Free tool · nothing leaves your browser
FAQ schema generator
Valid FAQPage markup is the easy half and every generator does it. This one also scores whether each answer still makes sense once something lifts it off your page — which is what decides whether it gets quoted or ignored.
What FAQ schema is
FAQ schema is a JSON-LD block declaring a page's questions and answers in a fixed vocabulary, so a machine can read them as pairs rather than inferring structure from headings. It is simple markup with one real requirement: every answer has to make sense on its own, because anything that quotes it will quote it without the page.
Set expectations honestly. Google restricted FAQ rich results to well-known authoritative government and health sites, so an ecommerce store adding this today should not expect expandable questions in search results. The markup is still worth adding — for models, for structure, and for the discipline it imposes on your answers — but any generator implying a search feature is describing a world that has changed. Checked 2026-09-09.
Write the pairs.
Prefilled with two answers: one that travels, and one that falls apart the moment it leaves the page. Take yours from the support inbox rather than inventing them.
FAQPage JSON-LD
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "How long does delivery take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Orders placed before 2pm ship the same working day from our Ohio warehouse. Standard delivery arrives in 2–5 working days across the contiguous US, and express in 1–2. Shipping is free over $50 and $6.95 below it."
}
},
{
"@type": "Question",
"name": "Can I return it?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes. As mentioned above, the window is 30 days from delivery and the process is the same as described in the returns section below."
}
}
]
} Will each answer survive being quoted?
- How long does delivery take? 55
- 37 words — in the range that gets quoted whole.
- Depends on the surrounding page: “below”. Lifted out of context this answer stops making sense, which is exactly what happens when a model quotes it.
- Can I return it? 40
- 24 words — in the range that gets quoted whole.
- Depends on the surrounding page in 2 places: “as mentioned above”, “below”. Quoted on its own, this answer says almost nothing.
- Leads with the answer. This is the shape that gets extracted.
What makes an answer travel
The rules this scores against, published in full. They are ours, and they are about writing rather than markup.
| Rule | Why |
|---|---|
| No reference to the page around it | “As mentioned above” or “see below” is fine in place and meaningless once quoted. This is the heaviest penalty in the score, because it is the one that breaks the answer completely. |
| Lead with the answer | The first sentence is the part most likely to be lifted. An answer that opens with “it depends” has spent it on nothing. |
| Forty to sixty words | Long enough to be useful, short enough to be quoted whole. Under fifteen carries no information; over a hundred and twenty gets truncated or skipped. |
| Name the brand, not just “we” | Quoted elsewhere, “we ship in two days” loses its attribution. The brand name travels with the sentence. |
| Ask it the way people ask it | Real queries are longer and messier than headings. “Can I return it?” matches nothing; “How long do I have to return an order?” matches what people type. |
The weights are our judgement, not a published standard, and the score is a writing aid rather than a measurement. Whether extractable answers get quoted more often is metric to confirm — we have not measured it and will not imply a correlation nobody has demonstrated.
Where to put it
Three rules, and one thing to check first.
- On the page the questions are visible on Structured data must reflect content a visitor can see. Marking up hidden questions is the same class of mistake as marking up reviews you do not display.
- One FAQPage block per page Two competing blocks is worse than one — the crawler picks, and it may not pick yours. Check the theme and any app before you add another.
- In the head or the body, either works JSON-LD is read wherever it appears. On Shopify, a snippet rendered from the template keeps it with the content it describes.
- Validate the rendered page, not the snippet Run the live URL through the schema.org validator once it is installed. A block that is valid in isolation can still be wrong in place.
What this will not do
Get you a rich result. Covered above, and worth repeating because most pages about FAQ schema still promise it. The restriction to authoritative government and health sites means an ecommerce store should treat rich results as out of scope.
Make bad answers good. The score checks structure, not truth. An extractable answer that is wrong is worse than an unquotable one — it travels.
Tell you which questions matter. That is in your support inbox and your search console, not in a generator.
Replace product data. FAQ markup answers questions around the product. What the product actually is belongs in Product markup, and a model answering a buying question needs both.
Definitions
- FAQPage
- The schema.org type for a page of questions and answers written by the site owner.
- Rich result
- An enhanced search listing. For FAQ markup, now restricted to authoritative government and health sites.
- Extractability
- Whether an answer still makes sense when lifted away from its page. What this tool scores.
- JSON-LD
- Structured data as a JSON block rather than attributes threaded through markup. The recommended format.
- Context dependence
- An answer leaning on surrounding content. The most common reason a well-written answer cannot be quoted.
Questions about FAQ markup
Will FAQ schema get me a rich result?
Almost certainly not. Google restricted FAQ rich results to well-known authoritative government and health sites, so an ecommerce store adding this markup today should not expect the expandable questions in search results that older articles promise. That is not a reason to skip it — the markup still gives models and other consumers a clean, labelled question-and-answer structure — but any tool implying you will get a search feature from it is selling something that no longer applies.
Then why add FAQ markup at all?
Because a labelled question and answer is the easiest thing in the world for a machine to lift and quote correctly. Search features come and go; the structure is durable, it costs an afternoon, and it forces you to write answers that stand on their own — which is the part that actually changes whether you get quoted accurately.
What makes an answer extractable?
It stands alone. An answer that says “as mentioned above” or “see the section below” is fine on the page and useless the moment anything quotes it, because the context it leans on is gone. The best answers lead with the answer in the first sentence, run 40 to 60 words, and name the brand rather than saying only “we”.
How many questions should I mark up?
The ones people actually ask, which for most stores is between four and eight per page. Pull them from your support inbox rather than inventing them — a page marked up with questions nobody asks is a page optimised for a machine and not a person, and it reads that way to both.
Can I mark up questions that are not visible on the page?
No. Structured data has to reflect content the visitor can see. Marking up hidden questions is against Google's guidelines and is the same class of mistake as marking up reviews you do not display.
Is this different from product schema?
Yes, and they complement each other. Product markup describes the thing you sell — price, availability, specs. FAQ markup describes the questions around it: shipping, returns, suitability, care. Most stores need both, and the product schema generator handles the first.
Related
Find out what you’re losing.
Before you commit to anything, we tell you exactly what you’re losing and what it costs to stop it. Two weeks. Fixed fee. Credited in full against any build you go ahead with.
- Fee
- $1,500–$3,000, fixed
- Duration
- Two weeks
- Credited
- In full, against any build
- You supply
- Read access + one 45-minute call