JSON-LD schema types that still work in 2026

JSON-LD is still the format Google prefers for structured data. The mistake in 2026 is shipping types that no longer have a Search feature, or inventing fields the page does not support.

A website audit should validate what is on the page, then recommend only types you can prove.

Which schema types should a SaaS site use?

For a product like Siteory, the honest graph is small:

• Organization for the company, with name, url, logo, and email
• WebSite for the origin
• SoftwareApplication for the product, including real offers
• WebPage or a more specific subtype on inner URLs
• BreadcrumbList on nested paths
• BlogPosting on articles, with dates and a publisher

That graph is enough for entity clarity. It is not a rich-result shopping list.

SoftwareApplication needs an operating system (Web is fine), a category, and offers that match pricing. If a plan is $9, the markup should say 9, not 0.

What Google retired and what you should do about it

HowTo rich results were removed in September 2023. Do not add HowTo. If you still have it, removing it is cleanup, not an emergency.

FAQPage rich results were retired for every site on May 7, 2026. Existing FAQ markup is an info-level finding. Do not add new FAQPage to chase a star or a dropdown in Search. If you have a genuine user Q&A thread, QAPage is the type that still describes that page.

CourseInfo, EstimatedSalary, LearningVideo, SpecialAnnouncement, and VehicleListing also left Search features in 2025. They are irrelevant to a website audit product unless you operate those businesses.

Keep Article and BlogPosting. Keep BreadcrumbList. Keep Organization. Those types still describe the page even when they do not paint a special SERP layout.

Rules that fail validation more than exotic types

Most schema bugs are mechanical:

• Missing @context
• Relative URLs in url, image, or logo
• Dates that are not ISO 8601
• Offer prices that do not match the visible page
• Review markup for reviews you do not have
• A rating on a node without a valid @type

Google can generate structured data from JavaScript, but time-sensitive offers should be in the first HTML response. Siteory emits JSON-LD from the server for that reason.

Do not mark up fake reviews. Do not put a 5.0 AggregateRating on a product with no public ratings. That is a policy problem, not an SEO tweak.

Breadcrumbs, articles, and canonical agreement

BreadcrumbList should match the visible trail and the URL path. Home > Features > SEO audit is correct for /features/seo-audit. A breadcrumb that says Home > Blog for a pricing page is a defect.

BlogPosting needs a headline, dates, and a publisher. The author can be an Organization when a named person did not write the piece. Use Person when a person did. Do not invent an author.

The mainEntityOfPage URL must be the canonical URL. If the article lives at /blog/json-ld-schema-types-that-still-work, do not point schema at a tracking URL.

How Siteory validates JSON-LD

The SEO audit extracts script type="application/ld+json" blocks from the crawled HTML. It reports parse errors, missing types that the page clearly is, and deprecated types that no longer have a Search feature.

It will not invent LocalBusiness markup for a company with no address. It will not tell you to add FAQPage. It will tell you when the homepage has no Organization node and when an article is missing a published date.

If you want the current allow list in one place: Organization, WebSite, SoftwareApplication, WebPage, AboutPage, ContactPage, CollectionPage, BlogPosting, BreadcrumbList. That is the set Siteory uses on siteory.co, and it is the set the audit is willing to recommend.