HowTo Schema

What is HowTo Schema?

HowTo¹ is the schema.org JSON-LD type for marking up step-by-step instructional content. Each HowTo entity contains a step array of HowToStep objects, each with a name, text, and optional image or url. The structure mirrors how a reader actually executes the procedure: discrete numbered actions, each independently parseable.

Minimum viable HowTo JSON-LD

{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to validate JSON-LD before deploy",
  "step": [
    {
      "@type": "HowToStep",
      "name": "Paste into the validator",
      "text": "Copy the rendered JSON-LD output and paste into Google's Rich Results Tester."
    },
    {
      "@type": "HowToStep",
      "name": "Check both validators",
      "text": "Repeat the paste in Schema.org's validator; they flag different issues."
    }
  ]
}

Status in 2026

Demoted in SERPs, potentially useful for machine readability. Google announced changes to HowTo and FAQ rich results on August 2, 2023²; HowTo rich results were removed from mobile SERPs that month and from desktop SERPs in the months that followed (the same wave that initially restricted FAQ rich snippets, which were later fully deprecated for all sites on May 7, 2026). The HowTo schema type is still valid and may still be parsed by AI engines, but no longer drives in-SERP visual treatment for most categories. The reasonable practice in 2026: ship HowTo on genuinely procedural content because step-by-step structure is good for human readers and machine-readability, and practitioners report AI Overview and ChatGPT surfacing structured procedural content; vendor confirmation that HowTo schema is a citation ranking signal is absent.

Like FAQ Schema, HowTo schema sits in vendor-canonical territory: Google and schema.org both publish official documentation for the type. The 2026-05-17 ChatGPT probe of the FAQ Schema entry illustrates what to expect for similar queries against this entry: AI engines tend to retrieve third-party glossary entries on vendor-canonical schema concepts but prefer vendor sources in citation slots. The strategic implication for HowTo content programs is the same as for FAQ Schema: clarity of prose alone will not displace Google's primary source.

How to apply

HowTo is straightforward when you have genuinely step-ordered content. Three concrete moves:

• Only mark up pages with real procedural content: a page like "5 things to do when X happens" is informational, not procedural; use Article schema. HowTo is for "do A, then B, then C" where order matters. Mismarking risks being flagged as misused markup.
• Give each step a meaningful name: practitioners report that AI engines often quote just the step name when condensing a HowTo. Whether step names are structurally privileged in the markup or simply double as concise summary phrases regardless of markup has not been isolated. Either way, "Step 1" is unhelpful; "Paste into the validator" carries information for readers and is easier for automated systems to interpret.
• Validate via the Rich Results Tester: Google's validator still accepts HowTo markup even after the rich-result treatment was demoted. It confirms the markup parses correctly; it does not imply SERP rich-result eligibility (that has been deprecated) or AI citation value (which is not vendor-documented). Treat the validator as a parse-correctness check, not a downstream-value indicator.

What to skip: HowTo on FAQ-shaped content (use FAQPage). Deliberate schema-type-to-content-type misuse can trigger Google's "Spammy structured markup" manual action; whether mismatched (but well-intentioned) pairings are independently down-weighted by AI engines is plausible but not vendor-documented.

How it relates to other concepts

• Sibling JSON-LD type to FAQ schema and DefinedTerm schema.
• Direct expression of procedural answer blocks. Each HowToStep is one extractable unit.
• Common pairing with Article schema on long-form how-to guides: Article for the wrapper, HowTo for the embedded procedure.
• Relevant to AEO for "how do I X" queries. AI Overview and voice assistants commonly surface procedural answers, and practitioners report HowTo-marked content appearing in such responses; no major vendor has documented HowTo schema as a citation ranking signal, so the role of the markup itself (vs the underlying procedural structure of the content) is not isolated.