FAQ schema generator: structured data, valid by construction.

Article, FAQPage and BreadcrumbList JSON-LD you can paste straight into your head. Schema describes what a page already says — the moment it describes something else, it is a liability.

By · Updated September 4, 2026

JSON-LD schema generator

Runs in your browser. Nothing you paste is sent anywhere.

Paste into your <head>

 

Schema describes what a page already says. It cannot add authority a page has not earned, and marking up a claim the page does not make is the one use of structured data that reliably backfires.

What you can do with it

  • Add Article schema to a blog post with a real author
  • Mark up an FAQ block that already exists on the page
  • Emit breadcrumbs that match your actual URL hierarchy
  • Hand a developer valid JSON-LD instead of a description of it

How it works, step by step

  1. 1

    Only mark up what is on the page

    This is the rule that matters. Structured data describing content a visitor cannot see is against Google's guidelines and is the single most common cause of a structured-data manual action. If the FAQ is not rendered, it does not get FAQPage markup.

  2. 2

    Name a real author

    A Person with a name and a URL that resolves to a real profile is worth considerably more than an Organization byline, both for E-E-A-T and for an assistant deciding whether a claim has a source. A fabricated author is worse than none.

  3. 3

    Keep dates honest

    dateModified should change when the content changes, not on every deploy. Bumping it without editing anything is a signal you are training search engines to ignore.

  4. 4

    Validate before you ship

    Paste the output into Google's Rich Results Test and Schema.org's validator. They disagree occasionally, and where they do, Google's is the one that governs what appears in Google.

  5. 5

    Put it in the head, once

    One block per type per page. Duplicate Organization or Article nodes on the same page are a common source of confusing validator output, usually because a plugin is already emitting one.

The one rule

Structured data is a description of the page for machines. Everything good about it and every way it goes wrong follows from that.

If the description is accurate, you have made the page easier to parse, easier to attribute, and eligible for richer presentation. If it is not — an FAQ that is not on the page, a rating nobody gave, an author who does not exist — you have made a false claim in a machine-readable format, which is the easiest kind to detect and act on.

The temptation is real because the markup is invisible to visitors. Treat it as public statement anyway, because that is how it is read.

Which types are worth the effort

Article on anything editorial, with a real named author whose URL resolves. This is the highest-value markup for most sites and the one most often left as a bare Organization byline.

BreadcrumbList where you have a genuine hierarchy. It is cheap, it is hard to get wrong, and it still affects how the URL line displays.

FAQPage only where an FAQ block genuinely exists. Rich results for it are now rare, so the return is machine-readability rather than a snippet — worth doing, not worth inventing questions for.

Everything else — Product, Recipe, Event, HowTo — is worth it when the page is genuinely that thing. Nothing is worth it when the page is not.

Where markup sits in the order of work

Schema is the last 10% of making a page readable by machines, not the first. A page that cannot be crawled, or that never states its answer in plain prose, gains nothing from perfect JSON-LD. Fix access, then structure, then describe.

To find out which of those three you are actually missing, test your homepage against the nine AI-readability signals. It reports whether any JSON-LD is present at all and, separately, whether an Organization or LocalBusiness type is among it, which is the block that turns a site into a resolvable entity rather than an unattributed document.

Frequently asked questions

Does schema markup improve rankings?

Not directly. Google has been consistent that structured data is not a ranking factor. What it does is make a page eligible for rich results and easier for a machine to parse, which affects click-through and citation rather than position. That is still worth having; it is just not the mechanism people assume.

Does FAQPage still show rich results?

Google reduced FAQ rich results substantially in 2023, so treat the visual snippet as unlikely rather than expected. The markup remains useful as machine-readable question-and-answer structure, which is exactly the shape an assistant extracts, so the value moved rather than disappeared.

Can I mark up an FAQ that is hidden behind an accordion?

Yes. Content inside a collapsed accordion is present in the HTML and visible on interaction, which satisfies the guideline. Content that is not in the HTML at all, or is display:none with no way to reveal it, does not.

What happens if my schema is wrong?

Usually nothing dramatic: invalid markup is ignored and you lose the eligibility you were hoping for. Markup that misrepresents the page is different — that is a guidelines violation and can attract a manual action, which is why the honest-description rule is not pedantry.

Do I need schema for AI search?

It helps and it is not sufficient. Schema tells a model what the page is unambiguously, which reduces the chance of being mischaracterised. But assistants extract from body content, so a page with perfect markup and no clear answer in the prose still does not get cited. Structure the writing first, then describe it.

Be the site the assistant names.

Add a domain and the first draft is written while you watch. Every publish is your decision, a click on the draft or a rule you set and can hold, and you can read the source or self-host it free.

Add a domain, it sets up your workspace