Free Tool — No Login Required

Schema Markup for
AI Engines & Voice Search

FAQ schema no longer shows rich results in Google (May 2026) — but it’s still essential for ChatGPT, Perplexity, Gemini, and voice assistants. Generate the right markup here, free and instant.

ChatGPT Perplexity Gemini Google Assistant Siri Bing Copilot
FAQPage Schema
FAQPage schema helps AI engines like Perplexity and ChatGPT extract and cite your answers directly. While Google deprecated FAQ rich results in May 2026, AI engines still use this data heavily.
Speakable schema tells voice assistants and AI search engines exactly which sections to read aloud or summarize. Best practice: target your intro paragraph and key answer sections.
Common: h1 .intro-paragraph #key-answer
Power User Package — generates FAQPage + Speakable + Article schema combined into one JSON-LD block. One paste gives you maximum AEO coverage across all AI engines.
Page Info
Speakable Selectors
FAQ Pairs
JSON-LD Output — Live Preview 0 lines · 0 chars 
// Fill in the form on the left to generate your schema →
AI Citation Readiness Score 0/100
Not Started
  • →Add at least 3 Q&A pairs for maximum AI citation coverage.
  • →Fill in your page URL to add an @id and improve attribution.
  • →Add Speakable selectors to optimize for voice search assistants.
AI
AEO Insider Editorial Updated May 2026 · 8 min read · GEO & AEO

Schema Markup for AI Engines: The Complete 2026 Guide

Schema markup for AI engines is structured data — written in JSON-LD and added to your page’s <head> — that tells AI systems like ChatGPT, Perplexity, and Gemini exactly how to read, extract, and cite your content. In 2026, it is the single most direct technical lever you have for improving how AI engines reference your site.

What Changed With FAQ Schema in May 2026?

Google officially deprecated FAQPage rich results for the vast majority of websites in May 2026. Only a narrow band of authoritative government and health sites retain eligibility. The accordion-style FAQ snippets that SEOs relied on for years are gone for almost everyone.

Most content teams stopped here and concluded FAQ schema was dead. That is the wrong conclusion. The deprecation only affects Google’s visual display layer — the rendered snippet in search results. The underlying structured data is still actively parsed and used by AI engines and voice platforms. Those platforms are now the primary battlefield for content visibility, and they have not deprecated anything.

Key distinction

Google deprecated the rich result display of FAQPage schema — not the schema itself. Perplexity, ChatGPT, Gemini, and voice assistants still read and use FAQPage markup to extract and cite answers. The audience for your schema has shifted, not disappeared.

How AI Engines Use Schema to Select and Cite Sources

AI engines like Perplexity and ChatGPT do not rank pages the way Google does. They extract content and decide which sources to cite based on a combination of signals: domain authority, content clarity, entity recognition, and — critically — structured data. Schema markup gives AI engines an unambiguous, machine-readable summary of your content without requiring them to interpret raw HTML.

When an AI engine encounters FAQPage schema, it reads your Q&A pairs as pre-packaged, citable answers. When it encounters Speakable schema, it treats the targeted CSS selectors as editorial guidance on which sections represent your most authoritative statements. When it encounters HowTo schema, it has a structured, step-by-step process it can summarise in a voice response or AI overview. Each schema type is a direct instruction to the AI — and most sites give none.

This is why schema markup for AI engines is not optional in 2026. Sites that implement it correctly are giving AI systems explicit citation-ready content. Sites that don’t are leaving that decision entirely to algorithmic heuristics.

The Four Schema Types That Matter for AI Search in 2026

Not all schema types carry equal weight with AI engines. These four are the highest-priority for citation eligibility and voice search visibility:

Most cited by AI

FAQPage

Structures Q&A pairs as direct, machine-readable answers. The primary format AI engines extract for conversational queries. Keep answers under 300 words and start each with a direct statement.

Voice & AI summaries

Speakable

Uses CSS selectors to explicitly mark your most authoritative paragraphs. Tells voice assistants and AI engines which sections to read aloud or include in summaries. The most underused schema type in AEO.

Process queries

HowTo

Structures step-by-step processes with time, cost, tools, and supplies. High-value for voice search and AI Overviews on procedural queries — “how do I”, “how to”, “steps to”.

Maximum coverage

Combined @graph

FAQPage + Speakable + Article schema in a single JSON-LD block using @graph architecture. One paste covers all AI citation surfaces simultaneously. The recommended approach for important pages.

Why Speakable Schema Is the Most Underused Signal in AEO

Answer Engine Optimization (AEO) is the practice of structuring your content so AI engines can extract, attribute, and cite it accurately. Speakable schema is your most direct signal to these systems — it uses CSS selectors to mark exactly which paragraphs on your page represent your definitive, citable answer.

Without Speakable markup, AI engines fall back to heuristics: they look at heading proximity, paragraph position, and word density to guess which content is most relevant. With Speakable markup, you eliminate the guesswork entirely. You tell the system: this paragraph is the answer, cite this. The practical effect is measurably higher citation rates in AI-generated summaries, Google AI Overviews, and voice assistant responses from Siri, Google Assistant, and Alexa.

Despite this, Speakable schema appears on fewer than 2% of pages in most content categories. It is the single highest-leverage, lowest-competition schema implementation available to content teams right now — and most are ignoring it entirely. [1]

How to Add Schema Markup for AI Engines in WordPress

There are three reliable methods for implementing schema markup on a WordPress site. Choose based on how your site is set up:

  1. RankMath (recommended): Go to the page editor → RankMath sidebar → Schema tab → Add Schema → Custom Schema. Paste the JSON-LD output from this tool directly — no script tags required. RankMath handles the injection into <head> automatically.
  2. WPCode plugin: Install the free WPCode plugin (formerly Insert Headers and Footers). Create a new snippet, set type to JavaScript, paste your JSON-LD wrapped in <script type="application/ld+json"> tags, set location to Header, and scope it to the specific page URL. This method bypasses WordPress content filters entirely.
  3. Custom HTML block: In the Gutenberg editor, add a Custom HTML block and paste your JSON-LD wrapped in script tags. This works for admin users with unfiltered HTML capability, but note that some security plugins may strip inline scripts on save.

Whichever method you use, validate the output at the Google Rich Results Test and Schema.org Validator before considering it live. Also check that Perplexity and ChatGPT can access your page — both require pages to be publicly indexed and not blocked by robots.txt.

Common Schema Markup Mistakes That Hurt AI Visibility

  • Answers longer than 300 words. AI engines prefer concise, direct answers. Long FAQ answers get truncated or deprioritised in extraction.
  • Questions that don’t match real user queries. Write FAQ questions exactly as your audience would search them — not as internal headings.
  • Starting answers with a question or hedge. Begin every FAQ answer with a direct statement. “Yes, FAQ schema still matters for AI engines because…” outperforms “It depends on…” every time.
  • Speakable selectors pointing to navigation or footer elements. Target only your core content sections — intro paragraphs, key answer blocks, summary sections.
  • No @id in FAQPage schema. Adding your page URL as the @id links your Q&A pairs to your domain as an entity — a critical attribution signal for AI engines.
  • Using separate schema blocks instead of @graph. Multiple disconnected schema blocks are harder for AI engines to process than a single unified @graph containing all entity types.

How to Use This Tool

Select your schema type from the tabs above. Start with the FAQPage tab if you have Q&A content, the Speakable tab if you want to target specific content sections for voice and AI extraction, or the HowTo tab for step-by-step guides. For maximum AEO coverage on a single important page, use the Combined AEO Package — it generates FAQPage, Speakable, and Article schema in a single @graph JSON-LD block.

Use the Load Preset button to see fully populated examples for e-commerce, SaaS, local business, and more. The AI Citation Readiness Score on the right updates in real time and tells you exactly what to improve. A score above 70 indicates strong AI citation eligibility. Use the WordPress Snippet button to get a ready-to-paste PHP function for functions.php or a code plugin. You can also check the AEO Readiness Checker to audit an existing page’s schema implementation, or use the AEO Schema Generator for Article, Author, Service, VideoObject, and WebSite schema types not covered here.

Frequently Asked Questions

What is schema markup for AI engines?
Schema markup for AI engines is structured data written in JSON-LD that tells AI systems like ChatGPT, Perplexity, and Gemini how to read, extract, and cite your content. Unlike traditional SEO schema focused on Google rich results, AI-targeted schema is designed to improve citation rates inside AI-generated answers and voice assistant responses. The most important schema types for AI engines are FAQPage, Speakable, HowTo, and Article.
Does FAQ schema still work after Google deprecated it in May 2026?
Yes. Google deprecated the rich result display of FAQPage schema — meaning the accordion-style snippets in Google search results are gone for most sites. However, AI engines including ChatGPT, Perplexity, Gemini, and Bing Copilot still actively read FAQPage schema to extract and cite answers. For Answer Engine Optimization (AEO), FAQ schema remains one of the most valuable markup types you can implement.
What is Speakable schema and how does it help with AI search?
Speakable schema uses CSS selectors to explicitly mark which sections of a page contain your most authoritative answers. AI engines and voice assistants — including Google Assistant, Siri, and Alexa — use these signals to identify content for extraction and citation. Without Speakable markup, AI systems rely on heuristics to guess your most relevant content. With it, you direct them precisely to your best paragraphs, improving citation rates in AI-generated summaries and voice responses.
Which schema type should I implement first?
Start with FAQPage schema if your page answers specific questions — it is the most directly cited format by AI engines. Add Speakable schema next to mark your intro paragraph and key answer sections for voice and AI extraction. If your page covers a step-by-step process, add HowTo schema. For maximum coverage, use the Combined AEO Package which generates FAQPage, Speakable, and Article schema in a single @graph block — one paste, full coverage across all AI platforms.
How do I add schema markup for AI engines in WordPress?
The three most reliable methods are: (1) RankMath — go to the page editor, open the RankMath sidebar, navigate to Schema → Custom Schema, and paste the JSON-LD directly with no script tags needed. (2) WPCode plugin — create a JavaScript snippet set to run in the header, scoped to your specific page URL. (3) Custom HTML block in Gutenberg — paste the JSON-LD wrapped in script type application/ld+json tags. Validate your implementation using the Google Rich Results Test after publishing.
What is the @graph format and why does it matter?
The @graph format is a JSON-LD architecture that combines multiple schema types — FAQPage, Speakable, Article, BreadcrumbList — into a single structured data block using a shared @context. It is recommended over separate schema blocks because it lets AI engines understand all entity relationships on a page in one read. The Combined AEO Package tab in this tool generates @graph output automatically.
Sources & References
  1. Google Developers — Speakable structured data. Google Search Central documentation on Speakable schema implementation and eligibility.
  2. Schema.org — FAQPage specification. Official specification for FAQPage, Question, and Answer entity types.
  3. Google Search Central Blog — Changes to HowTo and FAQ rich results. Google’s announcement on the restriction of FAQPage and HowTo rich results.
Want Your Full Site AEO-Audited?
AEO Insider offers hands-on schema audits and implementation. We fix your markup, restructure your content architecture, and wire up entity signals for AI citation.
Book a Free Consultation