FAQ schema in 2026: the rich result is gone, the markup is not

FAQ schema no longer produces anything you can see in Google Search. Google deprecated the FAQ rich result in May 2026 and removed its documentation the following month. The FAQPage type still exists in the schema.org vocabulary and still validates, so the decision is no longer "how do I get the rich result" but "is a machine-readable question list worth the markup". For most pages the answer is no.

Before you start

This chapter assumes you already know which structured data types are worth shipping at all, which is covered in schema markup for AI search, and that you have Article in place where it belongs — see article schema. A structured data block is a claim about the page written for machines; it does not add content and it does not change ranking on its own.

What changed, and when

Three dated steps, all from Google's own changelog, and the direction never reversed.

Both quotes are from the Search Central changelog (Latest Google Search documentation updates, read 2026-09-02). The 2023 restriction was announced in Changes to HowTo and FAQ rich results.

The vocabulary side did not move. schema.org still defines the type — "A FAQPage is a WebPage presenting one or more 'Frequently asked questions'" — and reports it in use on one to ten million domains (schema.org/FAQPage, read 2026-09-02). Nothing about your existing markup broke. The display it was written for is what disappeared.

A deprecated rich result does not make the markup invalid. It makes the reason you added it invalid.

This is the second type to travel that road. The same August 2023 announcement covered HowTo, and that one went further — the documentation was withdrawn outright rather than narrowed to a category of site. Two types announced together, one narrowed and then killed, one killed immediately. If you are deciding how much to invest in any single rich result today, that pair is the most useful precedent available.

Should you keep FAQ schema: four cases

Work out which row you are in before touching anything. Only one of the four calls for removal work.

The visible-content rule is the only line here that can actually cost you something. Google's structured data guidelines say: "Don't mark up content that is not visible to readers of the page." Breaking it can trigger a manual action, and the guidelines are precise about the consequence — "a page loses eligibility for appearance as a rich result; it doesn't affect how the page ranks in Google web search" (General structured data guidelines, read 2026-09-02).

Do it: four steps

Each step has a finish line you can check without waiting for anything.

1. Find out whether you already ship it. Run the command in the next section against your own page. Finish line: a count, not a guess.
2. If the count is above zero, open the page as a visitor and read the questions on screen. Finish line: every Question in the JSON has visible text on the page.
3. If any question is missing from the visible page, delete that entry from the markup rather than adding text to satisfy it. Finish line: the two lists match.
4. Move the effort to the visible structure — one question per heading, the answer's first sentence directly under it. Finish line: reading the headings alone gives someone the answers.

Deliverable: one command, one block, one checklist

Start with the count, because most teams do not know whether their CMS emits this type.

# Does this page ship FAQPage markup, and how many questions?
curl -sL --compressed -A 'Mozilla/5.0' https://example.com/page > page.html
grep -c '"@type" *: *"FAQPage"' page.html
grep -o '"@type" *: *"Question"' page.html | wc -l

If you are keeping or writing the block, this is the whole thing. Two questions shown; the pattern repeats.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "How long does indexing take?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Google states that crawling can take anywhere from a few days to a few weeks."
      }
    },
    {
      "@type": "Question",
      "name": "Does FAQ markup still show a rich result?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "No. Google removed the FAQ rich result from Search in May 2026."
      }
    }
  ]
}
</script>

Six lines to check before it goes out. Every one of them is answerable in under a minute.

Three ways this goes wrong

All three are common and all three are visible from outside the site.

The first is markup that describes an accordion nobody expands, where the answers exist in the HTML but the page shows only the questions. That is the visible-content rule, and it is the one case here with a defined penalty.

The second is a plugin emitting FAQPage on every page of a site, including a pricing page whose "FAQ" is three sentences of copy. Nothing breaks, but it tells any machine reading the site that its author does not distinguish between page types — the same signal problem covered in structure a page for AI answers.

The third is writing questions for the markup rather than for readers. Invented questions produce answers nobody needed, and those answers are what an engine quotes when it quotes anything at all. The tell is easy to spot in your own copy: a question phrased the way a marketer would ask it rather than the way a person types it into a search box. "What are the benefits of our platform?" is not a search term anyone has ever entered.

There is a fourth pattern that is not quite a mistake but wastes the same afternoon: teams auditing every page for FAQ markup after reading that the feature was deprecated. The deprecation removed a display, and a display that no longer appears cannot be appearing wrongly. Unless a page breaks the visible-content rule, the block sitting in its head is inert, and inert markup is not a backlog item.

Where this stops

Two boundaries, and the second one is the honest limit of this chapter.

Removing FAQ markup will not improve anything. There is no evidence that a deprecated rich result type harms a page, and treating cleanup as an optimisation is how afternoons disappear. Delete it only when the visible-content rule is broken.

Whether AI answer engines read FAQPage markup, we do not know. We have not tested it, and no engine we know of documents an answer either way. What is testable is that the same content in headings and short paragraphs is read by everything, which is why step 4 above points at the visible page and not at the JSON. If you want to confirm that an engine can reach and read the page at all before worrying about either, run the AI crawler check.

Common questions

Is FAQ schema still worth adding in 2026?

For the rich result, no — it does not exist any more. For a machine-readable list of questions on a page that genuinely has one, it costs almost nothing and validates cleanly. Neither reason is strong enough to make it a priority.

Will removing FAQ schema hurt my rankings?

Google's guidelines state that a structured data manual action "doesn't affect how the page ranks in Google web search", which is the closest thing to an answer on record. We have no first-party data on removals and would not claim a direction from anecdotes.

Does FAQPage still validate?

Yes. The type is defined in the schema.org vocabulary and parsers accept it. Validating and producing a search feature are separate things, and this type now does only the first.

What replaced the FAQ rich result?

Nothing did. Google removed the feature without a successor, and the practical replacement is the visible page: a question as a heading, the answer in the first sentence under it.

Should I use an FAQ schema generator?

You can, but a generator will happily produce questions your page does not answer, which is exactly the failure this chapter is about. Write the visible questions first and generate from those.