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

# Enrichment

> AI enrichment that adds sentiment, emotions, and a consolidated language to open-text Feedback Records.

Enrichment automatically adds AI-generated context to your open-text feedback. When a Feedback Record has a text answer (`value_text`), Formbricks analyzes it and attaches three kinds of enrichment:

* **Sentiment**: how positive or negative the feedback is.
* **Emotions**: which emotions the text expresses.
* **Language consolidation**: a translated copy of the text in one common language, so feedback written in different languages can be compared and analyzed together.

These fields turn free text, which is normally hard to count, into structured signals you can filter, chart, and trend over time.

## How enrichment works

Enrichment is **automatic and per-record**. Individual records are never triggered or configured by hand; the only setting is the translation target language, which is configured once at the service level (see [Language consolidation](#language-consolidation)):

1. A Feedback Record is ingested with a text answer in `value_text`.
2. Formbricks enriches the record **asynchronously**. The enrichment fields are absent until processing finishes, so a record may show as **Not enriched** for a short time after it arrives.
3. Once complete, the sentiment, emotion, and translation fields appear on the record and in charts.

Only records with open text receive AI enrichment. Records that only carry a rating, NPS score, or categorical choice have no text to analyze. Categorical answers are still consolidated across languages through their **Value ID**, which is not AI enrichment. See [Consolidating the same option across languages](#consolidating-the-same-option-across-languages).

<Note>
  Enrichment fields are read-only. Editing a record's `value_text` clears its
  enrichment and re-queues it for processing. Editing a record's `language`
  re-runs only the translation.
</Note>

## Sentiment enrichment

Sentiment classifies each text answer on a polarity scale and assigns a signed score.

**Sentiment label** is one of:

| Label             | Meaning                                     |
| ----------------- | ------------------------------------------- |
| **Very positive** | Strongly positive feedback                  |
| **Positive**      | Positive feedback                           |
| **Neutral**       | Neither positive nor negative               |
| **Negative**      | Negative feedback                           |
| **Very negative** | Strongly negative feedback                  |
| **Mixed**         | Contains both positive and negative signals |

**Sentiment score** is a number from **-1.0** (very negative) to **1.0** (very positive). `Mixed` records score near 0.

Where you see it:

* **Record detail drawer**, under the **AI enrichment** section, e.g. `Positive (0.82)`.
* **Topics & Subtopics**, as a colored sentiment badge on each record card.
* **Charts**, where you can group by **Sentiment** or chart the average **Sentiment Score**. Records that haven't been enriched yet fall into a **Not enriched** bucket.

## Emotion enrichment

Emotion enrichment detects which emotions a text answer expresses. It is **multi-label**: a single record can carry several emotions at once (or none, before it's enriched).

The six recognized emotions are:

* **Joy**
* **Anger**
* **Sadness**
* **Fear**
* **Surprise**
* **Disgust**

Where you see it:

* **Record detail drawer**, listed in the **AI enrichment** section.
* **Topics & Subtopics**, as one colored badge per detected emotion on each record card.
* **Charts**, where you can chart the count of each emotion.

<Note>
  Because emotions are a multi-label set, filter charts on a single emotion
  using the **contains** operator rather than an exact-match equals.
</Note>

## Language consolidation

Feedback often arrives in many languages, which makes it hard to read and analyze together. Language consolidation solves this in two ways.

### Translation into a common language

Every open-text answer is translated into your organization's configured target language and stored alongside the original:

* The **translated text** is shown throughout Unify Feedback in place of the original, with a **Translated to {language}** chip. The original text is always available in the tooltip and in the record detail drawer.
* In **Topics & Subtopics**, a **Comment language** toggle lets you switch between the original and the translated text.

This means an analyst working in English can read, cluster, and chart feedback that was originally submitted in German, French, Spanish, and more, all consolidated into one comparable language.

<Note>
  The target language is set at the service level for your organization. On
  Formbricks Cloud it is managed for you; self-hosted instances configure it in
  the Hub service. See [Formbricks Hub](/docs/unify-feedback/formbricks-hub).
</Note>

### Consolidating the same option across languages

For multiple-choice answers, Formbricks also carries a stable **Value ID**, the option's identifier from the source system (for example, a survey choice id). This lets charts group the **same option** together even when its label was captured in different languages or later edited, instead of splitting it into a separate category per localized label.

When charting choice questions, group by **Value (Option)** rather than the raw text value to keep categories consolidated across languages.

## Using enrichment in charts

All enrichment fields are available in [Dashboards & Charts](/docs/unify-feedback/dashboards-charts) as dimensions and measures. Records that haven't finished enriching are grouped under a **Not enriched** value, so you can see enrichment coverage at a glance.

## Requirements

* A Feedback Dataset with records that contain open text.
* Enrichment runs on the Formbricks Hub. No local AI provider configuration is required to read enrichment fields.

<Note>
  Unify Feedback is an enterprise feature. Enable it on Formbricks Cloud with a
  paid plan, or self-host with a license.
</Note>