FAQ Schema (FAQPage)
FAQ schema marks up question and answer pairs on a page. Learn the rules, what changed with rich results, and why AI engines still read it.
TL;DR
FAQ schema is the structured data type that marks up question and answer pairs on a page, using the FAQPage type from schema.org. Each entry pairs a question with its accepted answer in a form machines read without parsing prose. Its value shifted when Google narrowed FAQ rich results to a small set of site categories, but the markup still earns its place for a different reason.
Why FAQ schema still matters
The rich result payoff shrank for most sites. The machine-readability payoff did not, and it now serves a larger audience than search alone.
- Answer shape: A question paired with a direct answer is exactly the structure AI engines produce, which makes marked-up pairs easy to lift.
- Explicit boundaries: Markup states where an answer starts and ends, removing the guesswork of extracting it from surrounding prose.
- Voice and assistant surfaces: Question-answer pairs remain a natural fit for spoken responses.
- Internal discipline: Writing markup-worthy FAQs forces genuinely direct answers rather than paragraphs that circle the point.
How to implement FAQ schema
Write real questions first
Use the phrasing people actually search or ask. Invented questions that exist to carry markup read as filler to humans and engines alike.
Answer directly in the first sentence
The answer property should resolve the question immediately, with elaboration after rather than before.
Mirror the visible page
Every marked-up question and answer must appear on the page for users. Hidden FAQ markup is a guidelines violation.
Keep answers self-contained
An answer that depends on the paragraph above it makes no sense when an engine extracts it alone.
Use plain text in answers
Limited HTML is permitted, but complex nested markup inside answer values causes validation failures.
Limit to genuine FAQs
Three to six real questions beat fifteen manufactured ones. Padding dilutes the page without adding eligibility.
FAQ schema vs. HowTo schema
FAQ schema: Independent question and answer pairs. Order does not matter and each entry stands alone.
HowTo schema: A sequence of steps toward one outcome. Order is the point, and steps do not make sense out of sequence.
Choosing wrongly produces valid markup describing the wrong thing. If your content is five common questions, it is FAQPage. If it is how to do X in five steps, it is HowTo.
Marked-up answers only help if engines fetch and read the page. Findrix checks structured data alongside crawler access and rendering in a 31-point audit, then tracks how often seven AI engines cite you against named competitors. Every gap comes with the fix already written: technical, content and off-site. The audit is free, takes about a minute, and requires no signup.
What FAQ schema requires
- mainEntity: An array of Question items, each one a distinct question.
- name: The question text, phrased as a user would ask it.
- acceptedAnswer: An Answer type containing the response.
- text: The answer body, self-contained and direct.
- Visible equivalence: Every pair must appear on the rendered page.
Why the rich result change did not kill it
Google restricted FAQ rich snippets to a narrow set of site types, and a lot of teams concluded the markup was dead. That reading was too narrow, because it treated blue-link rich results as the only consumer of structured data.
AI engines produce answers shaped like questions and answers. Content already organised that way, with explicit machine-readable boundaries, is unusually easy for them to extract and attribute.
Stop adding FAQ blocks to pages with no real questions. Keep the markup where the content genuinely is a set of questions your buyers ask.
Frequently asked questions
Does FAQ schema still produce rich results?
For most sites, no. Google limited FAQ rich snippets to a small set of authoritative site categories. Government and health sites still commonly see them; typical commercial sites generally do not.
Should I still add FAQ schema?
Where the page genuinely contains questions people ask, yes. It makes the answers machine-readable for AI engines and voice surfaces, and it costs almost nothing. Adding fabricated FAQs purely to carry markup was never worthwhile and is less so now.
How many questions should an FAQ section have?
Enough to cover what people genuinely ask, which is usually three to six. Long lists of manufactured questions dilute the page and give engines weak material to work with.