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

# Personas

> Add personas to measure how AI engines answer different types of buyers, and compare the results.

A persona is a system prompt that tells the AI engine who asks the question, for example "an IT manager at a large company". The same prompt can get a different answer for a different persona. Use personas to see which buyers AI engines recommend your brand to.

To open the page, click **Personas** in the sidebar.

## How personas affect scans

* Each scan asks every prompt under the baseline and under each persona that you add, on every tracked AI engine.
* The baseline is the default persona of all charts. When the baseline is the raw query, it asks the prompts with no persona and shows as **Raw**. You cannot edit or remove the baseline.
* A new persona has no historical data. It collects data from the next scan.
* A change to a persona applies from the next scan.
* Each persona that you add increases the number of answers that a scan collects.

> [!NOTE]
>
> The number of personas that you can add depends on your plan. See [pricing](https://perquery.ai/en/pricing).

To see the results of one persona on other pages, select it in the **Persona filter** at the top of the page.

## Add a persona manually

**1. Open the dialog**

Click **Add persona**. The **Manual** tab opens.

**2. Name the persona**

Type a name in **Persona name**, for example "Enterprise IT Manager".

**3. Select the kind**

In **Persona kind**, select **Demographic**, **Need-based**, **Behavioural** or **Custom**.

**4. Write the system prompt**

Type the system prompt. Describe who the persona is, their budget, what they weigh and what they fear. To get a draft, click **Write a detailed voice**. Edit the draft before you save.

**5. Save**

Click **Add**.

If you leave the system prompt empty, Perquery writes a detailed voice in the language of your site. It uses the name, the kind and the information about your brand.

## Add a suggested persona

**1. Open the suggestions**

Click **Auto-suggest** at the top of the page, or click the **Auto-suggest persona** card in the persona list.

**2. Read the suggestions**

Each suggestion shows a name, a kind, a system prompt and example prompts. The suggestions use your brand and its category.

**3. Add a persona**

Click **Add** next to each persona that you want. **Add** adds only the persona, not the example prompts.

To get new suggestions, click **Regenerate**. The number of suggestions that you can get in a period depends on your plan.

## The persona list

### Persona findings

Above the list, **Persona findings** compares your personas. It needs at least two measured personas. It can show these findings:

* Your visibility is almost the same for every persona. Nothing shows which buyers choose you.
* You are almost not visible for a persona that you marked as a target.
* You are strongest for a persona that is not a target, and a target persona is behind.

The last two findings show only when at least one persona is marked as a target. See [Mark a target persona](#mark-a-target-persona).

### Persona cards

Use the kind filter to show **All kinds** or one kind. Each card shows the name, the kind, the **Target** badge for a target persona, the system prompt and the **Summary visibility**. Click a card to open the persona detail.

The page uses the **Date range** at the top of the page. The list does not show the baseline.

### Persona × LLM matrix

The **Persona × LLM matrix** has one row for each persona and one column for each AI engine. Each cell shows the visibility of your brand for that persona on that engine. The **Avg.** column shows the average. Red means that your brand is rarely mentioned. Green means that your brand is mentioned in most answers. Click a row to open the persona detail.

## Persona detail

The detail page shows these values:

* **Summary visibility**: the visibility of your brand for this persona.
* **Tracked prompts**: the number of prompts that were measured for this persona.
* **Avg. LLM visibility**: the average of the visibility values of the engines.
* **Strongest platform**: the engine with the highest visibility.

The page also shows the **System prompt** and **Visibility by LLM**, with the **Weakest** and the **Strongest** engine.

### Mark a target persona

A target persona is a buyer that your brand is positioned for. **Persona findings** uses target personas.

To mark a persona, open it and click **Mark as target**. To remove the mark, click **Unmark target**.

### Edit a persona

**1. Open the form**

Open the persona and click **Edit**.

**2. Change the fields**

Change **Persona name**, **Persona kind** or **System prompt**.

**3. Save**

Click **Save**. The change applies from the next scan.

### Remove a persona

> [!WARNING]
>
> When you remove a persona, its tracking stops. The historical data stays in reports.

**1. Open the persona**

Click the persona card.

**2. Remove**

Click **Remove**, then click **Yes, remove**.

## Related pages

* [Performance Overview](https://docs.perquery.ai/en/monitor/dashboard.md)
* [Prompts](https://docs.perquery.ai/en/monitor/prompts.md)
* [Scans and schedule](https://docs.perquery.ai/en/sites/scans-and-schedule.md)


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