The Promise That Didn’t Quite Materialise
Voice search was supposed to transform B2B marketing. By 2020, according to the predictions that circulated around 2016-2018, 50% of all searches would be voice-based. Speakable schema markup — structured data that tells search engines which parts of your content are suitable for text-to-speech playback — was positioned as the technical foundation for capturing this wave.
The 50% prediction never materialised. Not even close. But the conversation has evolved in ways that make a straightforward “voice search is dead” conclusion wrong for B2B marketers in 2026. The picture is more nuanced — and more commercially relevant — than the binary narrative suggests.
Where Voice Search Actually Stands in 2026
Voice search adoption has grown, but not in the way the early predictions imagined. The growth has been in consumer contexts — smart speakers, in-car assistants, mobile voice queries — and skews heavily towards simple, informational, and local queries. “What time does Tesco close” works well by voice. “Compare enterprise ERP systems for mid-market manufacturers” does not.
For B2B, voice search usage remains marginal as a direct research tool. The overwhelming majority of B2B research — particularly in regulated sectors — happens via typed queries in browser-based search or AI chat interfaces. Professionals researching financial advisers, law firms, or engineering consultancies are not doing so via Alexa.
The honest assessment: voice search is not a meaningful B2B discovery channel in 2026 for most regulated and professional services businesses. If your GEO strategy is built around voice search optimisation, you are optimising for a channel that delivers negligible B2B commercial impact.
What About Speakable Schema?
Speakable schema (the speakable property in schema.org markup) tells search engines which sections of a page are appropriate for audio playback or text-to-speech reading. Google supported it in a limited beta for news publishers but has not expanded it broadly.
The current state:
- Google’s documentation still lists speakable as available for Google Assistant on Google Home devices, primarily for news content
- Adoption outside news publishers remains extremely low
- There is no evidence that speakable schema influences rankings, AI citations, or visibility in ChatGPT, Perplexity, or Google AI Overviews
- The schema type is not deprecated, but it is not being actively promoted or expanded
For B2B businesses, implementing speakable schema is not harmful — but it should not be a priority. The time and technical resources required to implement it deliver virtually no measurable return for B2B audiences.
What Has Actually Replaced the Voice Search Promise
The interesting development is that the underlying user behaviour predicted by voice search advocates — conversational, natural language queries — has manifested, just not through voice. It has manifested through AI chat interfaces.
When a buyer types “which accounting firms in Manchester specialise in R&D tax credits for SaaS companies” into ChatGPT, they are using exactly the kind of natural language, conversational query that voice search was supposed to enable. The interface is text, not voice. But the query structure — and the optimisation requirements — are remarkably similar to what voice search evangelists described.
This is where the connection to GEO becomes direct:
- Natural language query patterns — the long, specific, conversational queries that voice search predicted are now standard in AI chat. Optimising for these patterns is core GEO work.
- Answer extraction — voice assistants were supposed to read out a single, definitive answer. AI chat interfaces do exactly this, in text form. Content structured for AI extraction serves both use cases.
- Entity recognition — voice search required strong entity signals so that assistants could identify and recommend specific businesses. AI chat interfaces require the same signals for the same reason.
What B2B Marketers Should Actually Do in 2026
Deprioritise voice-specific optimisation. Unless your business has a significant consumer-facing component or operates in a sector where mobile voice queries are genuinely common, dedicated voice search optimisation is not a productive use of resources.
Do not invest in speakable schema as a priority. Implement it if you have completed all higher-value schema types (Organisation, FAQPage, HowTo, Article, LocalBusiness) and have technical capacity to spare. Do not implement it instead of those higher-value types.
Invest in the query patterns voice search predicted. Natural language, conversational, question-format content is exactly what AI models extract and cite. The voice search promise was right about user behaviour — it was wrong about the interface.
Focus on structured data that AI actually uses. Organisation schema, FAQ schema, HowTo schema, and Article schema with proper author attribution are all demonstrably used by AI models when constructing answers. These should be fully implemented before considering speakable.
Build for AI chat, not voice assistants. The B2B buyer’s conversational search interface in 2026 is ChatGPT, Perplexity, and Google AI Overviews — not Alexa or Google Home. Optimise accordingly.
The Bottom Line
Speakable schema and voice search optimisation for B2B are not dead, but they are not commercially significant for most UK professional and regulated sector businesses in 2026. The conversational search revolution that voice search predicted has arrived — through AI chat interfaces, not voice assistants. The winning strategy is to optimise for the behaviour, not the original interface.
Find out how AI answers questions about your business today. Request your free AI Visibility Audit →