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

# Sentiment: how AI describes your brand

> The sentiment score rates from 0 to 100 how positively an AI answer describes your brand. Learn when it is scored and how to act on it.

Sentiment in Trackbase measures tone: when an AI model mentions your brand, does it describe you positively, neutrally or negatively? This page explains when sentiment is scored and how to use it.

## How it is scored

For every answer that mentions your brand, Trackbase judges the text around each mention and gives:

* a **label**: positive, neutral or negative
* a **score** from 0 to 100, where higher is more positive

Sentiment is only scored when your brand is mentioned. An answer without a mention has no sentiment. It is not counted as neutral, so answers that skip you do not pull your average toward the middle.

## Where you see it

* On the **AI Visibility** page, in the **Sentiment** tab.
* On the **AI Visibility** page, in the **Mentions** tab, as average sentiment.

## How it relates to visibility

Sentiment is part of the [visibility score](https://academy.trackbase.ai/visibility-score.md), so a more positive description raises that score a little. The two still answer different questions. You can be mentioned first in an answer that mainly lists your downsides.

## Acting on weak sentiment

1. Open the prompts where sentiment is lowest and read the answers in full.
2. Note which claims come back. Repeated criticism usually traces to a small number of sources.
3. Check the cited sources. An outdated review or comparison page is often the origin.
4. Fix what is true, and publish clear, current information for what is not.


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