FAQ schema vs HowTo schema — what each does for AI search

Google's 2023 Search Central updates deprecated HowTo rich results entirely (mobile in August, desktop in September) and restricted FAQ rich results to well-known government and health websites. For most sites, the “which performs better” question is moot: neither type earns rich results today. What remains is parse hygiene — labeling content so crawlers extract it cleanly — and neither type has ever caused AI citations.

What each type is

FAQPage (schema.org/FAQPage)

FAQPage is a schema.org type that marks up a page containing one or more question-and-answer pairs where both the question and answer are authored by the same party. It maps each Q&A pair as a Question with an acceptedAnswer. Valid use: a support page where your team wrote both the questions and the answers. Invalid use: a page of marketing claims formatted to look like questions, or a community forum where users supply the answers.

HowTo (schema.org/HowTo)

HowTo marks up a page that describes how to accomplish a task through a sequence of steps. Each step is a HowToStep with a name and optional text, image, and url. Valid use: a numbered guide with discrete, actionable steps toward a specific outcome. Invalid use: a blog post with loosely related tips, a listicle, or a feature description page.

The 2023 deprecation — what it means

Google Search Central announced two changes in August 2023. First, HowTo rich results were deprecated — on mobile in the August announcement, with the desktop deprecation following in September 2023; today no site, regardless of authority, earns HowTo carousel or step cards in Google search results. Second, FAQ rich results were restricted to well-known, authoritative government and health websites. For the overwhelming majority of commercial sites, both changes mean the same thing: these schema types will not produce any visual enhancement in Google search results.

The practical effect: any SEO guide written before August 2023 recommending FAQPage or HowTo schema for rich-result benefit is describing a path that no longer works for most sites. The optimization advice you may have read — “add FAQ schema to win the FAQ accordion in Google” — is outdated.

The honest summary: using these types is still fine where the content is genuinely Q&A or procedural. Adding them just for rich results no longer has a payoff.

Use-case matrix — question type to markup choice

The decision is not “FAQ vs HowTo” for performance — it is “does this content fit either type at all?”

Content type	Correct markup	What it actually does now
Real Q&A pairs authored by the same party	FAQPage	Parse hygiene — labels Q&A structure for crawlers; enforces answer-shaped content discipline; no rich result for most sites
Procedural steps toward a specific outcome	HowTo, or a plain `<ol>`	Step-extraction clarity for parsers; HowTo rich results deprecated — plain list achieves similar clarity without schema overhead
Marketing bullets or feature list formatted as questions	Neither — omit schema	Applying FAQPage here is a false claim to parsers; schema/visible-text divergence is a quality signal problem
Community Q&A (user-supplied answers)	QAPage, not FAQPage	Correct labeling; FAQPage requires a single authoritative answerer
Government or accredited health site with real Q&A	FAQPage	FAQ rich result still possible per Google's post-2023 policy

Neither type causes AI citations

The schema types themselves — FAQPage or HowTo — do not cause an answer engine to cite your page. The underlying answer-shaped content is what matters. A page with clearly written Q&A pairs will surface in AI responses because the answers are crawlable, extractable, and appropriately indexed — not because FAQPage markup is present. A page with weak, vague answers will not be cited, regardless of how precisely the schema validates.

One genuine side effect of using FAQPage markup correctly: to write valid FAQPage schema, you must write a real question and a direct answer in the visible page text (markup that diverges from visible text is a quality problem). That discipline — writing actual direct answers, not marketing hedges — is what improves extractability. The schema is the side effect; the answer quality is the cause.

For a deeper treatment of why schema generally is not a citation lever, see why schema markup is not enough to get cited. This page focuses on the type-choice decision specifically.

Common mistakes

FAQ-stuffing for rich results that no longer exist

The most common pattern we audit: commercial pages with 8–12 FAQPage-marked questions that were added specifically to win the FAQ accordion in Google search results. That accordion is gone for most sites. The markup remains, often covering questions that are actually marketing copy, not genuine Q&A. This is dead schema weight — it adds no value and creates divergence risk.

Marking up content that is not real Q&A

FAQPage requires that the question and answer are authored by the same party and that the content is genuinely a question-and-answer pair. “Why choose us?” followed by three marketing bullet points is not a Q&A pair. Applying FAQPage to it tells parsers the page contains structured Q&A when it does not — that divergence between markup and content is a quality signal problem, not just a missed optimization.

Schema/visible-text divergence

FAQPage schema must match the visible text exactly. If the JSON-LD contains a question and answer that does not appear in the rendered page, or if the answer text in the schema is substantially different from what the user reads, the markup is making a false claim to parsers. Every FAQPage question and its acceptedAnswer must be present verbatim in the visible HTML.

Using HowTo for general blog posts or tips lists

HowTo is for discrete steps toward a specific outcome. A blog post titled “10 tips for better email open rates” is not a HowTo — tips are not steps, and there is no single defined outcome achieved by completing the list in order. Applying HowTo to this content mislabels the page type and, given that HowTo rich results are deprecated anyway, provides no benefit.

Frequently asked questions

Does FAQ schema or HowTo schema still earn rich results in Google?

For most sites, no. Google's 2023 Search Central updates deprecated HowTo rich results entirely (mobile in August, desktop in September) and restricted FAQ rich results to well-known government and health websites. If your site is not in those categories, neither type will trigger a rich-result enhancement in Google search.

Is there any remaining reason to use FAQPage schema if rich results are gone?

Yes, but a narrow one: parse hygiene. FAQPage markup labels question-and-answer pairs so crawlers and parsers can extract them cleanly. It does not cause AI citations — the underlying answer-shaped content is what matters — but it does reduce structural ambiguity. Only mark up real Q&A pairs; never use it on marketing bullets.

What does HowTo schema do now that rich results are deprecated?

HowTo markup signals that a page contains procedural steps, which helps parsers extract step text as structured content. A plain ordered list achieves similar step-extraction clarity without schema overhead. Neither approach causes AI citations; the quality and specificity of the step content is the actual signal.

Is it harmful to add FAQPage or HowTo schema to a page that does not qualify?

Yes. Marking up content that is not actually Q&A pairs (for FAQPage) or procedural steps (for HowTo) is a false claim to parsers. Schema-to-visible-text divergence — where the markup describes content that does not match the rendered page — is a quality signal problem, not just a missed opportunity.

Sources cited on this page

The August 2023 Search Central post restricted FAQ rich results to government/health sites and deprecated HowTo on mobile; the desktop HowTo deprecation followed in September 2023. Both are sourced from Google Search Central (linked above). Schema.org type definitions are sourced from schema.org directly. No citation counts, rank positions, or performance claims appear on this page.

What this does not guarantee

FAQPage and HowTo schema — like all schema.org markup — are parse hygiene, not citation levers. No action described on this page promises citation by an AI assistant, a rich result in Google, or any improvement in AI-response frequency.
No specific citation count, Bing rank position, Google search position, or AI-response outcome is promised by any recommendation on this page.
Applying the correct schema type removes mislabeling noise. It does not guarantee that an answer engine will surface your content within any particular timeframe.
Where Prompt Goblin engagement is mentioned: the refund covers the delivered work — audits, technical fixes, copy, measurement loop. It never covers a citation number or ranking position.

Not sure whether your structured data matches your visible content? Run a free scan and we will flag any schema/visible-text divergence, wrong type use, and the citation-gap levers that actually matter.