FAQPage schema is a structured data format that tells search engines and language models: this section of the page contains questions, and these are their answers. Without it, a machine has to infer the relationship from headings, spacing and visual hierarchy — which works sometimes and fails often.

The markup is JSON-LD, a block of structured data placed in your page's head. It does not change how your page looks. It changes how confidently a machine can parse it.

That confidence is the entire point. When an AI assistant assembles an answer, it favours sources it can extract cleanly. A clearly labelled question with a self-contained answer is the easiest possible thing to lift.

This deserves a direct answer, because the honest one is more nuanced than most pages admit.

In August 2023 Google narrowed FAQ rich results to well-known government and health websites. If you run an ecommerce store or an agency site, adding FAQPage markup will almost certainly not produce the expandable question boxes in search results that it once did. Anyone promising you rich snippets from FAQ schema is selling an outdated benefit.

So why bother?

• Answer engines rely on structure. ChatGPT, Perplexity, Claude and Google's AI Overviews assemble answers from sources. Explicitly marked question-answer pairs are the cleanest extraction target available.
• It disciplines your writing. Marking content as an FAQ forces genuine questions and self-contained answers rather than rambling prose.
• It clarifies page meaning. Structured data helps search engines understand what a page covers, even without a visual reward.
• The cost is near zero. Generating and pasting the markup takes minutes.

This is the rule that breaks most implementations. You cannot mark up questions and answers that do not appear in the visible page content. Hidden FAQ markup is a structured data violation and can trigger a manual action. Whatever goes in the JSON-LD must be readable by a person who visits the page.

Each Question object holds a single question and a single accepted answer. Do not bundle several questions into one entry, and do not repeat the same question across multiple pages on your site — pick the page where it genuinely belongs.

Answers should answer. A question like "how long does shipping take" should be met with shipping times, not a paragraph about your commitment to excellence. Promotional content in FAQ markup is both a guideline violation and less likely to be quoted.

FAQPage is for a page containing a list of questions and answers. If your page is a single question with a single answer — a support article, for instance — QAPage is the more accurate type. Choosing the wrong type is a common and avoidable error.

Valid markup gets you parsed. Good writing gets you cited. The two are separate skills and the second matters more.

Patterns that consistently perform:

• Lead with the answer. First sentence resolves the question. Qualifications come after, not before.
• Two to four sentences. Long enough to be complete, short enough to lift whole.
• Include a specific figure where one exists. "Within 24 hours" beats "quickly" every time — concrete numbers are far more quotable.
• Make each answer self-contained. An answer that depends on the paragraph above it cannot be extracted.
• Use the phrasing customers use. Match the question to how it is actually asked, not to internal terminology.

Generate the markup above, then place it inside a script tag with the type set to application/ld+json, ideally in your page head. On Shopify you can add it to the page template, to a section, or through your theme's custom head content.

Once it is live, validate. Google's Rich Results Test and the Schema.org validator will both confirm the syntax parses and flag missing required properties. Validation catches trailing commas and unescaped quotes — the two errors that account for most broken structured data.

Three to ten is the practical range for a page-level FAQ. Fewer than three rarely justifies the block. More than ten suggests the content deserves a dedicated FAQ page, or that some questions should become full articles in their own right.

Most broken structured data fails for a handful of repeated reasons. Checking against this list before you publish saves a round of debugging.

The most serious and most frequent error. Some teams generate FAQ markup from a spreadsheet and paste it into templates without adding the visible questions. Google treats this as a violation, and it can result in a manual action against the site.

JSON is unforgiving. A quotation mark inside an answer must be escaped, and a comma after the final item in an array breaks the whole block. These are invisible in a browser and obvious in a validator, which is why validating is not optional.

Copying an identical FAQ block onto fifty product pages dilutes rather than strengthens. Each question should live on the page where it is genuinely relevant.

Question-shaped H2s are good practice and help both readers and models. They are not a substitute for structured data, and structured data is not a substitute for them. The strongest pages do both: a question as a heading, a concise answer beneath it, and markup that confirms the relationship.

FAQPage rarely stands alone. On a well-marked-up page it sits alongside other types that each answer a different machine question:

• Organization — who publishes this content
• Article or WebPage — what kind of content this is
• Product — what is being sold, at what price, with what availability
• BreadcrumbList — where this page sits in the site
• FAQPage — which parts are questions and answers

Together these give a model enough context to describe your page accurately rather than guessing. Entity clarity — a machine knowing who you are and what you do — is one of the strongest predictors of whether AI assistants mention you at all.