Speakable Schema: What It Is, Who Needs It, and How to Add It

What speakable schema actually is

Speakable is a structured data type from Schema.org called SpeakableSpecification. Instead of describing your business or a product, it points at specific chunks of text on the page and says "these are the parts a machine should read out loud." You mark the headline and a sentence or two of summary, and a voice assistant knows which words to speak when someone asks for the news by voice.

You attach it to an Article (or NewsArticle) and locate the speakable sections two ways: a CSS selector or an XPath. The selector points at the element on your page that holds the speakable text. You use one method or the other, never both on the same property.

Who it is actually for

This was designed for news publishers. The use case Google built it around is a smart speaker. Someone asks their Google Assistant for news on a topic, the Assistant pulls up to three articles from different publications, and reads the speakable sections aloud using text to speech. If your content is not the kind of thing somebody asks a speaker to read to them, speakable does not have a job to do on your page.

There is one more limit worth stating plainly. Per Google's own documentation, the feature works for users in the United States who publish in English. Google has said it can expand to other countries and languages once enough publishers adopt it, but that is the stated scope today. If your audience is outside the US or you publish in another language, you are building for a door that is not open yet.

The honest scope: this is niche

I am not going to sell you on this one. Speakable has carried a BETA label in Google's documentation since it launched, and it remains tied to Google Assistant news playback rather than ordinary search results. Adding it will not move your rankings, will not earn you a rich result in the search listing, and will not help most sites at all. The honest read is simple: it is a small, optional feature for a specific kind of publisher. Our crawler flags pages that are eligible, meaning they look like article content, so the choice is yours to make rather than a box you must tick.

How to implement it

Add a SpeakableSpecification block inside your Article JSON-LD. Point the cssSelector at the elements holding your headline and summary. Here is a clean example using CSS selectors:

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "NewsArticle",
  "headline": "Your article headline here",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [".article-headline", ".article-summary"]
  },
  "url": "https://seoprocheck.com/your-article/"
}
</script>

Prefer XPath? Swap the cssSelector property for xpath and address the same content through the document structure:

"speakable": {
  "@type": "SpeakableSpecification",
  "xpath": [
    "/html/head/title",
    "/html/body//div[@class='article-summary']"
  ]
}

Mark the key points only. The headline plus a short summary of two or three sentences is the right size. Keep datelines, photo captions, and source credits out of it, because they sound like noise when read aloud with nothing on screen.

How to validate

Run the page through Google's Rich Results Test and the Schema Markup Validator. Confirm the cssSelector or xpath you used actually resolves to the text you meant, since a selector that points at nothing is the most common failure. Check that the speakable text is real text in the HTML, not an image or a chunk injected by script after load. Then read the selected sentences out loud yourself and ask whether they stand on their own without the page around them.

When to bother, and when to skip

Bother if you publish news or news style articles, your audience is in the US, you write in English, and you would genuinely like your stories read aloud through Google Assistant. In that narrow case it is cheap insurance and easy to add.

Skip it if you run a marketing site, a product catalog, a local service business, or anything that is not timely article content. Skip it if your readers are outside the US or you publish in another language. Spending time here when your other structured data is incomplete is effort in the wrong place. Get your Article, Organization, and Breadcrumb markup solid first.

Common mistakes

The usual problems are selectors that match nothing, marking the entire article body instead of the key points, and including both cssSelector and xpath on the same property. People also mark text that only appears after JavaScript runs, which a crawler may never see. And the biggest one of all is expecting it to do something it was never built to do. Speakable feeds voice playback for news. It is not a ranking lever.

FAQ