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

# Queries & Tags

> Create, import, and organize the questions you send to AI platforms

Queries are the questions MentionLab sends to AI platforms. Tags group them so you can read results
by topic rather than one question at a time.

## Managing queries

### Creating queries

Go to **Project settings → Queries**. You can:

* **Add one by hand** — click **New Query**.
* **Import a CSV** — the upload icon opens "Bulk import queries from a CSV file".
* **Accept AI suggestions** — during project creation, MentionLab analyses your website and proposes
  prompts.

<Tip>
  Write queries the way a customer would. Instead of *"Is Acme good?"*, try *"What is the best
  project management tool for remote teams?"* — that is how people actually ask AI platforms.
</Tip>

### Query properties

| Property     | Description                                                                                                                |
| ------------ | -------------------------------------------------------------------------------------------------------------------------- |
| **Text**     | The question sent to AI models                                                                                             |
| **Language** | The language the query is written in ([BCP 47](https://en.wikipedia.org/wiki/IETF_language_tag), e.g. `en`, `fr`, `zh-cn`) |
| **Country**  | The market the query targets ([ISO 3166-1 alpha-2](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2), e.g. `US`, `FR`)     |
| **Type**     | Optional: **Comparative**, **Informative** or **Perception**. Untyped queries show as "Untyped"                            |
| **Tags**     | One or more query tags                                                                                                     |

<Warning>
  The country you pick decides which models can run the query — see
  [Models Availability](/guides/models-availability).
</Warning>

### Bulk operations

Select rows, then use the **Actions** menu:

* **Add tags** / **Remove tags** on many queries at once
* **Set type**
* **Launch partial analysis** for just the selected queries
* **Delete**

<Warning>
  Deleting a query also removes its historical results.
</Warning>

<Note>
  There is no export button on this page. Query exports live on **Project settings → Export data**,
  as the **Queries Summary** export type.
</Note>

***

## Reading query results

The **Queries** analysis page grades every query:

| Metric            | Description                                                  |
| ----------------- | ------------------------------------------------------------ |
| **Mention rate**  | Percentage of this query's responses that mention your brand |
| **Avg. position** | Where your brand ranks in those responses (lower is better)  |
| **Sentiment**     | Average sentiment score across the query's responses (0–100) |
| **Source rate**   | Percentage of the query's responses that include citations   |

Each numeric column has a min/max range filter in its header. Click a row to open **Query Analysis**,
which adds:

* Five metric cards: Total Responses, Mention Rate, Avg. Sentiment, Source Rate, Shopping Rate
* **Top Sources** and **Sources by Model**
* **Response History** — every individual run, with the full response text, entity chips and its
  source list
* **Entities Mentioned** and **Top Products**

***

## Query tags

### Creating tags

**Project settings → Queries → Tags**. A query tag has a **Name** and a **Color** — nothing else.

### Using tags

* **Overview** — the **Tags Summary** card shows mention rate and average position per tag.
* **Tags page** — a full comparison table (Results, Mention rate, SoV, Avg. position, Sentiment)
  plus a **Mention rate by tag** trend chart. Click any row for a detail sheet with entity ranking
  and top sources for that tag.
* **Platforms page** — the **Tags performance** heatmap breaks each tag down by model.
* **Filters** — narrow any analysis page to specific tags.

<Tip>
  Build tags around decisions you actually make: product lines, buyer personas, funnel stages,
  markets.
</Tip>

### AND / OR mode

When you select more than one tag in the filter bar, a mode toggle appears:

* **OR** — responses matching *any* selected tag
* **AND** — responses matching *all* selected tags (the default)

***

## Response tags

Query tags label the **questions**. Response tags label the **answers** — useful for patterns you
only recognise once you read the response ("recommends us", "competitor only", "mentions pricing").

Manage the definitions at **Project settings → Response Tagging → Tags**, and apply them to
individual responses at **Project settings → Response Tagging → All responses**.

### Matching mode

Unlike query tags, a response tag has a **Matching mode**:

* **Exact** (default) — matches responses whose text contains the tag name.
* **Semantic** — matches by text *and* meaning, and indexes the project's recent responses.

<Info>
  Turning on Semantic asks you to confirm, then indexes your existing responses in the background.
  A banner reports progress, and tag matches keep growing until indexing finishes.
</Info>

Response tags work as filters everywhere query tags do — they sit in the **Filters** popover rather
than the main bar, with their own AND/OR toggle.