Do FAQ schemas still help SEO?

Yes, but indirectly. They no longer trigger accordion results for most sites, but they help AI Overviews and voice assistants extract answers cleanly.

Can I have FAQ schema on every page?

Technically yes, but only if each page has genuinely unique, useful Q&A. Pasted boilerplate FAQs across pages is treated as spam.

FAQ Schema: How to Implement and Why Most Sites Lose It

FAQPage vs QAPage

Use FAQPage when one author wrote both questions and answers (like this page). Use QAPage when users post questions and others answer (forums, Q&A communities). They're not interchangeable.

Every Q must have exactly one accepted A

Multiple acceptedAnswer entries break the schema. Use suggestedAnswer for additional perspectives.

Mark up only visible content

Schema for hidden text is a guidelines violation. Every Q&A in your JSON-LD must be visible to users on the page.

Test with the Rich Results Test

search.google.com/test/rich-results - it shows exactly what Google sees and flags errors before you deploy.

In-depth guide

A longer, practitioner-level breakdown of FAQ schema implementation - written for readers who want the full picture, not just the summary above.

The state of FAQ schema after the 2023 rollback

In August 2023, Google dramatically reduced the display of FAQ rich results in the SERP. Previously, any page with valid FAQPage schema could earn an accordion-style rich result that showed questions and answers directly under the SERP entry. Now, that display is reserved for well-known authoritative health and government sites. Most publishers no longer see the accordion at all.

This does not mean FAQ schema is useless. Google still parses it and still uses it as a signal for AI Overview inclusion, for voice assistant answer extraction, and for internal understanding of your page structure. Sites that removed FAQ schema after the 2023 change often saw AI citation rates drop, because the schema was doing more work behind the scenes than the visible accordion suggested.

The pragmatic recommendation: keep FAQ schema on pages that genuinely have Q&A content, remove it from pages where the FAQ was fabricated to game the accordion display. The former group still benefits; the latter group was risking a manual action even before the rollback.

FAQPage vs QAPage: the distinction Google enforces

FAQPage schema is for content where a single author has written both the questions and the answers - a support page, a landing page FAQ section, a documentation FAQ. QAPage schema is for user-generated Q&A where different users post questions and answers - forum threads, Q&A community sites, product review Q&A sections.

The two are not interchangeable. Using FAQPage on user-generated Q&A content is a schema violation and can trigger a manual action. Using QAPage on author-written FAQ content is also incorrect and gets ignored. The distinction is about content source, not content format.

The practical check: if you or your team wrote both sides of the exchange, use FAQPage. If your users wrote either side, use QAPage. For pages that mix both (a support article with a user-comments Q&A section below), you can nest QAPage inside an Article - do not use FAQPage across both sections.

One question, one answer: the exactly-one rule

Every FAQPage question must have exactly one acceptedAnswer. Multiple acceptedAnswers per question is a schema violation that causes Google to ignore the entire FAQPage markup. If you have multiple valid perspectives on a question, put the primary one in acceptedAnswer and additional ones in suggestedAnswer.

The acceptedAnswer text should be complete on its own. It should not depend on the surrounding page for context. If the answer is 'yes' or 'no,' pad it with the reasoning: 'Yes, because X, Y, and Z.' AI systems and voice assistants extract acceptedAnswer text verbatim - if it does not stand alone, it fails as an answer.

Answer length should be roughly 40 to 300 words. Shorter answers rarely get lifted into voice or AI results. Longer answers get truncated at extraction time and read awkwardly when read aloud. The 40-to-300 range covers the sweet spot for most extraction systems.

Only mark up visible content: the guidelines line

Google's structured data guidelines are explicit: only mark up content that is visible on the page. Adding FAQ schema for questions and answers that are not actually rendered in the DOM is a guidelines violation and can trigger a manual action for structured data spam.

This affects two common patterns. First, marking up content inside collapsed accordions is fine as long as the content is in the DOM and openable by users. Do not mark up content that is dynamically inserted after page load only for the purpose of appearing in schema.

Second, do not mark up questions and answers that are hidden behind CSS display:none unless they are meaningfully accessible (via a tab switch, an accordion toggle, or similar user action). Content permanently invisible to users must not be in the schema, even if it is technically in the HTML.

Beyond FAQ: the six schema types every content site needs

FAQ is one of many. For content sites, six schema types cover the majority of use cases: Article (for blog posts and long-form content), FAQPage (for FAQ sections), HowTo (for step-by-step tutorials), Breadcrumb (for navigation clarity), Organization (once on your homepage, for entity establishment), and WebSite with SearchAction (for the sitelinks search box).

Article schema unlocks the top-stories carousel eligibility for news sites. HowTo schema unlocks step-by-step rich results (with numbered steps and images). Breadcrumb schema replaces URL paths in the SERP with clean breadcrumb trails, which improves CTR. Organization schema helps Google's knowledge graph identify you as an entity and can trigger the knowledge panel.

Product, LocalBusiness, Recipe, Event, Course, and Job are the other high-impact schemas for their respective content types. Add the ones that fit your content genuinely. Do not add schemas that do not fit - that is spam and gets penalized.

JSON-LD vs Microdata vs RDFa: pick JSON-LD

There are three ways to embed structured data: JSON-LD (a script tag in the head), Microdata (inline attributes on HTML elements), and RDFa (a different flavor of inline attributes). Google supports all three but recommends JSON-LD as of 2015 and has repeatedly recommended it since.

JSON-LD wins because it decouples structured data from HTML markup. You can update your schema without touching your HTML templates, and you can inject it dynamically from any framework. Microdata and RDFa force you to entangle schema with layout, which makes both harder to maintain over time.

The one place inline schema still makes sense is when structured data has to match user-visible content exactly and dynamically (product prices that change based on user location, for example). Even then, generating the JSON-LD server-side is usually cleaner than maintaining inline markup.

Testing and validation workflow

Two tools are essential. Rich Results Test (at search.google.com/test/rich-results) shows exactly what Google can extract and what rich results your page is eligible for. Schema.org Validator (validator.schema.org) shows spec compliance without the Google-specific lens. Use both. Rich Results Test catches Google-specific issues; Schema.org Validator catches structural issues that Rich Results Test may skip.

Bake validation into your CI pipeline for structured data. Every template that emits schema should have an automated test that renders a sample page and validates the output against the schema.org spec. This catches template regressions before they hit production.

The Enhancements section of Search Console shows structured data errors at scale across your indexed pages. Review it weekly. Any spike in errors usually traces to a specific template change - roll back if the error pattern points at a recent deploy.