Most structured data advice lists types that no longer produce anything. The list of ones that still do is short enough to memorise.
What still shows in 2026
| Type | Visible result | Worth it |
|---|---|---|
| BreadcrumbList | Path instead of a URL | Always — trivial to add |
| Article / BlogPosting | Eligible for Top Stories, dates | Yes for publishers |
| Product | Price, availability, rating | Yes for e-commerce |
| Recipe | Image, time, rating, carousel | Yes |
| Event | Date and venue in results | Yes |
| LocalBusiness | Hours, address, map panel | Yes for physical premises |
| JobPosting | Google Jobs listing | Yes |
| VideoObject | Thumbnail, key moments | Yes for video |
| FAQPage | Almost never since 2023 | Rarely |
| HowTo | Retired on desktop and mobile | No |
Breadcrumb is the one to add first everywhere. It is a few lines, it always displays, and it replaces an ugly URL in the result with a readable path — which measurably affects click-through.
What happened to FAQ and HowTo
These are the two most widely recommended types and both were curtailed.
In August 2023 Google restricted FAQ rich results to "well-known, authoritative government and health websites" and retired HowTo entirely. The markup is still valid and Google still parses it; it simply stops producing the visual result that was the reason to add it.
A great deal of SEO writing has not caught up, and pages are still built around FAQ schema on the promise of an expanded result that will not appear. Adding it is not harmful — it may still help a machine understand the page, which matters for AI citation — but it is not the rich-result win it is sold as.
Why add it at all, then
Two reasons that survive the rich-result question.
Machine comprehension. Structured data states explicitly what a page is about, who wrote it, when it was updated and what it relates to. Search engines and language models both use that, and it removes ambiguity that inference has to resolve.
Eligibility, not entitlement. Valid markup makes a page eligible for a result. Nothing guarantees it. But a page without the markup is not eligible at all, which is a worse position.
So the useful framing is: add the types that display, add Article and Breadcrumb everywhere as a baseline, and treat anything beyond that as optional rather than as work owed.
JSON-LD, and getting it right
Google prefers JSON-LD over microdata or RDFa, and it is easier to maintain because it sits in a script tag rather than being woven through the HTML.
The failures are consistent and worth checking for.
- Markup describing content that is not on the page. This is a policy violation, not a technicality — a rating in schema that appears nowhere visible can trigger a manual action.
- Invented aggregateRating. The most common version of the above. If there is no genuine rating mechanism, there is no rating to declare.
- Missing required properties. Each type has them; without them the markup parses and produces nothing.
- Multiple conflicting blocks on one page describing the same entity differently.
- Dates in the wrong format. ISO 8601, or it is ignored.
That second point is worth being firm about. Fabricated review counts are one of the more reliable ways to attract a structured-data manual action, and the recovery is slow.
Testing
Two tools, and they answer different questions.
Google's Rich Results Test tells you whether a page is eligible for a specific rich result. That is the one that matters — it reflects what Google will actually do.
The Schema Markup Validator checks the markup against the schema.org vocabulary. Useful for correctness, and it will happily validate markup Google has no interest in.
A page can pass the validator and produce nothing, which is exactly the FAQ situation. Check the Rich Results Test before concluding that markup is doing anything.
A sensible baseline
- Organization once, on the homepage. Name, logo, URL, contact.
- WebSite with SearchAction if you have site search.
- BreadcrumbList on every page below the top level.
- Article or BlogPosting on every editorial page, with real published and modified dates.
- Then whichever type matches what you actually are — Product, LocalBusiness, Event, JobPosting.
That is most of the available benefit. The Schema Markup Generator produces valid JSON-LD for these, and generating from your real page data rather than typing it by hand is what stops the markup drifting away from the content it describes.
Frequently asked questions
Does FAQ schema still work?
Rarely. In August 2023 Google restricted FAQ rich results to well-known authoritative government and health sites. The markup remains valid and is still parsed, but on most sites it no longer produces the expanded result that was the reason to add it.
Is HowTo schema worth adding?
No — Google retired HowTo rich results on both desktop and mobile. A lot of SEO advice still recommends it. The markup is harmless but produces nothing visible.
Which schema should every site have?
Organization once on the homepage, WebSite with SearchAction if you have site search, BreadcrumbList on every page below the top level, and Article on editorial pages. Breadcrumb is the highest-value quick win — it always displays and replaces a URL with a readable path.
Can schema markup hurt my site?
Yes, if it describes content that is not visible on the page. That is a policy violation rather than a technicality, and invented ratings are the most common form. Fabricated review counts reliably attract a manual action.
What is the difference between the two testing tools?
The Rich Results Test tells you whether Google will show a rich result, which is the question that matters. The Schema Markup Validator checks correctness against schema.org and will validate markup Google has no interest in displaying.
Should I add schema if it produces no rich result?
Sometimes. It still states explicitly what a page is about, which both search engines and language models use, and it costs little when generated from page data. Just do not expect a visual result from a type that no longer produces one.