> ## 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.

# AI Score

> Understand friction AI's composite score for AI recommendations and brand discovery.

# AI Score

AI Score summarizes how strongly AI assistants recommend your brand across tracked prompts and model responses. It is not just a mention count. It favors outcomes that indicate preference: appearing in the answer, being described positively, and being recommended in purchase or selection contexts.

## What it combines

<CardGroup cols={3}>
  <Card title="Visibility" icon="eye">
    Whether your brand appears in the response and how consistently it appears across providers.
  </Card>

  <Card title="Sentiment" icon="smile">
    Whether the response frames your brand positively, neutrally, or negatively.
  </Card>

  <Card title="Purchase intent" icon="cart-shopping">
    Whether the response recommends your brand when a user is comparing, choosing, or buying.
  </Card>
</CardGroup>

## How to read it

<Tabs>
  <Tab title="High score">
    AI recognizes the brand, mentions it in relevant prompts, frames it positively, and recommends it in buyer contexts.
  </Tab>

  <Tab title="Low visibility">
    AI may like the brand when it appears, but the brand is absent from too many category or comparison answers.
  </Tab>

  <Tab title="Low sentiment">
    The brand appears, but the response is cautious, outdated, negative, or positioned behind competitors.
  </Tab>

  <Tab title="Low purchase intent">
    AI may mention the brand, but it does not guide the user toward choosing or buying it.
  </Tab>
</Tabs>

## Where AI Score appears

<CardGroup cols={2}>
  <Card title="Dashboard" icon="gauge" href="/dashboard/overview">
    The fastest top-level read on whether your AI recommendation position is improving or declining.
  </Card>

  <Card title="Analysis" icon="chart-line" href="/brand-audit/analysis">
    The place to inspect metric trends, provider differences, prompt evidence, sources, and exportable data.
  </Card>

  <Card title="Competitor" icon="code-compare" href="/brand-audit/competitor">
    A head-to-head view of your AI Score and metric trends against a promoted competitor.
  </Card>

  <Card title="Actions" icon="wrench" href="/actions/prompts">
    A practical list of prompts dragging down visibility, sentiment, or purchase intent.
  </Card>
</CardGroup>

## Common patterns

<AccordionGroup>
  <Accordion title="High visibility, low purchase intent">
    AI knows you, but it does not recommend you as the choice. Improve comparison pages, use-case positioning, reviews, and proof that supports buyer decisions.
  </Accordion>

  <Accordion title="Low visibility, high sentiment">
    When AI mentions you, the framing is strong. The issue is coverage. Expand content, citations, category pages, and prompts around the categories where you should appear.
  </Accordion>

  <Accordion title="Strong score on one model, weak on another">
    Different providers use different retrieval and answer patterns. Use provider filters in [Prompts](/tracking/prompts), [Analysis](/brand-audit/analysis), and [Entity](/brand-audit/entity) to find the source of the split.
  </Accordion>

  <Accordion title="Score changed after adding prompts">
    Adding prompts changes the measurement set. Compare the prompt-level evidence before interpreting the movement as a market-wide gain or loss.
  </Accordion>
</AccordionGroup>

## What to do next

<Steps>
  <Step title="Open the dashboard">
    Confirm whether the score movement is broad or isolated to a specific widget.
  </Step>

  <Step title="Inspect the metric">
    Use [Analysis](/brand-audit/analysis) to check visibility, sentiment, and purchase intent individually.
  </Step>

  <Step title="Open prompt evidence">
    Use [Actions: Prompts](/actions/prompts) to see the response text and recommendations for weak prompts.
  </Step>

  <Step title="Fix content or positioning">
    Use [Actions: Content](/actions/content) for website readiness gaps and [Prompts](/tracking/prompts) for query coverage gaps.
  </Step>
</Steps>

<Note>
  AI outputs can vary. Focus on trends, repeated evidence, and provider patterns rather than one isolated answer.
</Note>