Working with Prompts
Prompts are the questions you send to AI engines on behalf of your brand. Learn how to create, run, and interpret prompt results to understand your AI visibility.
What Is a Prompt?
A prompt is a question or query that you want to test against AI engines. Think of it as simulating what a potential customer might type into ChatGPT or Perplexity when researching your category.
For example, if you sell project management software, a prompt might be:
"What are the best project management tools for remote teams?"
AI Brand Report sends this question to each engine, captures the response, and analyses whether your brand was mentioned, how positively, and what sources were cited.
Creating Prompts
Manually
- Go to your project and click Prompts in the sidebar
- Click New Prompt
- Type your question
- Click Save
Auto-generated prompts
- From the Prompts page, click Generate Prompts
- The system analyses your domain and suggests relevant questions
- Review the suggestions and save the ones you want to keep
Running Prompts
Once prompts are saved, you can run them individually or in bulk:
- Run individually — click Run next to a specific prompt to fire it against all engines simultaneously
- Bulk run — use the bulk action to run all prompts at once (useful for a scheduled refresh)
Running a prompt takes 10–30 seconds per engine. Results appear as soon as each engine responds.
Understanding Prompt Results
Each prompt result page shows the full AI response from each engine, along with structured analysis.
Engine Cards
Each engine (ChatGPT, Perplexity, Gemini, Claude) has its own collapsible card showing:
| Field | Description |
|---|---|
| Presence | Whether your brand name, domain, or synonyms were detected |
| Stance | The AI's overall tone — Positive, Neutral, or Negative |
| Confidence | How confident the analysis is in that stance classification |
| Rationale | A brief explanation of why that stance was assigned |
| Score | A 0–100 visibility score for this engine/prompt combination |
| Journey Stage | Which stage of the buyer journey this prompt maps to |
| Topic Cluster | The topic category this prompt belongs to |
| Full Response | The raw AI response text, formatted with headings, lists, and tables |
| Citations | Sources cited by the engine (available for Perplexity) |
Previous Runs
Below the latest results, you'll see a history of all previous runs for this prompt — useful for tracking how AI knowledge about your brand changes over time.
Writing Good Prompts
The quality of your prompts directly affects the usefulness of your data.
Effective prompt patterns
Category comparison queries — these reflect real user research behaviour and often trigger competitive mentions:
- "What are the best [category] tools for [use case]?"
- "Compare [your brand] vs [competitor] for [use case]"
Buyer intent queries — questions asked when someone is close to making a decision:
- "Is [your brand] good for [enterprise / small business / specific use case]?"
- "What are the pricing options for [category]?"
Problem-first queries — how users describe pain points before they know which product to buy:
- "How do I [solve problem] without [common limitation]?"
What to avoid
| ❌ Avoid | Reason |
|---|---|
| "Tell me about [Your Brand]" | Too brand-specific; doesn't reflect real user search behaviour |
| Very generic queries | Won't reliably trigger category-relevant answers |
| Questions with obvious answers | Waste runs without generating useful visibility data |
| Queries unrelated to your category | Will skew your journey stage and topic distributions |
Prompt Organisation
Journey Stages
Each prompt result is automatically classified into a buyer journey stage:
| Stage | Description |
|---|---|
| Awareness | User is learning about a problem or category |
| Consideration | User is comparing options and evaluating features |
| Decision | User is ready to buy and comparing final choices |
| Retention | User is an existing customer looking for help or alternatives |
Topic Clusters
Results are grouped into topic clusters (e.g. Pricing, Integrations, Security, Customer Support) to help you spot where your coverage is strong or weak.
Monitoring Frequency
| Cadence | When to use |
|---|---|
| Weekly | For key "money" prompts in competitive categories |
| Monthly | Full prompt library re-run to catch gradual knowledge drift |
| After events | After a product launch, PR campaign, or major content push |
Re-running prompts is the only way to see whether your AI optimization efforts are paying off.