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

# Brand

> What AI holds against your brand and credits it with in answers to your tracked prompts, by attribute, topic, and source, with the sentences and pages behind each one.

![Image](https://mintcdn.com/peecai-b61a1960/DDWEdCiIj1Q00pIb/images/image-31.png?fit=max\&auto=format\&n=DDWEdCiIj1Q00pIb\&q=85\&s=b2bfdbae933ea39ff4edfc3b8cf96aa0)

The **Brand** tab helps you understand how AI models describe your brand in responses to your tracked prompts. It shows how often your brand is associated with positive or negative attributes, along with the specific statements and cited pages behind those associations.

Peec AI analyzes each response to your tracked prompts and identifies the words and phrases AI models use to describe your brand. These are grouped into **attributes**, such as pricing, fit, or customer support.

Each attribute is classified as:

* **Positive:** When AI models highlight an attribute as a strength of your brand.
* **Negative:** When AI models describe an attribute as a weakness or disadvantage of your brand.

> [!NOTE]
>
> The **Brand** and **Fact-checking** tabs analyze responses to your tracked prompts, while **Market** and **Objections** use Peec AI's own set of prompts.
>
> Because they analyze different data, the same attribute may show different results across these views.
>
> **Market** measures how strongly AI models associate your brand with an attribute using a score from 0 to 100. The **Brand** tab, on the other hand, counts how many responses describe that attribute positively or negatively. These metrics measure different things and shouldn't be compared directly.
>
> All filters at the top of the page apply to this tab, including **date range, model, topic, tag, and country**.

## Summary

The header shows that the data comes **From your tracked prompts**, along with when the latest AI response was collected. The headline highlights the **percentage of responses that describe your brand negatively**.

Below it, the subtitle identifies the attributes that receive more negative mentions than positive ones, helping you understand where AI models see weaknesses in your brand.

If no attribute has more negative mentions than positive ones, the subtitle highlights the attribute with the highest share of negative mentions instead.

* **Describing you:** Answers that describe you with at least one attribute, out of every answer to your tracked prompts.
* **Positive answers:** Share of all answers that credit you with at least one attribute.
* **Negative answers:** Share of all answers that hold at least one attribute against you.
* **Contradicted claims:** Claims AI makes about you that contradict your facts, taken from [Fact-checking](https://docs.peec.ai/brand-perception-fact-checking.md).

> [!NOTE]
>
> Both percentages are calculated using **all responses to your tracked prompts**, including those that don't mention any attributes of your brand.
>
> A single response can describe one attribute positively and another negatively. In this case, it counts toward both percentages.
>
> This means the **positive and negative shares can add up to more than 100%**.

## Working through the tab

1. Read the headline and the **Negative answers** share to see how often AI holds something against you.
2. Click the negative segment of the ring to filter the section to the answers that hold something against you.
3. In **By attribute**, open the attribute held against you most and read the passages behind it.
4. Open a row to load its **Sources**, the pages AI cites where it says it.
5. If the same point shows up as contradicted under **What AI claims about you**, open it to see which of your facts it contradicts.

![Brand tab walkthrough](https://mintcdn.com/peecai-b61a1960/v40XFqKnuHKJkoss/images/brand-perception-brand-loop.gif?s=1982a3369a3c1345b51e8d70da4b9d40)

## How AI describes you

The section includes two charts to help you understand how AI models describe your brand:

* **Most mentioned attributes:** Shows which attributes AI models most frequently describe positively or negatively, counting each attribute only once per response. Switch between **Most negative** and **Most positive** to explore your brand's strengths and weaknesses. The chart displays nine attributes by default, but you can expand it to see the rest.
* **How AI judges you:** Shows all attribute mentions during your selected time period, grouped into **positive**, **negative**, and **mixed** to give you an overview of how AI models describe your brand.

> [!TIP]
>
> Click a segment of the ring, or its legend, to filter the whole section to negative or positive answers. The **All**, **Negative**, and **Positive** toggle does the same, and the bar chart follows whichever is set.

Below the charts, three lenses read the same answers. Each has a search box and shows ten rows a page.

* **By attribute:** Every attribute AI uses for you, with the attributes held against you most listed first.
  * **Negative:** Answers that hold the attribute against you.
  * **Positive:** Answers that credit you with it.
  * **Share negative:** The share of the answers judging the attribute that hold it against you.
  * **Negative vs positive:** A red and green bar of how the answers judging the attribute split.
  * **Sources:** The pages the answers cite where they state the attribute, most cited first. They show once you open the row.
* **By topic:** Each of your topics with its **Answers**, the attribute **Held against you most**, and the same negative, positive, share, and split columns. Open a topic to see its attributes on the same grid.
* **By source:** The pages AI cites at the point where it describes you, with the pages tied to the most negative answers first, and the same negative, positive, share, and split columns. **Contradicted** counts the contradicted claims made right where the page is cited. Open a page to see the attributes and claims stated there.

> [!TIP]
>
> Hover over a bar to see the counts and the share held against you. Click a column header to sort by it.

> [!NOTE]
>
> Each response is counted only once per attribute and direction, even if AI mentions the same point multiple times. An attribute must appear in at least **five responses that judge it positively or negatively** before a percentage is calculated.
>
> If fewer than five responses qualify, a dash (**–**) appears instead of a percentage, and the attribute is listed below those with calculated percentages.
>
> Responses that mention an attribute without describing it positively or negatively are excluded from both the table and the chart.

### Attribute details

![Attribute details](https://mintcdn.com/peecai-b61a1960/v40XFqKnuHKJkoss/images/brand-perception-brand-attribute-details.png?fit=max\&auto=format\&n=v40XFqKnuHKJkoss\&q=85\&s=442b76aa75d3d521eb70323a5d205b85)

Click on any attribute to open a table showing all AI responses that describe it positively or negatively.

* **Attribute:** A short description of what AI models say about your brand in relation to that attribute.
* **In the answer:**  The exact passage where AI describes the attribute, displayed as it appears in the original chat.
* **Answer:** Whether the response describes your brand positively or negatively.
* **Model** and **Topic:** The AI model that generated the response and the topic of the tracked prompt.
* **Sources:** The pages cited in the relevant passage. These appear when you expand a row. Hover over a source to see the page and its type.

Some attributes may also appear in other sections of Brand Perception:

* **In Market:** This tag appears when an attribute is also tracked in the **Market** tab. Hover over it to see the attribute's association score, your brand's ranking, and the competitors included in Market. Click the tag to open the attribute directly in **Market**.
* **In Objections:** This tag appears when an attribute is also included in one of your tracked objections. Hover over it to view the objection's score, or click it to open the objection in **Objections**.

> [!TIP]
>
> Click any answer to open the full chat, with the relevant passage automatically highlighted and brought into view.

## What AI claims about you

This section shows the claim groups from [Fact-checking](https://docs.peec.ai/brand-perception-fact-checking.md), helping you identify statements AI models make about your brand that contradict your facts. Groups with the highest number of contradicted claims appear first.

The table includes:

* **Claim group:** Groups claims that communicate the same point during your selected time period. Claims that don't belong to a group appear under **Other claims**.
* **Claims:** The total number of claims within each group during your selected time period.
* **Contradicted:** The number of claims that contradict at least one of your facts.
* **Share contradicted:** The percentage of claims within the group that contradict one of your facts.

Click a **claim group** to explore its individual claims, then open a specific claim to see which facts it was checked against. You can also use the **search bar** and **verdict filter**, just like in Fact-checking, to narrow your results. These filters update both the claims displayed and their corresponding counts.

## Best practices

* **Start from the ring:** The split between negative and positive mentions tells you whether the problem is a few attributes or the overall tone. Click the negative segment and read the attributes it leaves.
* **Rank by the negative share:** An attribute with many negative answers can still be one AI mostly credits you with. The bar shows which way it leans.
* **Read the answers before deciding what to fix:** The attribute name is a summary. The passages show what AI says, and the pages cited on them show where it comes from.
* **Compare topics:** Use **By topic** to see where an attribute is held against you and where it's credited.
* **Check the claims next to the attributes:** When a negative attribute also shows up as a contradicted claim, AI is stating something your own facts say is wrong.


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