The 7 schema markup types every blog needs in 2026

Schema markup is JSON-LD structured data, embedded in a page's HTML head, that tells search engines and AI engines what the content is, who published it, and how it fits the site. For a blog in 2026 the question is no longer whether to add it, but which types matter and how they stack on one page.

Google deprecated HowTo and downgraded FAQ rich results in August and September 2023, and those restrictions are still in force in 2026 (Google Search Central). At the same time, content with proper schema is roughly 2.5x more likely to appear in AI-generated answers, and FAQ-structured content is showing 44% higher citation rates in ChatGPT, Perplexity, and Google AI Overviews (Position Digital, 2026). The blog stack must serve both surfaces. The seven types below cover what a serious blog ships in 2026, ranked by leverage.

What we measured

Each schema type was scored on three axes:

• Rich result eligibility. Whether Google still surfaces a visual rich result for it in 2026.
• AI citation utility. Whether ChatGPT, Claude, Perplexity, or Google AI Overviews use it as a structural signal during answer construction.
• Implementation cost. The amount of code per article and the maintenance burden over time.

Rich result eligibility comes from Google's official documentation. AI citation utility draws on recent industry analyses and the Princeton GEO study (Aggarwal et al., KDD 2024). Implementation cost is what we measure on actual blog deployments.

1. Article (or BlogPosting)

The spine. Every blog post needs an Article or BlogPosting block. The two are interchangeable for most use cases, but BlogPosting is the more specific subtype and is preferred for time-sensitive posts and personal insights. Article fits formal evergreen content like explainers and reference pages.

Google's recommended properties are headline (max 110 characters), image, datePublished, dateModified, author, and publisher (Google Search Central). Add mainEntityOfPage to declare the canonical URL, and description to shape the AI snippet. Without this block, no other schema on the page has a clear parent entity to attach to.

2. BreadcrumbList

Tells search engines and AI engines the navigational path from the site root to the current article. The visual rich result, where Google replaces the raw URL with a hierarchy like Studio > Blog > GEO, is one of the few rich results still consistently shown for content sites in 2026 (Schema.org).

For a blog post that lives under /blog/[slug], the BreadcrumbList carries three items: home, blog index, current article. Each item needs position, name, and item (the absolute URL). Tiny block, always present, low maintenance.

3. Organization

Identifies the publisher entity. Goes in the site-wide layout, not in each article. Google uses it to attach a knowledge panel and a logo to your search results. AI engines use it to disambiguate when a query mentions the brand by name.

Properties that pull weight: name, url, logo, sameAs (an array of authoritative external profiles like LinkedIn, GitHub, Crunchbase). The Article block's publisher field references this Organization by @id, so the two must agree. One mismatched id and the chain breaks.

4. Person (the author)

In 2026 authorship signals carry more weight in both classic search and AI citation scoring than in any year prior. A blog publishing under named authors should ship a Person block per author, hosted on a dedicated author page, and reference it from each article's author field.

Properties that matter: name, url (the author profile page), sameAs (Twitter/X, LinkedIn, ORCID, personal site), jobTitle, worksFor (linking back to the Organization). This is the chain that builds verifiable expertise: article to author to outside profiles. AI engines that score E-E-A-T weight this chain explicitly.

5. WebSite (with SearchAction)

One block, site-wide, in the homepage layout. Two jobs. First, declares the site name Google uses in search results. Second, when paired with a potentialAction of type SearchAction, it can trigger Google's sitelinks searchbox feature, which places a search box directly under your branded query result.

Skip the WebSite block and Google falls back to your domain name as the displayed site name, which often looks worse. The block is small. The win is small but compounding on branded searches.

6. FAQPage (AI utility, not rich result)

Since August 2023, Google only shows FAQ rich results for well-known authoritative government and health sites (Search Engine Journal, 2023). For everyone else, the rich result is gone. Adding the markup does not produce rich results and does not penalise the page.

Ship FAQPage anyway when the article carries a real FAQ block. AI engines extract structured Q&A blocks more cleanly than they extract free-form prose, and recent industry data shows FAQ-marked content earning 44% higher citation rates in AI answers (Position Digital, 2026). The rule: only emit FAQPage when the questions are real and not duplicates of body H2s. Empty or padded FAQ blocks dilute the signal. We argue the same on the editorial side in our piece on GEO best practices.

7. HowTo (AI utility only)

Google deprecated HowTo rich results across mobile and desktop in August and September 2023 (Schema App, 2023). The visual format is gone. The schema itself is not.

For procedural articles ("how to deploy", "how to migrate", "how to configure"), HowTo gives AI engines a clean step-by-step structure to extract. Each step carries a name, text, and optional url anchored to the in-page section. The format earns its place when the article is genuinely procedural; it adds noise when forced onto explanatory content.

How to stack them on one page

A typical blog article in 2026 ships four to six JSON-LD blocks in the <head>, each in a separate <script type="application/ld+json"> tag. Order does not matter. What matters is that the @id references line up: the Article's publisher points to the Organization's @id; the Article's author points to the Person's @id; the BreadcrumbList declares its own items inline.

A minimum viable stack is Article, BreadcrumbList, Organization, Person. A full stack adds FAQPage when the article has real Q&A, HowTo when it is procedural, and WebSite at the site level. Validate the result on Schema.org Validator and Google's Rich Results Test before shipping.

Comparison at a glance

The seven types side by side, in the order to ship them:

• Article / BlogPosting. Per article. Rich result: yes. AI utility: high. Required.
• BreadcrumbList. Per article. Rich result: yes. AI utility: medium. Required.
• Organization. Site-wide. Rich result via knowledge panel. AI utility: high. Required.
• Person. Per author. No direct rich result. AI utility: high (E-E-A-T). Required for named-author blogs.
• WebSite. Site-wide. Rich result via sitelinks searchbox. AI utility: low. Strongly recommended.
• FAQPage. Conditional. Rich result restricted. AI utility: high. Ship when a real FAQ exists.
• HowTo. Conditional. No rich result. AI utility: high. Ship for procedural articles only.

How to choose what to ship

If the blog is brand new and ships under one author, the first four types (Article, BreadcrumbList, Organization, Person) cover the foundation in roughly 60 lines of JSON-LD per article plus the site-wide blocks. Add WebSite the same day. FAQPage and HowTo wait until an article actually warrants them.

If the blog already has 20 to 50 published posts and zero structured data, retrofit Organization and WebSite first because they are site-wide, then walk every article and add Article plus BreadcrumbList plus Person. Only after all four are present is it worth introducing FAQPage and HowTo on the articles that genuinely carry that shape.

What you skip on a typical blog: Product, Review, Recipe, VideoObject, Course. They have their own rich result eligibility but rarely apply to a content blog. Add them only if the article is literally about a product, a review, or a recipe.

The maintenance question

Schema markup is not write-once. Google deprecates types (HowTo, FAQ rich results, others through 2026), introduces new properties, and tightens enforcement on AI-generated articles. Two habits keep the stack alive: validate on every deploy, and re-check Google's Search Central blog every quarter for retired or expanded types. The cost of stale schema is silent, the gain from fresh schema compounds. The broader playbook lives in our explainer on GEO vs SEO.

Sources

• Google Search Central: Article structured data
• Google Search Central: Changes to HowTo and FAQ rich results
• Google Search Central: FAQPage structured data
• Schema.org: BreadcrumbList
• Search Engine Journal: Google downgrades visibility of HowTo and FAQ rich results
• Schema App: How-to rich results removed on Google Search
• Position Digital: AI SEO Statistics 2026
• Schema.org Validator