Tagging & Classification Overview

Duckie automatically classifies conversations using categories, attributes, and resolution rules — giving you structured data for analysis and reporting.

Why Classification Matters

Classification transforms unstructured conversations into actionable data:

Benefit	Example
Understand trends	”40% of tickets are billing-related”
Measure performance	”Technical issues have 85% resolution rate”
Route intelligently	”High-priority tickets go to senior agents”
Improve over time	”Most escalations happen in shipping category”

Classification Components

Attributes

Custom fields with predefined options:

Attribute	Options
Priority	High, Medium, Low
Product Area	Dashboard, API, Mobile App
Customer Tier	Enterprise, Pro, Free
Sentiment	Positive, Neutral, Negative

Conversations can have multiple attributes.

Resolution Rules

Conditions that determine when a conversation is “resolved”:

Customer confirmed the issue is fixed
No response needed (informational)
Time elapsed without follow-up

How Classification Works

At the end of each agent run:

Conversation ends
       │
       ▼
┌─────────────────────┐
│ Assign Category     │ → "Billing"
└─────────────────────┘
       │
       ▼
┌─────────────────────┐
│ Extract Attributes  │ → Priority: High
└─────────────────────┘   Product: API
       │
       ▼
┌─────────────────────┐
│ Evaluate Resolution │ → Resolved: Yes
└─────────────────────┘

The AI analyzes the conversation context and applies appropriate classifications.

Using Classification Data

In Analytics

View breakdowns by category and attribute:

In Run History

Filter and search by classification:

For Routing (Future)

Use classification for intelligent routing based on category or attributes.

Getting Started

Create Categories

Define the high-level conversation types for your business.

Create Attributes

Add custom fields for the dimensions you want to track.

Configure Resolution

Define what “resolved” means for your team.

Assign to Agents

Select which categories and attributes each agent should apply.

Next Steps

Attributes

Create custom attributes

Resolution Rules

Define what “resolved” means

Analytics

Analyze by classification

Getting Started

Building Agents

Training Your Agent