# Speakable

## Quick answer

Speakable is a structured-data markup from schema.org that flags specific sentences or sections of a page as suited for reading aloud by a voice assistant or AI system. It is implemented the same way other schema markup is, by tagging particular content blocks rather than the whole page.

## Why it matters

Speakable was designed for news content read aloud by voice assistants, not as a general mechanism for getting quoted by AI answer engines, so a brand should not expect it to directly drive citations in ChatGPT, Claude, Perplexity or similar engines. It is worth adding where it fits because it is cheap and machine-readable, but it should not replace the more direct work of writing clear, well-structured answers on the page itself. The gap between what speakable actually does and what growth content often implies it does is where teams tend to waste effort.

## How to measure it

There is no brand-side metric specific to speakable markup, since its confirmed use case is narrow, so the useful check is the same one used for AI visibility generally: whether an AI engine's answer actually quotes or cites the brand's content, independent of whether speakable tags are present. Adding the markup and then watching citation behavior over following weekly checks is the realistic way to see whether it made any difference. If nothing changes after a reasonable number of checks, that result itself is useful information rather than a failure to interpret.

## Example

A news-adjacent blog adds speakable tags to its article summaries after reading that it helps with AI visibility, expecting a quick change in how often an AI engine quotes the site. Several weekly AI-visibility checks later, citation behavior looks unchanged, which matches the fact that the markup's confirmed use case does not extend to the engines the brand was hoping to influence. The team redirects the time they would have spent expanding the markup toward rewriting a page's opening paragraph instead.

## FAQ

### Does adding speakable markup help with AI-engine citations?

Not confirmed. Speakable is a beta feature for voice reading of news content on Assistant devices; it has no confirmed effect on whether ChatGPT, Claude, Gemini, Perplexity, Grok or DeepSeek cite a page.

### Is speakable worth adding anyway?

It is low-cost and machine-readable, so adding it where relevant does little harm, but it should not be prioritized over clearer page structure and content, which has a more established connection to being cited.

### Is speakable the same kind of thing as schema markup in general?

Yes. It is one specific type of schema.org structured data, narrower in scope than the general schema markup used to describe products, articles or FAQs on a page, and aimed specifically at voice reading rather than general indexing.

### Which pages are the best candidates for speakable markup?

News-style content with a clear, short summary near the top, since that matches the narrow use case it was actually designed for rather than a general product or pricing page.

## Related terms

- /glossary/ai-grounding
- /glossary/generative-engine-optimization

## Related

- Schema markup and AEO: /blog/schema-markup-and-aeo