Prompts

The prompt library used to query AI models — add, edit, tag, and organize by product or campaign

What prompts are

A prompt is the question or instruction sent to an AI model during each tracking run. The quality and coverage of your prompt set directly determines how useful the Intelligence data is.

Good prompts mimic what a real user in your target market would type into ChatGPT or Perplexity. Poor prompts (too generic, too narrow, or too biased toward your brand) produce data that doesn’t reflect real-world AI behavior.

Adding a prompt

1. Open Intelligence → AI Overviews → Prompts
2. Select Add prompt
3. Write the prompt text — aim for a natural question, not a keyword
4. Assign a product (required) and one or more tags (optional)
5. Select which AI models should run this prompt
6. Save

The prompt enters the queue immediately and will be included in the next scheduled run.

Prompt fields

Field	Required	Description
Prompt text	Yes	The question sent to AI models
Product	Yes	The internal product line this prompt belongs to
Tags	No	Labels for grouping and filtering (e.g. top-of-funnel, brand, competitor)
AI models	Yes	Which models run this prompt (ChatGPT, Gemini, Perplexity, Google AI Overviews)
Country	No	Locale used when querying models that support geo-targeting
Active	Yes	Inactive prompts are not run but their history is preserved

Tags

Tags are free-form labels you apply to prompts. They serve as the main filter across the entire Intelligence workspace — any chart or table can be filtered to show only prompts with a specific tag.

Common tagging strategies:

• By topic: pricing, comparison, how-to, review
• By funnel stage: awareness, consideration, decision
• By campaign: q1-launch, competitor-attack, rebrand

A prompt can have multiple tags. Tags are shared across your team.

Editing and deactivating prompts

Select any prompt to edit its text, tags, product assignment, or model list. Changes take effect on the next scheduled run — historical data is not retroactively recalculated.

To stop tracking a prompt without deleting its data, toggle it to inactive. Inactive prompts remain visible in the library with their full history intact.

Prompt run schedule

Runs are scheduled automatically based on your plan’s run frequency. Each active prompt is sent to each of its assigned AI models on every run. You can see run timestamps in the AI Overviews trend chart tooltip.