, especially for AI Mode.
How the two Google surfaces differ is covered in AI Mode vs AI Overviews.
How to spot a bad schema claim
Three questions settle most of them.
- Does it cite Google or the AI vendor directly? Or a study you cannot check?
- Does it separate correlation from cause? Well-structured sites tend to use schema and also tend to be cited. That does not mean the schema caused the citation.
- Does it contradict Google’s 15 May guidance? If so, it needs very strong evidence.
Question two catches most of them. Sites that bother with schema usually bother with everything else too.
What to do
- Leave existing FAQ schema alone. No action needed.
- Keep entity schema accurate. Organisation and Person markup pointing at real profiles.
- Stop paying for schema as an AI tactic. Spend it on content and access instead.
- Check AI crawler access for platforms outside Google, covered in how to get cited by ChatGPT, Perplexity and Claude.
Structured data still worth maintaining
- Organization or Person, with sameAs links to real profiles. This helps Google connect your site to who you are.
- Article, with a named author and accurate dates.
- BreadcrumbList, which still shapes how your URL path appears in desktop results.
- Product and Review, on pages that genuinely sell or review something.
- LocalBusiness, for businesses serving a physical area.
How to check it
- Google’s Rich Results Test for the types that still produce rich results.
- The Schema Markup Validator for the general validity of everything else.
- Search Console’s enhancement reports for errors across the whole site.
Accuracy beats coverage. Markup that contradicts the page, such as an author who never appears or a rating nobody gave, does more harm than having no markup at all.
Remove markup only when it is wrong. Inaccurate structured data is worth fixing or deleting. Accurate markup that simply no longer produces a rich result can stay, because Google says unused structured data causes no problems.
Review it whenever the page changes. Markup that described an older version of a page is one of the most common sources of quiet errors.
More claims like this one, each checked against what Google has actually said, are collected in Google SEO myths.
Common markup mistakes after a redesign
- Two Organization blocks, one from the theme and one from a plugin, with different details.
- An Article author who does not match the byline people can see.
- A modified date that never changes, or changes on every save.
- Product markup on pages that do not sell anything.
Run the Schema Markup Validator on one page of each template after any redesign. Template errors repeat across every page that uses them.
Paying for schema to win AI visibility?
Google’s own guidance says it is not required. The budget usually does more on content and crawler access. Send your Search Console access and within 48 hours I will tell you where it is better spent.
Get free recovery audit