> ## Documentation Index
> Fetch the complete documentation index at: https://docs.frictionai.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompts

> Create and analyze custom prompts and search queries for AI visibility tracking.

# Prompts

Prompts is where you track the specific questions and searches that should surface your brand. Use it for category discovery, competitor comparison, branded questions, and search queries.

<Frame>
  <img src="https://mintcdn.com/friction/GwaVVezuSSMLCKiT/images/prompts-tracking.svg?fit=max&auto=format&n=GwaVVezuSSMLCKiT&q=85&s=159580a8e381a9e5ac3588ef62b55efa" alt="Prompts table with filters, visibility scores, sources, and tags" width="960" height="520" data-path="images/prompts-tracking.svg" />
</Frame>

## What this page answers

<CardGroup cols={2}>
  <Card title="Does AI mention us?" icon="eye">
    Visibility score shows whether your brand appears in responses for the tracked prompt.
  </Card>

  <Card title="Who else appears?" icon="users">
    Brand mention stacks show competitors and other brands that appear in the same answer.
  </Card>

  <Card title="Which sources matter?" icon="link">
    Source panels show the domains AI cites when forming an answer.
  </Card>

  <Card title="Which models differ?" icon="brain">
    Provider filters let you compare OpenAI, Claude, Gemini, Perplexity, and Google AI Overviews where your plan allows.
  </Card>
</CardGroup>

## Prompt types

<Tabs>
  <Tab title="LLM prompts">
    Natural-language questions sent to AI assistants.

    ```text theme={null}
    What are the best AI visibility platforms for AEO teams?
    How does friction AI compare to Profound?
    ```
  </Tab>

  <Tab title="Search queries">
    Search-style queries used to evaluate AI search and retrieval surfaces.

    ```text theme={null}
    best AI visibility platform
    friction AI alternatives
    ```
  </Tab>
</Tabs>

## Add prompts

<Steps>
  <Step title="Open Prompts">
    Go to Tracking -> Prompts in the sidebar, then select **Add Prompts**.
  </Step>

  <Step title="Choose manual or suggested prompts">
    Paste your own prompts or use friction AI suggestions based on the selected brand and category.
  </Step>

  <Step title="Add tags">
    Use tags like `comparison`, `brand`, `category`, `product`, or `campaign` so filtering remains manageable.
  </Step>

  <Step title="Save and wait for analysis">
    Analysis starts automatically. Results populate the table as they become available.
  </Step>
</Steps>

## Filters

<AccordionGroup>
  <Accordion title="Search">
    Filter the table by prompt text or tags.
  </Accordion>

  <Accordion title="Brand mention">
    Switch between all prompts, prompts where your brand appears, and prompts where your brand is missing.
  </Accordion>

  <Accordion title="Time period">
    Review 7-day, 30-day, 90-day, 180-day, or 1-year windows depending on available data.
  </Accordion>

  <Accordion title="Model">
    Filter to specific providers when the plan includes those providers for prompt tracking.
  </Accordion>

  <Accordion title="Tags">
    Use tags to group prompts by campaign, funnel stage, product line, or market segment.
  </Accordion>
</AccordionGroup>

## How to interpret a weak prompt

<Steps>
  <Step title="Open the detail panel">
    Click a prompt row to inspect provider responses, aggregate scores, source citations, and brand mentions.
  </Step>

  <Step title="Look for pattern, not one answer">
    Check whether the problem repeats across providers or appears only in one model.
  </Step>

  <Step title="Check cited sources">
    If AI cites competitor or third-party pages, decide whether your owned pages answer the same intent clearly enough.
  </Step>

  <Step title="Send the issue to Actions">
    Use [Actions: Prompts](/actions/prompts) to review underperforming prompts with recommendations.
  </Step>
</Steps>

## Prompt writing guidance

Good prompts match real discovery language.

<CardGroup cols={2}>
  <Card title="Use buyer language" icon="message">
    Write prompts the way a prospect would ask before choosing a tool, brand, product, or service.
  </Card>

  <Card title="Cover comparison moments" icon="code-compare">
    Include direct comparisons and alternative searches where competitors are likely to appear.
  </Card>

  <Card title="Keep prompts stable" icon="history">
    Avoid changing prompt text constantly. Stable prompts create cleaner trend history.
  </Card>

  <Card title="Tag by intent" icon="tag">
    Intent tags make it easier to separate awareness, comparison, shopping, and retention prompts.
  </Card>
</CardGroup>

<Tip>
  Start with 10 to 20 high-quality prompts before adding more. A small, representative set is easier to interpret than a large noisy one.
</Tip>