FAQPage schema: a practical guide
FAQPage markup tells a machine which part of a page is a question and which part is its answer. Rich results for it have been restricted, but the interpretive value has not gone away, and it remains one of the cheapest things you can get right.
Structured data has had a confusing few years. Rich result eligibility for FAQ markup was narrowed sharply, a lot of teams stripped it out in response, and a lot of advice online still describes the previous situation. Worth separating two different things the markup does.
What it still does, regardless of rich results
- It states unambiguously which text is the question and which is the answer, so a retrieval system does not have to infer it from heading levels.
- It gives every answer an addressable identity, which matters when a page carries several.
- It is read by systems well beyond one search engine — other engines, assistants, internal search, and anything built on a crawl.
- It survives a redesign. Markup describing meaning outlives markup describing layout.
FAQPage or QAPage?
| Type | Use when | Do not use when |
|---|---|---|
| FAQPage | You wrote both the question and the answer, and the answer is settled | The page is a forum thread or has competing answers |
| QAPage | One question, potentially several answers, typically user-submitted | You are marking up your own settled help content |
For a company knowledge base the answer is almost always FAQPage on the index and, where each question has its own page, a single-question graph on that page. A hub page that lists thirty questions can carry all thirty; the individual pages carry their own.
Mistakes that get markup ignored
The single most common cause of markup being disregarded. Whatever is in the JSON has to be visible in the rendered page.
"Ready to get started?" is not a question the page answers. Systems detect this pattern and it costs more than it gains.
It works often enough to feel safe and fails silently when a page is fetched without executing scripts, which is precisely when it mattered.
The answer gets edited, the JSON does not. Generate the markup from the same source as the copy so they cannot diverge.
A page that describes itself as an article, an FAQ, a product and a how-to is describing itself as none of them.
Markup does not win an argument with the text. It only clarifies it.
What to add beside it
Three graphs earn their place on almost every knowledge page: an Organization graph, declared once and referenced everywhere, so that a machine can connect the pages to a company; a WebSite graph, so the site itself is an entity; and a BreadcrumbList, so the position of the page in the structure is stated rather than guessed. Author and date fields should be real people and real dates.
Questions this settles
Eligibility has been narrowed substantially and most sites no longer see the expanded FAQ display. The markup continues to serve interpretation by search and assistant systems, which is a separate benefit.
No. The marked-up text must be present in the visible page. Partial or paraphrased markup is the usual reason structured data is disregarded.
Yes, where the answer deserves it. The index lets a machine see the whole library at once; the individual page is where a citation can usefully land.