FAQ, HowTo and Breadcrumb schema

FAQ, HowTo, and Breadcrumb are the three structured-data types site owners historically added for one reason: to win a rich result in Google Search. In 2026 that reasoning no longer works as a bundle, because the three have separated into a clear gradient of usefulness. Breadcrumb still earns a visible rich result. FAQ's rich result is gone, but the question-and-answer shape it labels is still worth marking up. HowTo sits at the optional end: valid and harmless where it fits, with no visible payoff left to expect.

All three are specific types from the schema.org vocabulary. The general mechanics of structured data, what schema.org is, why JSON-LD is the recommended format, and why valid markup matters, are covered separately in our guide to structured data for AI search. This article covers the three types themselves: what each one currently does, what changed, and where each is still worth adding.

Breadcrumb: the one that still earns a result

Breadcrumb markup, the BreadcrumbList type, describes where a page sits inside a site: the homepage, then a category, then the page itself. Google is explicit about what it does with that trail: "Google Search uses breadcrumb markup in the body of a web page to categorize the information from the page in search results" (Google's breadcrumb documentation). The payoff is visible today: instead of a raw URL above the page title, searchers see the readable path, and the feature guide states that "This feature is available on desktop in all regions and languages where Google Search is available." Of the three types this article covers, Breadcrumb is the only one whose rich result never went away.

The same documentation explains why the trail matters beyond the pixels: "A breadcrumb trail on a page indicates the page's position in the site hierarchy, and it may help users understand and explore a site effectively." That second half is easy to skim past, but it is the part with staying power. A BreadcrumbList is an explicit, machine-readable statement of your site's hierarchy. A crawler that reads it does not have to reverse-engineer your structure from URL segments or navigation markup; the markup hands it the tree directly: this page belongs to this category, which belongs to this section, which belongs to the homepage.

That makes Breadcrumb the clear yes of the three. On any page below the homepage with a real position in a hierarchy, such as a product inside a category, an article inside a topic, or a doc page inside a manual, add BreadcrumbList. You get a live rich result in Google Search and an unambiguous hierarchy signal for any machine that parses the page. One honest limit: whether engines like ChatGPT or Perplexity read the trail is not documented anywhere.

FAQ: no rich result now, but the shape still helps

FAQPage markup labels a page, or a section of one, as a series of questions, each followed by its complete answer. For years the reward was the accordion of expandable questions under your listing in Google Search. That reward is gone. Google's Search Central changelog states that "This feature will no longer appear in Google Search starting May 7, 2026" (Google Search Central updates); the FAQPage guide itself was removed in June 2026. The feature had been narrow for years before that. A 2023 announcement restricted it to a small class of sites: "FAQ (from FAQPage structured data) rich results will only be shown for well-known, authoritative government and health websites" (Google's 2023 announcement). For most sites the accordion disappeared long ago; now it is gone for everyone, including the government and health sites that kept it longest.

If the rich result was your only reason for adding FAQ schema, that reason no longer exists. It was never the only reason. A self-contained question followed by a direct answer is a passage that can be read without the rest of the page for context, and FAQPage is a clean, machine-readable label for exactly that structure. Whether any answer engine uses either the structure or the label is undocumented; treat the benefit as mechanistic, not measured. The markup only labels what the content already does; writing Q&A sections that earn retrieval is its own subject.

So the practical position is: keep FAQ schema where it is accurate, and treat adding it as optional rather than as a job worth scheduling. Not for a search feature that no longer exists, but because it puts an unambiguous label on a structure worth having anyway. The old eligibility rule survives as a correctness rule: every question and answer in the markup has to be visible to the reader on the page. Markup describing questions and answers a visitor cannot see was always against the guidelines, and it is exactly the kind of mismatch that gets structured data ignored.

HowTo: the most optional of the three

HowTo markup describes a task as an ordered sequence of steps: do this first, then this, then this. Its rich result has been gone the longest. Google announced the removal in an August 2023 Search Central blog post titled "Changes to HowTo and FAQ rich results" (Google's announcement), the same change that restricted FAQ rich results to the government and health sites mentioned above. The HowTo result was deprecated on mobile in August 2023 and gone from desktop by that September. No engine has documented a replacement payoff since.

What remains is a balanced, modest case. The markup is valid schema.org, it costs almost nothing, and it labels your steps as ordered steps for any parser willing to read them. But it adds little over what clean HTML already says: a numbered list of steps under a clear heading conveys the same sequence to any machine that reads the page, which is why the honest description of HowTo in 2026 is optional rather than either valuable or dead. Add it where your content is a genuine sequential procedure and you want the structure to be explicit in the markup as well as on the page. If you skip it, you are not losing any documented benefit.

One caution if you do add it: HowTo asserts genuine sequence, a set of steps where each depends on the one before it, matched to what is visible on the page. A page of unordered tips, seven independent ways to improve something, is not a HowTo, and marking it up as one misdescribes the content to every machine that reads it.

Do AI engines actually use these types?

Google is the only engine that documents anything per type, and its documentation is the source of every status above. Beyond rich results, Google also states in general terms that "Google uses structured data that it finds on the web to understand the content of the page" (Google's structured data intro). But it does not present these types as a lever for its AI surfaces. Its guidance on generative AI is direct: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add. However, it's a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search" (Google's AI optimization guide). Its AI features page says the same about the surfaces themselves: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary" (Google's AI features page).

Bing has long encouraged schema.org markup in general, and its guidance stops there: it makes no per-type claim about FAQ, HowTo, or Breadcrumb. ChatGPT, Claude, and Perplexity publish no documentation at all saying they consume these types. That is not a gap in this article's research; "not documented" is the finding, and it is worth stating plainly because the third-party field fills the silence with confidence it has not earned. The nearest thing to evidence is one agency test, published by SearchVIU in October 2025, reporting that ChatGPT, Claude, Perplexity, and Gemini process schema markup when they access a page. It was not peer-reviewed or controlled, its numbers are not reproduced here, and processing markup on fetch is not citing a page because of it.

The defensible position is mechanistic. JSON-LD is trivially parseable, and an unambiguous label on a Q&A block or a hierarchy trail can in principle help any machine that chooses to read it. That is an inference from how these systems could work, not a description of how any of them documents working. If someone quotes you a number for how much a schema type lifts AI citations, ask for the vendor documentation behind it. There is none.

Getting it right

One rule governs all three types, and it decides whether any of them does anything at all: the markup must match the visible content. Google's structured data guidelines put it directly: "don't add structured data about information that is not visible to the user, even if the information is accurate" (structured data guidelines). The same idea appears on the AI features page, which lists "Making sure your structured data matches the visible text on the page" among the fundamentals worth doing. A FAQPage block containing questions the visitor never sees, or a HowTo whose steps differ from the steps on the page, is not a clever addition; it is a mismatch that gets the markup disregarded.

Correctness fails in more than one way. Malformed JSON-LD does not parse at all. Markup that parses but describes the wrong thing is a different failure: "Violating a quality guideline can prevent syntactically correct structured data from being displayed as a rich result in Google Search, or possibly cause it to be marked as spam" (Google's structured data policies), while what other engines do with it is undocumented. Markup that is simply unused is not a penalty, but it is not doing anything either. Validate before you publish. The format question, at least, has a documented answer: "Google recommends using JSON-LD for structured data if your site's setup allows it", so a small JSON-LD block in the page head or body is the normal way to ship any of these types.

FAQ

Should I still add FAQ schema in 2026?

Yes, where you have a real Q&A section. The FAQ rich result was removed from Google Search in May 2026, so do not add the markup for that feature; add it because it puts a machine-readable label on a genuine question-and-answer structure, and keep it matched to the visible content.

Does Breadcrumb schema still do anything?

Yes. It is the only one of the three that still produces a visible rich result: Google uses breadcrumb markup to categorize your page in search results, and the markup hands any machine your site hierarchy. Add it to every non-root page with a real place in that hierarchy.

Is HowTo schema worth adding?

It is optional. The HowTo rich result was removed in 2023, and the markup adds little over clean HTML steps. Add it where your content is a genuine step-by-step procedure and you want the explicit structure, but expect no rich-result or citation payoff; skipping it costs you nothing documented.

Do ChatGPT, Claude, and Perplexity use these schema types?

None of them documents doing so. Google uses structured data to understand content and still shows the Breadcrumb result; for the pure-LLM engines, treat any benefit as unproven inference, and ignore the confident third-party citation numbers built on uncontrolled tests.

If FAQ and HowTo rich results are gone, why not delete the markup?

Because valid markup still labels your content unambiguously for any machine that reads it, at almost no cost. Remove it only if it is wrong or no longer matches the page; otherwise leave it in place and keep it accurate.