Voice search
Voice Search Optimization for Cannabis
Your customers are asking Alexa, Google Assistant and their phones. They are not typing. A voice result returns one to three answers, not ten, so ranking for voice is foundational rather than incremental.
- Alexa
- 100M+
- Google Assistant
- 500M+
- Siri
- 900M+
- Voice share of search
- 20 to 30%
devices reached
devices reached
Apple devices reached
of all searches now
What changes
A spoken query is not a typed query with more words in it
| Criteria | Typed search | Voice search |
|---|---|---|
| Query form | Keyword strings, for example anxiety cannabis strain | Natural language questions, for example what is a good strain for anxiety |
| What comes back | Ten results to scan | One to three answers, spoken, usually pulled from the featured snippet |
| Keyword research method | Standard keyword tools | Support transcripts, review language, forum threads and People Also Ask data |
| Monthly volume per term | Concentrated on commercial head terms | Typically 20 to 50 searches per long tail question |
| Conversion | The commercial keyword baseline | Long tail conversational terms convert 2 to 3x higher |
| Devices in play | A screen the customer is already looking at | Smart speakers, smart displays, phones, in car assistants and wearables |
The constraint
Voice platforms will not sell your product. They will still send you the customer.
Alexa restricts cannabis product transactions, Google Assistant suspends cannabis shopping integration, and Siri blocks cannabis transactions. That is not an oversight. Cannabis transactions require age verification, licensing checks and compliant payment processing, and voice platforms cannot enforce those at scale. So the strategy is decided by the constraint rather than around it: win the voice query, drive the customer to your site, and handle the transaction there, compliantly.
Local discovery
Real time accuracy is the local voice ranking factor
Voice local queries assume urgency. Customers ask what is open now, not which dispensaries have hours listed, which makes real time hour accuracy non negotiable. Google Assistant and Alexa both weight review count, rating, hours consistency and distance. Outdated hours, an incorrect address or an inaccurate delivery status can trigger de listing from voice local results entirely.
The platforms
Three assistants, three different jobs
Related hubs
Every surface a spoken answer touches
Listen to the result
A spoken answer only ever names one business
Ask the question your customer asks and hear who wins it
- 01 IfYou ask an assistant about your own brand and it reads a competitor's pageAnswer engine optimizationA voice result returns one to three answers. Owning the answer is the whole contest.
- 02 IfYou rank on page one and are still never read aloudFeatured snippets and position zeroGoogle reads the snippet text aloud. If you own the snippet you own the voice answer.
- 03 IfAssistants cannot tell what your pages are actually aboutSchema and structured dataWithout schema markup, voice assistants cannot understand your content context.
- 04 IfNobody nearby gets your address, hours or phone number from a speakerLocal SEO and GBPVoice weights review count, rating, hours consistency and distance.
- 05 IfYour pages answer keywords rather than questionsContent strategy for AI searchCustomers ask what is a good strain for anxiety, not anxiety cannabis strain.
- 06 IfYou are invisible inside AI assistants generally, not only in voiceAI search optimizationVoice is one surface of the same retrieval problem.
What gets asked
The shape of cannabis voice demand
- 40%
- Informational voice queries
- 35%
- Local discovery voice queries
- 15%
- Transactional voice queries
- 10%
- Comparison voice queries
- 30+
- Pages carrying speakable markup
- 50
- Voice queries mapped to page and schema
Share of cannabis voice search
Share of cannabis voice search
Share of cannabis voice search
Share of cannabis voice search
Implementation target
Implementation target
The answer format spec
What a voice answer has to be, mechanically
- Speakable block
- 50 to 80 words
- Paragraph snippet target
- 40 to 60 words
- Definition snippet
- 60 words
- Step snippet
- 5 to 8 steps
- List snippet
- 3 to 5 items
- Table snippet
- 2 to 3 columns
- Product detail placement
- First 100 words
- Mobile load
- Under 2 seconds
Longer blocks lose listener attention. Shorter blocks lack sufficient information.
Google reads the featured snippet aloud, so the snippet is the voice answer.
What is queries return definitions.
How to queries return steps.
Cannabinoid ratios, terpene profiles and strain comparisons.
Strain name, cannabinoid percentages, terpene profile, effects and usage instructions.
Voice to web transitions have to load instantly.
The full discipline
Every piece of this discipline, in 8 deep dives
- 01Alexa Optimization for Cannabis BusinessesRead the breakdown
- 02Voice Commerce for Cannabis BusinessesRead the breakdown
- 03Conversational Keyword Strategy for Cannabis BusinessesRead the breakdown
- 04Google Assistant Optimization for Cannabis BusinessesRead the breakdown
- 05Local Voice Search for Cannabis BusinessesRead the breakdown
- 06Voice Search Schema Markup for Cannabis BusinessesRead the breakdown
- 07Apple Siri Optimization for Cannabis BusinessesRead the breakdown
- 08Smart Speaker Strategy for Cannabis BusinessesRead the breakdown
First deliverable
Hear what the assistants say about you
We ask Google Assistant and Alexa your target queries, record which competitors answer and in what content format, then return the gaps as a prioritised queue rather than a report.