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

# Post-Chat Analysis

> Automatically extract insights and sentiment from completed conversations

**Post-Chat Analysis** uses AI to automatically analyze conversations after they end. It can extract structured data, summarize the conversation, and assess customer sentiment — giving you actionable insights without manual review.

## How It Works

1. A conversation ends (closed by an agent or auto-closed).
2. The AI analyzes the full conversation.
3. It generates:
   * **Summary** — A concise overview of what was discussed.
   * **Sentiment** — The customer's overall mood (Positive, Neutral, Negative).
   * **Extracted Data** — Custom fields you define (see below).

## Viewing Analysis Results

1. Open any completed conversation from the **Chat History** page.
2. In the **Live Chat Panel**, click the **Analysis** tab.
3. View the AI-generated summary, sentiment, and any extracted fields.

<Frame>
  <img src="https://mintcdn.com/quensultingai/Vk0F3QdCtadjJvc9/images/post-chat-analysis.png?fit=max&auto=format&n=Vk0F3QdCtadjJvc9&q=85&s=0e1d0fb950dd7c371bdc8938202688c4" alt="Post-chat analysis tab showing conversation summary, sentiment, and extracted data" width="543" height="596" data-path="images/post-chat-analysis.png" />
</Frame>

## Configuring Analysis Fields

You can define custom fields that the AI will extract from each conversation. This is configured per agent.

### Adding a Field

1. Open the **Agent Editor** for your chatbot.
2. Scroll to the **Post-Chat Analysis** section.
3. Click **Add Field**.
4. Configure the field:

| Setting         | Description                                                                                |
| --------------- | ------------------------------------------------------------------------------------------ |
| **Key**         | A unique identifier (e.g., `issue_category`)                                               |
| **Type**        | `text`, `number`, `boolean`, or `enum`                                                     |
| **Description** | Instructions for the AI on what to extract (e.g., "The main category of the user's issue") |
| **Options**     | For `enum` type only — list of allowed values (e.g., "Billing, Technical, General")        |

### Example Fields

| Key                  | Type    | Description                                                 |
| -------------------- | ------- | ----------------------------------------------------------- |
| `issue_category`     | enum    | Main category: Billing, Technical, General, Feature Request |
| `resolved`           | boolean | Whether the user's issue was fully resolved                 |
| `product_mentioned`  | text    | The specific product or feature the user asked about        |
| `satisfaction_score` | number  | Estimated satisfaction from 1-10 based on conversation tone |

<Tip>
  Well-written descriptions help the AI extract more accurate data. Be specific about what you want it to look for.
</Tip>

### Editing & Removing Fields

* Click on a field to edit its configuration.
* Click the delete icon to remove a field.

<Note>
  Changes to analysis fields apply to **future** conversations only. Previously analyzed conversations retain their original extracted data.
</Note>