What Is Speakable Schema?
Speakable schema is a structured data specification developed by Schema.org that allows web publishers to identify specific sections of their content as suitable for audio narration. When implemented correctly, it tells Google Assistant and other voice-enabled platforms which passages are clear, concise, and appropriate to read aloud in response to a voice query.
The markup uses either CSS selectors or XPath expressions to point to parts of a page, such as the introduction paragraph or a key summary section.
Google currently applies speakable schema primarily to news content when delivering audio briefings through Google Assistant on smart speakers and mobile devices. The standard is defined under the SpeakableSpecification type in the Schema.org vocabulary and is implemented within a NewsArticle or Article JSON-LD block.
For South African publishers and brands producing regular news, commentary, or informational content, speakable schema is an important part of a comprehensive SEO and answer engine optimisation strategy.
Voice search adoption is growing across South Africa, driven by smartphone penetration and the popularity of Google Assistant on Android devices. Marking up content for speakable delivery positions your brand to be heard, not just read, in the emerging voice search landscape.
The schema does not guarantee your content will be read aloud. Google still evaluates the quality and relevance of the marked section before including it in a voice response. Short, direct, factual passages perform best.
Avoid marking up paragraphs that contain navigation cues, promotional language, or content that requires visual context to understand. A well-written two- to four-sentence summary at the start of an article is typically the strongest candidate for speakable markup.
Speakable Schema In Practice
The two scenarios below are illustrative examples, not Juicy Designs client results. They indicate the kind of effect that speakable schema work typically produces, so treat them as indicative rather than measured.
Picture a South African news portal covering local business and finance news that implements speakable schema on its daily market updates. Each article's opening summary paragraph, describing the key development and its immediate impact, would be tagged using a CSS selector pointing to the .article-summary class.
When a user asks Google Assistant "What is happening with the JSE today?", the Assistant could draw on the tagged passage and read a concise answer aloud.
For a Pretoria-based digital agency like Juicy Designs, speakable schema could be applied to FAQ-style content or service overview pages where a clear introductory sentence answers a common question directly.
The process involves adding a speakable property inside the existing Article or NewsArticle JSON-LD block, specifying either a CSS selector or an XPath value pointing to the relevant HTML element.
Because Google currently limits speakable support to news content, non-news publishers should still implement it as a forward-looking optimisation while ensuring all other structured data is accurate and complete.
What speakable schema is
Speakable schema is a type of structured data (schema.org's speakable property, typically in JSON-LD) that identifies specific sections of a page's content as particularly suitable for being read aloud by text-to-speech, such as by voice assistants, marking the parts of the content best suited to be spoken in response to a voice query. The idea is to indicate to search engines and voice platforms which portions of a page (for example, a concise summary or the key answer) are appropriate to be read out by a voice assistant, so that voice-based results can use those marked sections. Speakable schema was introduced in connection with voice search and voice assistants, aiming to help identify content suitable for audio delivery. It is important to understand the status and applicability of speakable schema: it has been available primarily in a limited context (Google's documentation associated it particularly with news content and specific contexts, and it has been in a limited or beta status rather than a broadly-applicable feature), so its practical use and support have been narrower than some other schema types, and it is not a widely-applicable, guaranteed feature for general content, but rather a specialised markup with limited support and application. Understanding speakable schema matters mainly to know what it is, structured data marking content as suitable for text-to-speech, and to have a realistic view of its limited applicability and support, so that a business does not over-invest in it expecting broad voice-search benefits, but understands it as a specialised, limited-support markup, while focusing its voice-search efforts on the broader, more impactful approach of clear, question-answering content that serves voice search (and featured snippets and AI answers) regardless of speakable markup.
Using speakable schema and voice-search reality
Using speakable schema realistically means recognising its limited applicability and support, and focusing voice-search efforts on the broader, more impactful content approach, rather than relying on speakable markup for voice visibility. On speakable schema's limited status: speakable has been available in a limited context and status (associated particularly with news and specific use cases, and not a broadly-supported, guaranteed feature for general content), so its practical impact for most businesses' general content is limited, and it is not a reliable, widely-applicable route to voice-search visibility, meaning a business should not expect broad voice benefits from adding speakable markup to general content, since its support and application are narrow. On what actually drives voice-search visibility: voice search visibility comes largely from the broader approach of providing clear, direct, concise answers to the questions people ask, on pages that rank, since voice assistants typically read out a single, concise answer often drawn from featured snippets and top results, so the clear, question-answering, featured-snippet-optimised content that serves voice search is what matters most, and this is the same answer-focused content that serves featured snippets and AI answers, applied with voice's conversational, question-based queries in mind, rather than depending on speakable markup. So for voice search, the impactful work is clear, direct answers to conversational questions on ranking pages (voice-search optimisation), not speakable schema, which is a limited-support specialised markup. On the specific questions: whether speakable schema helps with Google Assistant results depends on its limited support and applicability, and it is not a broadly-reliable route, so its help is limited and context-dependent, and voice visibility is better pursued through clear, question-answering content; and the content types that would most benefit from speakable, in principle, are those suited to being read aloud as concise answers or summaries (such as news summaries or clear, concise answer content), but given speakable's limited support, the broader point is that concise, clear, answer-focused content benefits voice search generally, whether or not speakable markup is used. For a South African business, the practical guidance on speakable schema is to be aware of it as a specialised, limited-support markup for identifying text-to-speech-suitable content, but not to over-invest in it for general content expecting broad voice-search benefits, and instead to focus voice-search efforts on the broader, more impactful approach: providing clear, direct, concise answers to the conversational questions its audience asks, on ranking pages (which serves voice search, featured snippets and AI answers alike). Because speakable schema has limited applicability and support while clear, question-answering content is what genuinely drives voice-search visibility, the sound approach is to prioritise the latter, treating speakable schema as a minor, specialised markup rather than a key voice-search lever, which is the realistic way to approach speakable schema and voice search.
FAQ
Does speakable schema help with Google Assistant results in South Africa?
Speakable schema is currently used by Google Assistant for news content. South African publishers and brands producing news or briefing-style content can benefit by marking up summary sections, increasing the chance of their content being read aloud in voice responses.
Which content types benefit most from speakable schema?
News articles, product summaries, how-to introductions, and FAQ answers benefit most. Concise, factual paragraphs that answer a specific question directly are the best candidates for speakable markup on any page.