Is the HowTo schema still worthwhile even though Google has reduced the rich results?

Yes. Even without the former instructional blocks in classic search, the value remains machine readability: AI systems and crawlers understand your steps more unambiguously. For AI visibility, a clean structure is still helpful, even if it guarantees no placement.

How does the HowTo schema differ from the FAQ schema?

The FAQ schema marks up individual question-answer pairs that stand independently of one another. The HowTo schema describes an ordered sequence of steps with a shared goal. Rule of thumb: if it is about a sequence to achieve a result, you use HowTo, otherwise FAQ.

HowTo schema

HowTo schema is a schema.org markup format you add to a page's code to identify a set of instructions as an ordered sequence of steps, each with its own text, and optionally its materials, tools, duration, and cost. It doesn't change what a visitor sees. It gives crawlers and AI systems a machine-readable map of a process that would otherwise have to be inferred from paragraph text and photo captions.

Why HowTo schema matters for AI visibility

A large share of what people ask AI assistants is procedural: "how do I do X." HowTo markup labels each step, so a model parsing your page doesn't have to guess where one instruction ends and the next begins. That matters more now than it used to: an Ahrefs study found only 6-8% of URLs cited by ChatGPT overlap with a query's Google top 10, and around 80% of ChatGPT citations don't rank in the top 100 at all — citation is a separate selection process from ranking, and machine-readability is one of the few levers you control directly on the page itself. Be clear about what markup does and doesn't buy you, though: Google's own guidance for AI Overviews and AI Mode states no special schema is required, and structured data is not documented as an independent ranking or citation factor. Treat HowTo schema as an aid to correct interpretation, not a mechanism for getting picked.

How HowTo schema works technically

You implement it as JSON-LD, a script block in the page's head or body that doesn't touch the visible layout. The container is an object of type "HowTo" with a name for the overall task and an ordered array of "HowToStep" entries, each carrying its own step text and, optionally, an image or a link to the step's location on the page. You can layer in "HowToSupply" for materials, "HowToTool" for equipment, total time, and estimated cost. The one rule that isn't optional: the markup has to mirror what a visitor actually sees on the page. Steps that exist only in the JSON-LD and not in the visible content are a structured-data policy violation, and engines that catch it will disregard or penalize the markup.

Common mistakes

The most common failure is markup with nothing behind it: a HowTo block sitting on a page that doesn't actually lay out numbered steps for a reader. Close behind that is thin steps — entries with a title but no real instructional text, which give a parser an empty shell instead of content. Teams also conflate HowTo with FAQ schema: FAQ is for independent question-and-answer pairs, HowTo is for an ordered sequence that builds toward one outcome, and the two aren't interchangeable. Worth knowing before you invest heavily here: Google has pulled back the HowTo rich result from classic search results, so the payoff today is mostly the machine-readability benefit described above, not a visual result block. Validate your markup with a structured-data testing tool before you publish it, every time.

Relation to generative engine optimization (GEO)

GEO, the practice of optimizing content to be surfaced and cited by AI systems, was named and studied in the 2024 paper "GEO: Generative Engine Optimization" (Aggarwal, Murahari, Narasimhan, Deshpande, Rajpurohit, and Kalyan; ACM SIGKDD 2024). HowTo schema sits inside that practice as one supporting technique, not the core of it. It doesn't earn a citation on its own; it removes ambiguity from content a model is already inclined to draw on. Ahrefs analysis across roughly 75,000 brands found that how often a brand is mentioned across the web correlates with AI citation rate at about 0.664 — roughly three times the correlation seen for backlinks at 0.218. That's the bigger lever. Structured data supports substance; it doesn't replace it.

Example

Picture a bike shop in Leipzig publishing a guide, "Clean your bicycle chain in 5 steps." On the page, each step gets its own paragraph and photo: slacken the chain, brush off grime, apply degreaser, rinse, re-oil. In the source code, a JSON-LD block of type HowTo lists five HowToStep entries plus the required tools (brush, rag) and supplies (chain cleaner, oil). If someone asks an AI assistant how to clean a bike chain, the model can parse the five discrete steps cleanly and, if it draws on the page, attribute the shop as the source. This is an illustrative scenario, not a documented case study.

Common questions

Is HowTo schema still worth implementing now that Google has scaled back the rich result?

Yes, but for a different reason than before. The visual search-result block is largely gone, so the case for HowTo schema today rests on machine-readability for AI systems and crawlers, not on winning a rich snippet. Google is explicit that no special schema is required for its AI features, so treat this as a clarity aid you layer on top of genuinely good instructional content, not as a requirement.

How is HowTo schema different from FAQ schema?

FAQ schema marks up standalone question-and-answer pairs that don't depend on each other or on order. HowTo schema marks up an ordered sequence of steps that build toward a single outcome. If the content is a series of separate questions, use FAQ; if it's a process someone follows start to finish, use HowTo.