Prompt
How do I measure AI visibility by topic and geography?
Latest observation
To measure AI visibility by topic and geography, treat it like an SEO/brand-monitoring problem with two layers:
- Topic visibility: how often your brand appears for specific subjects or prompts.
- Geographic visibility: how that appearance changes by country, city, language, or market.
Here’s a practical framework.
1) Define what “AI visibility” means for you
Pick the AI surfaces you care about, because measurement differs by platform:
- Chatbots / LLMs: ChatGPT, Claude, Gemini, Perplexity
- AI search summaries: Google AI Overviews, Bing Copilot
- Vertical AI tools: travel, shopping, local discovery, healthcare, etc.
Then define the outcome you want to track:
- Mentioned or not
- Recommended or not
- Ranked in top N
- Cited/source-linked
- Sentiment of the mention
- Share of voice vs competitors
2) Build a topic taxonomy
Create a list of topic clusters that match your business and audience.
Example:
- Product category
- Use case
- Pain point
- Competitor comparisons
- Industry verticals
- Local intent topics
For each topic, create a set of prompts/questions users might ask.
Example topic: CRM for small businesses
- “Best CRM for small businesses”
- “What CRM is easiest for a 10-person team?”
- “Affordable CRM in [country]”
- “CRM alternatives to Salesforce”
This becomes your measurement corpus.
3) Build a geography taxonomy
Decide how geography is defined:
- Country
- Region
- State/province
- City/metro
- Language locale
- Market-specific variants
Important: AI answers can vary based on:
- User location
- Browser language
- Search locale
- Local sources available to the model
- Regional regulations and availability
So you should measure by both:
- Locale settings in the query environment
- Geo-specific prompt wording
e.g., “best restaurant POS in Toronto” vs “best restaurant POS in the UK”
4) Create a repeatable prompt set
For each topic × geography combination, generate a standard prompt set.
Example matrix:
| Topic | Geography | Prompt |
|---|---|---|
| POS for restaurants | US | Best POS systems for restaurants in the US |
| POS for restaurants | UK | Best POS systems for restaurants in the UK |
| POS for restaurants | Toronto | Best POS systems for restaurants in Toronto |
Keep prompts consistent so you can compare over time.
5) Measure key visibility metrics
Use these metrics for each topic and geography:
A. Mention rate
How often your brand appears in responses.
- Formula:
Mention rate = prompts where brand appears / total prompts tested
B. Recommendation rate
How often your brand is included among recommended options.
- Formula:
Recommendation rate = prompts where brand is recommended / total prompts tested
C. Rank position
Where your brand appears in the response list.
- Track average position:
- #1
- Top 3
- Top 10
- Not present
D. Share of voice
Your visibility vs competitors for the same topic/geo set.
- Formula:
Brand mentions / total mentions across brands
E. Citation rate
How often the AI cites your site or trusted third-party sources mentioning you.
F. Sentiment / framing
Are you described positively, neutrally, or negatively?
G. Coverage breadth
How many topic × geo combinations you appear in.
- Example: visible in 12 of 30 market-topic combinations.
6) Run tests across models and geos
Measure on multiple dimensions:
- Model/platform: ChatGPT, Perplexity, Gemini, etc.
- Locale: language, country, city
- Device/context: desktop/mobile if relevant
- Time: weekly or monthly
Use a controlled process:
- Same prompt set
- Same date/time window
- Same model version if possible
- Same locale settings
7) Capture results in a dashboard
Store outputs in a structured table:
| Date | Platform | Topic | Geography | Prompt | Brand mentioned | Rank | Cited | Sentiment | Competitors |
|---|
Then build dashboards by:
- Topic
- Geography
- Platform
- Trend over time
- Competitor comparison
8) Compare against search and local signals
AI visibility is often influenced by:
- SEO rankings
- Local listings
- Reviews
- Wikipedia / knowledge panels
- News coverage
- Structured data
- Third-party directories
- Regional PR
So correlate AI visibility with:
- Organic search visibility
- Local pack visibility
- Review volume/ratings
- Regional backlink mentions
- Country-specific content performance
9) Identify content gaps by topic and region
If you’re visible in one geography but not another, ask:
- Do you have local landing pages?
- Is the content localized?
- Are local sources mentioning you?
- Is your brand available in that market?
- Do you have region-specific pricing, shipping, support, or compliance info?
If you’re visible for one topic but not another, ask:
- Do you have authoritative content for that subtopic?
- Are competitors stronger on that use case?
- Does the AI consider you relevant for the prompt wording?
10) Use a simple scorecard
A practical scoring model:
AI Visibility Score =
- 40% mention/recommendation rate
- 25% rank position
- 15% citation rate
- 10% sentiment
- 10% coverage breadth
Then segment the score by:
- Topic
- Geography
- Platform
11) Best practices
- Test in native language where relevant.
- Separate brand visibility from category visibility.
- Don’t rely on one model; visibility differs widely.
- Track prompt variants because user wording changes results.
- Re-test regularly; answers shift frequently.
- Include competitors to benchmark share of voice.
12) Tools and methods
You can do this manually or automate it.
Manual
- A spreadsheet
- Repeated prompt tests in each platform
- Human coding of responses
Automated
- Scripts to query APIs where allowed
- Prompt orchestration tools
- Response parsing and classification
- BI dashboard for trends
If you want a more enterprise-style setup, you can combine:
- A prompt library
- Locale-specific browser sessions
- Response extraction
- NLP classification for mention/rank/sentiment
Example workflow
- Choose 10 topics and 5 geographies.
- Write 3 prompts per topic-geo pair.
- Run prompts in ChatGPT, Perplexity, Gemini, and Google AI Overviews.
- Record whether your brand appears, its rank, whether it’s cited, and the sentiment.
- Calculate mention rate and share of voice by topic and geography.
- Compare results month over month.
- Fix content/SEO/local signals where visibility is weak.
A simple starting template
Track these columns:
- Date
- Platform
- Model/version
- Topic
- Geography
- Prompt
- Brand present? (Y/N)
- Position
- Citation? (Y/N)
- Competitors mentioned
- Sentiment
- Notes
If you want, I can also give you:
- a spreadsheet template for AI visibility tracking, or
- a step-by-step measurement framework for a specific industry or market.