Add Speakable schema for content designed to be read aloud by AI assistants and voice search.
Mark your best 1-2 sentences per page as Speakable — the sentences that perfectly answer the page's core question in a natural, spoken-aloud tone. Think: "If a voice assistant read one sentence from this page, which would be most useful?"
Look for concise, factual summaries that answer specific questions. Good candidates: definition paragraphs, key statistics, direct answers to common questions. The content should sound natural when read aloud — no jargon, acronyms, or complex sentence structures.
Use JSON-LD with @type: "WebPage" and speakable property pointing to CSS selectors or XPaths of the target sections. Example: "speakable": {"@type": "SpeakableSpecification", "cssSelector": [".summary", ".key-takeaway"]}. Add to your page's <head>.
Speakable content should use: short sentences (under 20 words), active voice, no parenthetical asides, spelled-out numbers ("three hundred" not "300" for small numbers), and no visual references ("see the chart below").
Voice responses need to be brief — 20-30 seconds of audio maximum. That's roughly 2-3 sentences or 40-60 words. Pack the maximum information into the minimum words. Every word should earn its place.
Go to search.google.com/test/rich-results, enter your URL, and verify the Speakable markup is detected and valid. Note: Speakable is currently a Google-supported beta feature primarily for news content, but it signals AI-readiness for all content types.
Generate Speakable schema markup for my page: Page URL: [URL] Page topic: [TOPIC] Sections suitable for AI voice responses: 1. [Section heading and CSS selector or XPath] 2. [Section heading and CSS selector or XPath] Generate: 1. Complete JSON-LD Speakable schema markup 2. Explanation of which sections to mark as speakable and why 3. Best practices for writing "speakable" content (sentence length, clarity, pronunciation)
Track your progress and get guided through every step.
Open Interactive Tool