Speakable schema is structured data that marks specific sections of your page as high-quality candidates for voice-based AI answer delivery. Instead of describing the content like most schema types, Speakable uses CSS selectors or XPath expressions to point at existing HTML elements - telling Google "this specific section of my page is the best part to read aloud when answering a voice query." It's one of the lightest-weight schema types to implement because you don't need to duplicate any content in JSON-LD.
Currently, Google's Speakable feature is in limited beta primarily for approved Google News publishers using NewsArticle schema. But implementing Speakable on any web page is still recommended - it acts as a permanent content quality signal that identifies your most answer-ready paragraphs, and positions your site for when Google expands voice feature eligibility. The implementation effort is minimal (adding one schema block to your page template), and the signal is durable regardless of whether Google's voice features expand in scope.
Speakable Implementation: CSS Selector vs XPath
Speakable schema uses two selector syntax options - CSS selectors and XPath - to point at specific sections of your HTML content rather than duplicating text in JSON-LD. Choose the approach that best fits your CMS and template architecture.
CSS Selector approach
CSS selectors target elements by class, ID, or structural pattern - the most maintainable approach when your CMS generates consistent class names for article summaries and definition blocks.
{
"@context": "https://schema.org",
"@type": "WebPage",
"name": "What is FAQPage Schema?",
"url": "https://acme.com/aeo/structured-data/faq-schema",
"speakable": {
"@type": "SpeakableSpecification",
"cssSelector": [
".article-lede",
"#key-takeaways",
".definition-block",
".voice-summary"
]
}
}Advantages
Limitations
Is Your Content Speakable-Ready? Pre-flight Checklist
Before marking a content section with Speakable schema, verify it meets these criteria. Check each item that your content satisfies.
Speakable content quality score
0/6