Complete Guide to Speakable Schema — AEO-Direct Structured Data

What Is Speakable Schema

Speakable schema specifies content areas on a page suitable for voice reading or AI summarization and citation. It delivers a structured data signal: "this part of the page is the core summary—take your answer from here."

Unlike other schemas that report content attributes (author, date, product info), Speakable fundamentally differs by specifying which area of content matters.

⚠️ Google's Current Status — Limited Beta

Official support for Korean content is not specified as of June 2026.

TL;DR

Speakable schema: ① Google beta, limited (US English only) ② Specify speakable area with cssSelector or xpath, not both ③ Nest as speakable property in Article or WebPage schema ④ Immediate Google voice citation effect for Korean sites unconfirmed, but low implementation cost makes adoption worth considering from an AEO perspective ⑤ Apply only to core areas of 20–30 seconds reading (2–3 sentences).

Decisive Difference from Other Schemas

Significance from an AEO Perspective

Speakable schema explicitly tells search engines and AI systems "which part of this page is suitable as a direct answer."

When AI answer engines crawl pages, they do not treat all text equally. Areas marked with Speakable become a signal that "this site considers this the important summary area." Specifying headline, intro summary, and conclusion with Speakable may favor AI answer extraction.

However, there is no official evidence that Speakable guarantees AI citation. Evaluate adoption based on potential AEO value.

Implementation: cssSelector vs xpath

Speakable specifies speakable areas in two ways. Choose one only; do not mix them.

cssSelector (recommended — more intuitive):

"cssSelector": [".article-summary", "h1", ".tldr-box"]

xpath:

"xPath": ["/html/head/title", "/html/body/article/p[1]"]

Google official guide describes cssSelector as "using class attributes" and xPath as "using XML path notation."

Minimal Example (cssSelector)

{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "name": "What Is AEO",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": ["h1", ".article-summary"]
  }
}

Recommended Example (Nested in BlogPosting)

{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "What Is AEO: Content Strategy for the AI Search Era",
  "datePublished": "2026-06-04T09:00:00+09:00",
  "author": {
    "@type": "Person",
    "name": "Jane Smith"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Kroffle"
  },
  "inLanguage": "ko-KR",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [
      "h1",
      ".article-tldr",
      ".article-definition"
    ]
  }
}

In this example, h1 (headline), .article-tldr (TL;DR box), and .article-definition (definition section) are specified as speakable areas. Those classes must exist in the actual HTML.

Content Writing Guide (Speakable Areas)

Google official documentation recommends the following for Speakable areas:

• Limit to 20–30 seconds (about 2–3 sentences)
• Exclude dates, photo captions, source attribution, etc. from Speakable areas
• Clear sentences understandable by voice alone
• Do not include the full article; core summary only

Adaptation for the Korean Market

Google Speakable currently supports US English content only. Applying it to Korean sites is unlikely to produce Google voice search effects at present.

However, adoption can be considered for these reasons:

1. Low implementation cost (add a few lines to existing Article schema)
2. Potential for expanded Korean support in the future
3. May be used by non-Google AI systems for content structure understanding

Common Mistakes

Wrong example: Mixing cssSelector and xpath

"speakable": {
  "@type": "SpeakableSpecification",
  "cssSelector": ["h1"],
  "xPath": ["/html/body/p[1]"]
}

Correct example: Choose one only

"speakable": {
  "@type": "SpeakableSpecification",
  "cssSelector": ["h1", ".article-summary"]
}

Wrong example: Specifying the full article as Speakable

"cssSelector": [".article-full-content"]

Correct example: Specify core summary areas only

"cssSelector": ["h1", ".tldr", ".definition"]

Validation Methods

1. Schema Markup Validator: https://validator.schema.org — verify Speakable schema structure validity
2. Actual behavior verification: Verifiable only within Google's official support scope (US English). Direct verification is currently difficult in Korean environments.

Frequently Asked Questions

Q. Should I use Speakable schema now that it's in beta?
A. Immediate effects for Korean sites are hard to expect given Google's official support is limited to US English. However, adding a few CSS selectors to existing Article schema is low cost, and value may increase as the AEO landscape evolves, so adoption is worth considering.

Q. What happens if cssSelector classes don't exist?
A. The schema is valid, but if selectors don't match actual HTML, there is no effect. After writing schema, confirm the classes exist in HTML.

Q. Must I add Speakable to every page?
A. Prioritize informational content suitable for voice and AI answers (What is~, How to~, guides, etc.). Unnecessary for product listings, form pages, etc.

Q. What is the difference between Speakable and FAQPage?
A. FAQPage reports question-answer pair content. Speakable reports which CSS areas are suitable for speech. Both schemas can be used together.

References

• Google official Speakable documentation: https://developers.google.com/search/docs/appearance/structured-data/speakable (verified June 2026, beta and limited status noted)
• Schema.org SpeakableSpecification: https://schema.org/SpeakableSpecification