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

# Taxonomy

Every metric in Nebuly is powered by a taxonomy.

A taxonomy is the set of groups Nebuly uses to organize interactions and conversations into meaningful business concepts. Instead of you having to analyze thousands of individual interactions, you can use Nebuly to automatically classifies each conversation into structured dimensions—such as **Topics**, **Business Risks**, **User Intent**, **Failure Types**, and **Sentiment**—so you can measure, compare, and explore your AI usage at scale.

Without a taxonomy, analytics can tell you **how many** conversations occurred. With a taxonomy, they tell you **what users were trying to do, where they struggled, and what matters most to your business.**

## Why taxonomies matter

AI conversations are unstructured by nature. Two users may ask the same question using completely different words, while similar-looking prompts may actually represent different business needs.

Nebuly's taxonomy automatically transforms these conversations into structured business data, allowing you to:

* **Understand what users are trying to accomplish**, rather than simply counting conversations.
* **Measure business outcomes** such as adoption, failures, satisfaction, and ROI by topic or use case.
* **Identify trends** as customer needs evolve over time.
* **Compare performance** across products, departments, teams, or business processes.
* **Build reports** using meaningful business dimensions instead of raw conversation data.

Because every report uses the same taxonomy, metrics remain consistent across the entire platform.

## How it's organized

Taxonomy has two levels:

* **Groups** are classification axes, such as *Topics* or *Car models*. Each group answers one question about an interaction.
* **Classes** are the values inside a group. For example, the *Topics* group might contain *Billing*, *Returns*, and *Technical support*.

Every interaction is sorted into the relevant class of each group, and that classification is what your reports aggregate.

## Group types

Groups come in three types, shown in separate sections with a type badge:

* **System** groups are defined by Nebuly and populated automatically. These include Topics, User Intent, Business Risks, Type of Failure (failure intelligence), Conversation Health, User Sentiment, Language, User Emotion, and others.
* **Tag** groups are created automatically from the tag keys you send during integration. Their classes are the distinct values received for each tag. Read more about tags [here](/guides/tags).
* **Custom** groups are the ones you create yourself to cluster interactions along a dimension that matters to you.

### System taxonomies

Nebuly includes a set of built-in taxonomies that are automatically populated using proprietary AI models. These provide immediate visibility into the most important aspects of your AI applications without requiring any configuration.

System taxonomy in detail:

* **Topics** – What users are trying to accomplish.
* **User Intent** – The intent behind each interaction.
* **Business Risks** – Conversations that indicate potential business issues.
* **Failure Intelligence** – Why conversations failed from the user's perspective.
* **Conversation Health** – Overall conversation outcome.
* **User Sentiment** – How users felt during the interaction.
* **User Emotion** – The emotions expressed throughout the conversation.
* **Language** – Languages detected across conversations.
* *(and other built-in taxonomies as they become available).*

### Custom taxonomies

In addition to the built-in taxonomies, you can create your own taxonomy groups and classes that reflect your organization's terminology, products, services, workflows, or business processes.

Custom taxonomies let you organize conversations around the dimensions that matter most to your business, making them available throughout Nebuly's reporting and analytics engine alongside the system taxonomies.

For a complete guide to creating and managing taxonomies, see [**Taxonomy Setup**](/guides/setup/taxonomy).

### Tag taxonomies

Tag taxonomies are created automatically from the metadata you send through the **Interaction API**. Every unique tag key (for example, `department`, `country`, or `customer_tier`) becomes its own taxonomy, while the distinct values become its classes.

Unlike system taxonomies, which Nebuly infers from conversations, tag taxonomies provide **business context** that cannot be derived from the conversation itself. They can be used throughout Nebuly to filter, group, and compare analytics across teams, regions, customer segments, or any other metadata you provide.

To learn how to send tags and best practices for designing your tagging strategy, see **[Tags](/guides/tags).**