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

# Prompt Performance

> Analyze individual monitored prompts and period-level mention state in Looker Studio

Detailed prompt-level metrics for analyzing individual query performance.

<Tabs>
  <Tab title="Dimensions">
    | Field | Type | Description |
    | - | - | - |
    | `date` | Date | Response date (YYYY-MM-DD) |
    | `provider` | Text | AI provider (ChatGPT, Claude, Perplexity, etc.) |
    | `funnel` | Text | Query intent stage (TOFU, MOFU, BOFU) |
    | `topic` | Text | Associated keyword/topic |
    | `tag` | Text | Tag names joined with " \| " (or "Untagged") |
    | `prompt` | Text | Prompt text (truncated to 200 chars) |
    | `brandMentioned` | Boolean | True if your brand was mentioned |
    | `brandMentionedLabel` | Text | `Oui` or `Non`, useful for dropdown controls |
    | `promptBrandMentioned` | Boolean | True if the brand was mentioned for this prompt anywhere in the selected period |
    | `promptBrandMentionedLabel` | Text | Period-level `Oui` or `Non`, useful for isolating prompts never mentioned in the selected period |
  </Tab>

  <Tab title="Metrics">
    | Field | Type | Description |
    | - | - | - |
    | `score` | Percent | % of brand-mentioning responses among responses mentioning any tracked brand (null when none) |
    | `brandMentions` | Number | Total brand mentions count |
    | `promptBrandMentions` | Number | Period-level brand mention total for the prompt; aggregated with MAX because it repeats on each date/provider row |
    | `sourceCitations` | Number | Citations of your own sources |
    | `shareOfVoice` | Percent | Brand / (Brand + Competitors) mentions (null when no mentions) |
    | `sentiment` | Percent | Average sentiment score (0-100), null when the brand is not mentioned |
    | `brandPosition` | Number | Average brand mention position (null when not mentioned) |
  </Tab>
</Tabs>

<Note>
  Each row covers one prompt per date and provider. Prompts carrying several
  tags appear once, with all tag names joined in the `tag` field, so SUM
  metrics are never double-counted. To filter on a specific tag, use a
  `CONTAINS_TEXT` condition instead of an exact match.
</Note>

<Note>
  Aggregated `shareOfVoice` here is a prompt-level average (each prompt weighs
  the same regardless of volume), so it will not match the brand-level Share of
  Voice. For the volume-weighted brand figure shown in the Qwairy interface,
  use the Performance Overview connector's Share of Voice.
</Note>

<Note>
  Use `brandMentioned` or `brandMentionedLabel` for a specific date/provider row.
  Use `promptBrandMentioned` or `promptBrandMentionedLabel` to isolate prompts
  where the brand never appears anywhere in the complete selected period.
</Note>

## Use this source for

* Comparing prompt-level observations
* Comparing funnel stages (TOFU, MOFU, BOFU)
* Tag-based filtering and analysis
* Reviewing prompt-level visibility trends


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.