How to Test Whether AI Systems Mention Your Brand

If you want to understand your current AI visibility, the most direct approach is prompt testing: systematically querying AI systems with relevant prompts and documenting how (or whether) your brand appears in the generated answers.

Prompt testing is not a one-time activity. AI systems update their models, refresh their retrieval data, and change their answer generation patterns regularly. Meaningful measurement requires structured, repeatable testing over time.

Why Prompt Testing Matters

Traditional SEO provides clear metrics: rankings, traffic, click-through rates. AI visibility does not have equivalent standardized metrics yet. You cannot check a dashboard to see how often AI systems cite your brand.

Prompt testing fills this gap. It gives you direct observational data about:

• Whether AI systems mention your brand at all in response to relevant queries
• How accurately AI systems describe your brand, products, or services
• Whether your brand is recommended, cited as a source, or merely mentioned in passing
• How your brand visibility compares to competitors for the same queries
• Whether visibility is improving, declining, or staying flat over time

Designing Your Prompt Set

The quality of your prompt testing depends on the quality of your prompts. A well-designed prompt set covers several categories:

Category-Level Prompts

These test whether your brand appears when someone asks about your industry or service category without naming specific brands. Examples:

• "What are the best [your service category] companies?"
• "Who provides [your service type] services?"
• "What companies specialize in [your industry niche]?"

Brand-Specific Prompts

These test how accurately AI systems represent your brand when it is specifically asked about. Examples:

• "What does [your brand] do?"
• "Tell me about [your brand] and their services."
• "Is [your brand] good for [specific use case]?"

Comparison Prompts

These test how your brand is positioned relative to competitors. Examples:

• "What is the difference between [your brand] and [competitor]?"
• "[Your brand] vs [competitor]: which is better for [use case]?"

Problem-Solution Prompts

These test whether your brand appears when someone describes a problem you solve. Examples:

• "How do I improve my [problem your service solves]?"
• "I need help with [challenge you address]. What are my options?"

Which AI Systems to Test

Test across multiple platforms because each AI system draws on different training data and retrieval methods:

• ChatGPT (OpenAI) -- The most widely used consumer AI assistant. Test with both free and paid tiers if possible, as they may use different models.
• Google AI Overviews -- Appears directly in Google search results. Test by searching relevant queries in Google and noting whether AI Overviews appear and what they contain.
• Perplexity -- An AI-powered search engine that provides sourced answers with citations. Particularly useful for evaluating citation visibility.
• Claude (Anthropic) -- Another major AI assistant with different training data and response patterns.
• Microsoft Copilot / Bing Chat -- Integrated into Microsoft's search and productivity tools.

Recording and Scoring Results

Consistent documentation is essential for tracking changes over time. For each prompt-platform combination, record:

1. Date and platform -- When the test was run and on which AI system.
2. Exact prompt used -- Record the exact wording to enable repeatable testing.
3. Mention status -- Was your brand mentioned? Options: Not mentioned, Mentioned, Recommended, or Cited as source.
4. Accuracy -- If mentioned, was the description accurate? Note any inaccuracies.
5. Position and context -- Where in the answer did your brand appear? Was it the primary recommendation or an afterthought?
6. Competitor mentions -- Which competitors were mentioned in the same answer?

Establishing a Testing Cadence

A single round of testing provides a snapshot. Meaningful measurement requires regular testing:

• Baseline assessment: Run your full prompt set across all platforms. This is your starting measurement.
• Monthly monitoring: Re-run the same prompt set monthly to track changes. Use identical prompts for consistency.
• Post-change testing: After making significant changes to your website (structured data updates, content restructuring, etc.), run targeted tests to see if there is an observable effect.
• Quarterly review: Every quarter, review your prompt set and update it to reflect any changes in your services, competitors, or market positioning.

Interpreting Results

Prompt testing produces qualitative data that requires careful interpretation. Some important caveats:

• AI responses are not deterministic. The same prompt can produce different answers on different days, even on the same platform. Look for patterns across multiple tests, not single data points.
• Correlation is not causation. If you update your structured data and your AI mentions improve next month, the improvement may or may not be directly caused by the update. Other factors (model updates, competitor changes, web crawl timing) also play a role.
• Absence is not failure. Not being mentioned in a specific response does not mean your optimization efforts are failing. AI visibility improvements are gradual and compound over time.