Extract ERD – Data Model & Relationship Documentation
Updated
This document provides an overview of the Extract Entity Relationship Diagram (ERD) for all Voice, Digital, and Agent data extracts. The ERD shows the physical data model created from the available extracts and acts as the primary reference for understanding how data connects across domains. It explains how the extract datasets relate to each other, clarifies relationships within each extract, defines accurate data navigation paths, and supports reliable reporting and analytics.
Architectural Overview
The extract ecosystem is structured into three logical domains:
Domain | Anchor Entity | Purpose |
Voice | Conversation | Call-level lifecycle tracking |
Digital | Case | Case-level life cycle tracking. |
Agent | Agent | Agent Adherance and cross-domain bridge. |
Note: Each domain has a single anchor entity from which all related entities branch.
Domain Architecture
The domain architecture is explained for three domains. Each one of these is explained below:

Voice Domain Architecture
Digital Domain Architecture
Agent Domain Architecture
An anchor entity allows all extracts to be mapped consistently by providing one common reference point across the system, i.e. it can be used to map all the extracts.
Composite Key Combination
Apart from anchor entities, such as a primary key like Conversation_ID, Case_Number, or Agent_ID—all extracts rely on composite key combinations. These combinations are designed to:
Uniquely identify every event, interval, or record.
Prevent duplicate records by enforcing structural uniqueness.
Maintain the correct level of granularity for each entity.
Ensure accurate joins between related datasets.
Protect data integrity during reporting and aggregation.
Composite keys typically pair the anchor identifier with contextual attributes, such as timestamps, agent identifiers, queue identifiers, or event‑specific fields. This structure ensures that repeated lifecycle actions, like transfers, queue entries, macro applications, or status updates, are captured clearly and without ambiguity.
Voice Domain Architecture
Anchor Entity
Within the Voice Domain Architecture, the Conversation ID stands as the principal entity. This Conversation entity denotes a single voice interaction (or call) and functions as the foundational element for all voice-related reporting.
Composite Key Combination
You can refer to Composite Key Combination section to understand how it works. The examples of compostite keys for Voice Domain Architechture are enlisted below:
The composite keys for the Detailed Call State Extract are as follows:
Conversation ID
Call States
Caller for the particular conversation state
Call States Actual Time (Epoch)
Note: When reviewing individual articles, you can consult the column labeled "Unique Key" for further information on composite keys.
Voice Navigation Guide
Starting from Conversation_ID, you can access the complete interaction footprint, including:
All agent‑handling metrics
The full transfer and escalation history
Queue transitions and corresponding SLA behavior
The detailed IVR path navigated by the customer
Every call lifecycle state
Any associated callback records
Digital Domain Architecture
Anchor Entity
Within the Digital Service Architecture, the Case entity serves as the primary anchor. Identified by its Case_Number as the principal key, this entity represents a single customer case created across digital channels and forms the foundational element for all case‑related tracking, workflows, and reporting.
Composite Key Combination
You can refer to Composite Key Combination section to understand how it works. The examples of compostite keys for Digital Domain Architechture are enlisted below:
The composite keys for the Case Summary Extract are as follows:
Case Number
Case Creation Time
Channel (All)
Case Associated Customer
Digital Navigation Guide
Starting from Case_Number, you can retrieve:
Complete message history
Queue dwell time analysis
Agent assignment intervals
Macro usage behavior
Survey performance
Processing clock tracking
SLA adherence behavior
Agent Domain and Cross Domain Logic
Anchor Entity
Within the Unified Service Architecture, the Agent entity is defined by Agent_ID as its primary key and serves as the connective layer between the Voice and Digital domains. While these domains do not directly integrate with one another, they both anchor their operational alignment through the Agent_ID, making the Agent entity the pivotal bridge across the two ecosystems.
Composite Key Combination
You can refer to Composite Key Combination section to understand how it works. The examples of compostite keys for Digital Domain Architechture are enlisted below:
The composite keys for the Login and Logout Extract are as follows:
Agent ID
Session ID
Agent as Cross-Domain Bridge
The Voice and Digital domains don’t connect directly.
They link through:
Agent_ID
Note:
Voice and Digital are separate lifecycle domains.
No direct Conversation to Case mapping unless explicitly configured.
Comma-separated list fields should not be treated as relational entities.
Custom fields reflect snapshot/ current values as specified in each extract.
Best Practices
• Always join using defined pairs.
• Understand entity granularity before joining.
• Joining event tables may increase row counts.
• Aggregate carefully when combining interval and event entities.