What it is
Speakable is a property in schema.org JSON-LD that marks specific paragraphs or CSS selectors as the canonical text for voice readback. When Google Assistant answers a voice query and finds a Speakable-tagged page, it reads the marked content rather than the page title or first paragraph.
{
"@type": "Article",
"speakable": {
"@type": "SpeakableSpecification",
"cssSelector": ["h1", "[data-speakable]"]
}
}Why it's underused in B2B
Voice-search SEO in B2B / technical content is functionally empty. Most marketing teams treat voice as a consumer-only channel ("ok Google, set a timer"). The result: pages that DO get queried via Assistant for technical/founder questions return generic AI-summarized snippets instead of opinionated, specific content.
A founder asking their phone "how much does an MVP cost in 2026" today gets a blended AI summary. With Speakable on a real opinion-piece, you become the cited source.
When to use it
On pages that have one clear answer-shaped section. FAQs (the answer paragraph). Glossary terms (the definition). Pricing pages (the headline number). Don't use it on long essays — Assistant won't read a 1500-word article aloud.
When NOT to use it
When the page is mostly navigation, multiple unrelated topics, or marketing copy. Voice search wants a specific answer to a specific question.
The actual implementation
Add the schema to the page's JSON-LD. Optionally tag the relevant DOM nodes with data-speakable (or whatever CSS selector you put in the schema). Verify in Google Rich Results Test.
