HowTo schema guide for AI search
By Abhijay Tondak, Founder · Updated July 1, 2026 · 5 min read
HowTo schema is structured data that marks up a set of step-by-step instructions as a procedure, telling engines your content is a tutorial with ordered steps, supplies, and an end result. It helps engines parse and present your instructions for 'how do I do X' queries. Implement it to mirror your visible numbered steps exactly, include the step text and any tools/materials, and validate it - like all schema, it supports well-structured on-page steps rather than replacing them.
Key takeaways
- HowTo schema marks up instructions as an ordered procedure engines can parse and present.
- It maps to 'how do I do X' queries, one of the most common AI question types.
- Key properties: the ordered steps (name/text), plus tools, supplies, and total time where relevant.
- Markup must mirror the visible numbered steps on the page.
- It supports clear on-page steps; it doesn't substitute for them.
What HowTo schema does
HowTo structured data explicitly tells engines 'this content is a procedure' and lays out its ordered steps in machine-readable form. For instructional queries - 'how to set up X', 'steps to do Y' - this helps engines understand your tutorial as a sequence they can extract and potentially present as steps. It reinforces the extractable structure that already makes how-to content citation-rich.
Key properties
Capture the procedure's essentials:
- step: each step as an ordered item with clear text (and optional name).
- tool and supply: what the reader needs, where relevant.
- totalTime: how long the whole procedure takes, when useful.
- An overall name and description matching the page's title and intro.
Mirror the visible steps
Your HowTo markup should reflect the actual numbered steps shown on the page - same steps, same order. Marking up steps that don't appear, or in a different sequence, is the kind of mismatch that gets schema ignored. The schema and the visible content should be two representations of the same procedure.
Use it where it fits
HowTo schema is right for genuine step-by-step procedures - not for every article. Applying it to content that isn't really a sequential how-to is misuse that engines discount. Where you do have a real tutorial, pair clean numbered on-page steps with validated HowTo markup for the strongest extractable result, and confirm it with a structured-data testing tool.
Frequently asked questions
When should I use HowTo schema?
For genuine step-by-step procedures - real tutorials with ordered steps. Don't apply it to content that isn't sequential instructions; misapplied schema gets discounted. Where you have a true how-to, it reinforces the extractable structure.
What are the key HowTo properties?
The ordered steps (each with clear text), plus tools, supplies, and total time where relevant, and an overall name/description matching the page. The steps are the core.
Does HowTo schema guarantee rich results?
No schema guarantees a specific presentation - engines decide. HowTo markup helps engines understand and potentially present your steps, but the reliable win is that it reinforces clean, extractable on-page structure.
Must the markup match the on-page steps?
Yes - same steps, same order as shown on the page. Mismatched or invisible steps in markup get ignored. Schema and visible content should represent the same procedure.
Put this into practice — free.
Get your free AI-visibility audit and see where engines find you today.
More from this topic
Keep building your expertise with related GEO content in the same cluster.
Structured data (JSON-LD) for AI search
Structured data helps AI engines understand and cite your pages. Here are the JSON-LD schema types that matter for AI search and how to implement them.
ReadHow to write a TL;DR that gets cited
A citable TL;DR answers the page's core question in 1-3 self-contained sentences at the top. Here's how to write one AI answer engines will lift verbatim.
ReadWhy original data and statistics win AI citations
Original statistics and data give AI answer engines something concrete and attributable to cite. Here's why proprietary data outperforms recycled claims in GEO.
Read