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

# Intent classification

> The nine-bucket intent system that drives demand eligibility and format selection.

Every conversation turn is classified into one of nine intent buckets. The bucket determines which demand tiers are eligible, which format renders, and what outcome is measured.

| Bucket        | Signal                                      | Example                                       |
| ------------- | ------------------------------------------- | --------------------------------------------- |
| Purchase      | Active buying intent                        | "I want to buy a standing desk"               |
| Transactional | Ready to complete a transaction             | "Where can I order this tonight"              |
| Consideration | Evaluating options                          | "Should I get the hybrid or the foam one"     |
| Comparison    | Directly comparing products or services     | "Explorer vs Highlander for a family of five" |
| Research      | Gathering information before a decision     | "What matters in a mattress for back pain"    |
| Informational | General knowledge seeking                   | "How do memory foam mattresses work"          |
| Navigational  | Looking for a specific brand or destination | "Tempur-Pedic official site"                  |
| Support       | Help with an existing product               | "My robot vacuum won't dock"                  |
| Entertainment | Non-commercial conversation                 | Casual chat, games, creative writing          |

## What publishers see

Reporting breaks down impressions, engagement, and revenue by intent bucket, so you can see where your surface's commercial moments are. High Purchase and Transactional share correlates directly with RPM.

## What is not documented here

Classification model internals are proprietary. Publishers do not need to pass an intent value; classification runs on the conversational context your integration already provides.