Is FAQ Schema Still Worth Adding?

FAQ schema is where the schema-for-AI pitch gets most specific: mark up your questions and answers, the story goes, and answer engines will lift them. We are well positioned to check the story, because we ship FAQPage markup on every page of three production builds — and the honest accounting is that the visible questions earn their keep while the markup itself has never demonstrably earned anything. Here is what FAQPage does, what the evidence says it doesn't, and why we keep it anyway.

What does FAQPage markup actually do?

It describes, in machine-readable form, a Q&A block that must already exist on the page. schema.org defines the type in one line: "A FAQPage is a WebPage presenting one or more 'Frequently asked questions'" — a subtype of WebPage, adjacent to QAPage, which schema.org reserves for pages where one question collects community answers. The markup states each question and its accepted answer; it asserts nothing the visible page doesn't.

That mirror relationship is a rule, not a style choice. Google's structured-data guidelines require markup to reflect the visible content of the page, so a FAQPage block whose answers appear nowhere in the rendered text is out of spec by definition. Whatever the markup may earn, it earns only as a companion to visible words.

Does FAQ schema get you cited by AI engines?

No evidence on our source list says so, and the best available test says no for JSON-LD generally. Ahrefs tracked 1,885 pages that added structured data between August 2025 and March 2026 against roughly 4,000 controls and found "no major uplift in citations on any platform" — the per-platform numbers were −4.6% on Google AI Overviews, +2.4% on AI Mode, and +2.2% on ChatGPT, the latter two statistically zero. The study did not identify FAQ markup as an exception to the null result.

Google closes the loop from the platform side: eligibility for AI Overviews and AI Mode is indexed-plus-snippet-eligible, with "no special schema.org structured data that you need to add." And at answer time the markup appears to go unread entirely — retrieval testing reported in the same Ahrefs study found every AI system extracting only visible HTML, a finding unpacked in do AI assistants read JSON-LD. Anyone selling FAQ markup as an AI-citation tactic is selling against the platform docs and the only controlled test on record.

What about FAQ rich results in classic Google?

Rich results are a real, documented mechanism with an undocumented grant rate — structured data makes a page eligible for enhanced display, and Google decides the rest. Its structured-data introduction is explicit that markup enables eligibility for rich result features, never that it guarantees them, and which enhancements actually show, for which sites and queries, has always been Google's call.

So the operational advice is empirical, not archival: check your own live SERPs for what FAQ markup currently produces for sites like yours, and price the markup against what you actually see, not against screenshots from someone's 2021 deck. The evidence ledger, period-stamped:

Why do we still ship FAQPage on every page?

Because the discipline it mirrors is the product, and the mirror costs nothing. Every page across our 3 production builds — 400+ published pages — ships a visible block of 4 to 6 questions phrased the way people actually ask them, each answered in 30–60 words, and a FAQPage block generated from that same content [our data]. Our build templates emit the markup automatically, so its marginal cost per page is 0.

The measurable value has come from the visible half. Writing real questions in the asker's phrasing forces pages to answer things readers actually type — the same self-contained, liftable passages that every answer surface quotes, which is why the pattern sits at the center of how we structure pages for AI Overviews. We cannot attribute a single citation to the FAQPage markup itself, and we say so on the record; what we can say is that no page with a good FAQ block has ever been worse for having one.

When is FAQ schema not worth it?

When the markup is the deliverable instead of the mirror. Concretely, skip it in three cases. If your page has no genuine frequently-asked questions, don't manufacture a block to carry markup — Google's guidelines already prohibit markup that outruns visible content. If your stack makes structured data a paid line item — per-page agency work, plugin subscriptions bought for this alone — the controlled evidence says that money buys roughly nothing; templated-once or not at all. And if a proposal justifies FAQ markup with AI-visibility numbers, ask for the controlled study, because the one that exists found zeros.

Against our own interest, plainly: we sell builds that include all of this markup, and you do not need us — or anyone — to add FAQPage to a page. It is among the simplest schema types to template in an afternoon. Spend instead where the evidence points: visible questions, direct answers, extractable passages — the system documented across our generative engine optimization guide, with the broader schema evidence weighed in does schema markup help AI citations.