What is Speakable and How Can it Help Voice Search SEO?

What Is Speakable?

Speakable is a Schema.org property (speakable) that identifies sections of an article or webpage that are best suited for audio playback via text-to-speech (TTS). Google introduced Speakable support in 2018 as a beta feature to help news articles surface on Google Assistant when users ask for news on smart speakers.

The idea is straightforward: news articles contain a lot of text that doesn’t read well aloud (image captions, bylines, navigation, embedded quotes, legal boilerplate). Speakable markup lets publishers tell Google, “these sentences are the ones you should read aloud when a user asks about this story.” Google Assistant then plays back up to three articles on a topical query, using the speakable-marked passages for audio.

The Reality in 2026: Still Beta, Still Narrow

Speakable sounds great in theory. In practice, its adoption and utility have stayed limited:

• Google’s documentation still labels Speakable as BETA — unchanged since the 2018 launch.
• Eligibility is restricted to news content from publishers registered in Google News, serving users in the U.S., on English-language Google Home devices.
• Google periodically prunes structured data types that don’t drive results — seven schema types were officially retired in January 2026 (Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, Vehicle Listing, Practice Problems). Speakable was not in that list, but it’s been in “beta” for seven years with no feature expansion.
• For non-news sites, Speakable does essentially nothing in 2026 — Google Assistant doesn’t use Speakable markup to choose answers for non-news voice queries.

If you run a news publisher that’s registered in Google News and serves U.S. English-speaking audiences with smart speakers, Speakable is still worth implementing. For everyone else, it’s a low-priority tactic at best.

Why Speakable Still Matters (a Little)

One genuine 2026 use case has emerged: Speakable-marked content appears to be a signal for AI retrieval systems. Perplexity, ChatGPT Search, and Google’s own AI Overviews extract citable passages from web pages to generate answers. Speakable markup — which explicitly identifies the most important, self-contained passages — lines up naturally with how these AI systems want to extract content.

The direct voice search value has faded; the “clean summary passage for AI extraction” value has quietly become the main reason to use it on eligible content.

How to Implement Speakable Markup

If you’re a news publisher and eligible, Speakable is easy to add. Google accepts two implementation styles:

CSS selector (the common choice):

{
  "@context": "https://schema.org/",
  "@type": "Article",
  "headline": "Example Article Title",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [".headline", ".summary"]
  }
}

XPath (less common):

"speakable": {
  "@type": "SpeakableSpecification",
  "xpath": ["/html/head/title", "/html/head/meta[@name='description']/@content"]
}

Guidelines from Google’s current documentation:

• Mark sections that are self-contained — they make sense read aloud without visual context.
• Keep each speakable section reasonably short — Google recommends 20-30 seconds of TTS playback per section.
• Don’t mark whole articles as speakable — choose the most informative passages (usually the headline + the first 1-2 sentences of the lede).
• Test with Google’s Rich Results Test before publishing.

What Voice Search SEO Actually Looks Like in 2026

Voice queries account for 30%+ of all searches in 2026, but the techniques that drive voice visibility are mostly not Speakable. The real voice search playbook:

1. Own the Featured Snippet

Google Assistant pulls roughly 40-60% of voice answers directly from featured snippets — the answer box at the top of regular search results. Optimizing for featured snippets (direct question-and-answer structure, concise paragraph-length answers, properly-structured headers) is by far the highest-ROI voice search tactic. A page ranking at position zero on desktop is almost always what Google Assistant reads aloud for that query.

2. Use FAQ Schema Thoughtfully

FAQPage structured data signals question-and-answer relationships to Google and other search engines. Google has tightened rich-result eligibility for FAQ schema (mostly government and authoritative sites get the rich results now), but the structured content still helps with AI extraction and provides question-answer pairs that voice assistants can surface.

3. Optimize for Google Business Profile (Local Voice Searches)

Local voice queries — “pharmacy near me,” “coffee open now,” “plumber in Chicago” — are dominated by Google Business Profile (formerly Google My Business) data, not website content. A complete, accurate, review-rich GBP is the foundation of local voice search. 76% of local voice searches result in a visit within 24 hours, so this is also the highest-converting voice traffic.

4. Write Conversationally

Voice queries average 29 words compared to 4-6 for typed searches. They’re full questions (“What’s the best running shoe for flat feet?”) rather than fragments (“best running shoe flat feet”). Content that answers long-tail conversational queries — starting with Who/What/When/Where/Why/How — captures far more voice traffic than keyword-stuffed pages.

5. Focus on Page Speed and Mobile

Voice assistants skew to fast, mobile-friendly pages — Google has confirmed Core Web Vitals (LCP, INP, CLS) factor into what gets surfaced in voice results. A page that takes 4 seconds to load on mobile is often filtered out in favor of a faster competitor, regardless of content quality.

6. Optimize for AI Assistants, Not Just Voice

The blurring line between voice search and AI chatbots is the most important 2026 trend. When someone asks ChatGPT or Claude a question on their phone, they’re doing something that looks a lot like voice search but runs on a completely different stack. The content attributes that work for both: strong E-E-A-T signals, clear structured data, direct answers early on the page, and authoritative citations. Allowing AI crawlers in robots.txt (GPTBot, ClaudeBot, PerplexityBot) determines whether your content can be cited at all — see our robots.txt guide for the specifics.

Risks and Limitations of Relying on Speakable

A few caveats on over-investing in Speakable specifically:

• Eligibility gates. If your site isn’t registered in Google News as a news publisher, Speakable markup does nothing for you. Adding it anyway causes no harm but also no benefit.
• Locale restrictions. U.S. English only on Google Home. Non-English, non-US audiences get no Speakable-based voice surfacing.
• Beta status. Seven years of beta with no expansion should set realistic expectations for future investment.
• No guarantee of selection. Even eligible, well-marked-up content isn’t guaranteed to be chosen for Google Assistant news answers — Google picks up to three articles per topical query from the entire eligible pool.

The risk isn’t that Speakable hurts anything. It’s that time spent on Speakable is time not spent on featured snippets, FAQ content, GBP optimization, Core Web Vitals, or AI crawler management — which move voice-search-adjacent traffic much more.

Frequently Asked Questions

Does Speakable still work in 2026?
Technically yes — Google still supports the speakable Schema.org property, and Google Assistant still uses it for topical news answers on smart speakers. But it’s still in beta (as it has been since 2018), still limited to U.S. English news content, and it’s not how most voice search answers are selected. For non-news sites, Speakable has essentially no effect on voice search visibility.

Is Speakable good for SEO generally, or just voice search?
Historically, just voice search — and a narrow slice of it. In 2026, one emerging benefit is that Speakable markup correlates with being cited by AI retrieval systems (Perplexity, ChatGPT Search, AI Overviews) because the markup explicitly identifies the most extractable passages. But there’s no confirmed ranking boost for regular search.

How does voice search differ from AI chatbot queries?
The user experience is converging. Both are conversational, question-based, and return a single (or very few) answers. The back-end is different — voice assistants like Google Assistant pull primarily from Google search and local data; AI chatbots use LLMs with either training data, live web retrieval, or both. For content strategy, the tactics overlap heavily: direct answers, structured content, strong E-E-A-T, fast load times, and allowing AI crawlers to access your content.

What’s the single highest-ROI voice search tactic?
Winning featured snippets for questions your audience actually asks. Google Assistant pulls 40-60% of voice answers from featured snippets. Ranking at position zero for a voice-friendly query typically means that’s the answer Google reads aloud. For local businesses, a complete Google Business Profile is an equally high-ROI move. See our on-page SEO tips for the foundational work.

Bottom Line

Speakable is a real Schema.org property that real news publishers can use to improve Google Assistant news playback for U.S. English audiences with smart speakers. If that describes your site, implement it — it’s easy, it’s documented, and it might still help on the margin (plus the emerging AI-retrieval citation benefit). If it doesn’t describe your site, skip Speakable and focus on what actually drives voice search and AI assistant visibility: featured snippets, FAQ structure, Google Business Profile, conversational long-tail content, Core Web Vitals, and proper AI crawler access.

Voice search in 2026 isn’t really about Speakable — and for most sites, it never was. It’s about being the concise, authoritative answer to the question someone’s asking, on whatever device or assistant they’re asking it through. For the broader arc of how voice and AI search fit into SEO’s evolution, see our history of SEO and search engines.