Voice QA Setup Guide

This guide walks you through setting up Voice QA from start to finish — from connecting your phone system to configuring evaluation rules and interpreting results.

---

Table of Contents

1. Overview
2. Step 1: Connect Your Phone System
3. Step 2: Create Departments
4. Step 3: Configure Evaluation Rules
5. Step 4: Set Up Categories
6. Step 5: Add Staff
7. Step 6: Manage Clients
8. Understanding Evaluations
9. Scoring System Explained
10. Reading Evaluation Results
11. Best Practices
12. Troubleshooting

---

Overview

Voice QA is an AI-powered call quality assurance system that:

• Transcribes every call automatically
• Evaluates calls against your custom rules
• Scores performance as a percentage (e.g., 86%/100%)
• Provides detailed feedback for each rule

Pricing

• 12 credits per minute of audio analyzed
• Includes transcription, AI evaluation, and all analytics

---

Step 1: Connect Your Phone System

Voice QA integrates with PBX phone systems to automatically receive call recordings.

Supported Platforms

Platform	Region	Description
Orange Moldova	Moldova	Direct PBX integration
Moldcell	Moldova	Direct PBX integration

Other PBX Systems

Voice QA can be configured to work with any PBX system. Contact Support for guidance on integrating your specific phone system.

Configuration Steps

1. Navigate to Voice QA → Integration
2. Select your platform (Orange or Moldcell)
3. Enter your API credentials:
   • API Key
   • API Secret
   • Account ID
4. Configure department mapping
5. Enable synchronization
6. Test the connection

Once connected, calls automatically flow into Voice QA for processing.

---

Step 2: Create Departments

Departments organize your call center by team or function. Each department has its own:

• Staff members
• Clients
• Evaluation rules
• Performance metrics

Creating a Department

1. Navigate to Voice QA → Departments
2. Click Add Department
3. Fill in:

Field	Description
Name	Department identifier (e.g., "Customer Support")
Description	Purpose and scope
Scoring Type	Fixed or Weighted (see Scoring System)

4. Click Create

Department Structure

Department
├── Staff (agents assigned to this team)
├── Clients (customers who call this team)
├── Evaluation Rules
│   ├── General Category (always applied)
│   └── Specific Categories (applied when relevant)
└── Performance Statistics

---

Step 3: Configure Evaluation Rules

Evaluation rules define what agents should do during calls. The AI evaluates each call against these rules and provides a score.

Rule Types

Voice QA supports two scoring types:

Fixed Scoring (0-10 Scale)

Every rule is evaluated on a consistent 0-10 point scale.

Rule	Score	Max
Used proper greeting	8	10
Verified customer identity	10	10
Resolved the issue	7	10

Best for: Consistent evaluation where all rules have equal importance.

Weighted Scoring (Custom Scale)

Each rule has its own scale based on importance:

Rule	Scale	Importance
Used proper greeting	0-3	Medium
Verified customer identity	0-5	High
Resolved the issue	0-10	Critical

Best for: Nuanced evaluation where some behaviors matter more than others.

Creating Rules

1. Open your department
2. Navigate to Rules section
3. Click Add Rule
4. Configure:

Field	Description
Text	What the rule evaluates (e.g., "Agent used customer's name")
Weight/Max Score	Maximum points for this rule
Call Type	Incoming, Outgoing, or Both
Category	Which category this rule belongs to

Example Rules

Customer Support Rules:

Rule	Scale	Type
Agent greeted customer professionally	0-3	Incoming
Agent verified customer identity	0-5	Both
Agent listened actively to the issue	0-5	Both
Agent provided accurate information	0-5	Both
Agent resolved the issue or escalated properly	0-10	Both
Agent thanked customer at end of call	0-2	Both

---

Step 4: Set Up Categories

Categories group related rules and define when they should be applied.

How Categories Work

• General Category — Always applied to every call (default)
• Specific Categories — Applied only when relevant to the call type

The AI automatically matches the call to relevant categories based on the conversation content.

Creating Categories

1. In your department, go to Rules → Categories
2. Click Add Category
3. Configure:

Field	Description
Name	Category identifier (e.g., "Sales Calls")
Description	When this category applies — be specific!

4. Add rules to this category

Example Categories

Category	Description	Applied When
General	Core quality standards	Every call
Sales	Sales-specific behaviors	Customer shows purchase interest
Technical Support	Technical issue handling	Customer reports technical problem
Complaints	Complaint handling	Customer expresses dissatisfaction
Billing	Payment and account issues	Financial topics discussed

Category Descriptions

Write clear descriptions that help the AI understand when to apply each category:

✅ Good: "Apply when the customer is inquiring about purchasing a product or service, comparing options, or asking about pricing."

❌ Bad: "Sales calls"

---

Step 5: Add Staff

Staff members are the agents whose calls are evaluated.

Adding Staff

1. Navigate to Voice QA → Staff
2. Click Add Agent
3. Fill in:

Field	Required	Description
First Name	Yes	Agent's first name
Last Name	Yes	Agent's last name
Email	No	For notifications
Department(s)	Yes	Assign to one or more departments
Alias	No	Display name

4. Click Add

Staff Performance Tracking

Each agent's profile shows:

• Overall average score
• Score trend over time
• Category breakdown
• Rule-by-rule performance
• Recent evaluations

---

Step 6: Manage Clients

Clients are customers whose calls are tracked over time.

Adding Clients

1. Navigate to Voice QA → Clients
2. Click Add Client
3. Fill in:

Field	Description
Name	Customer or company name
Type	Individual or Company
Phone Numbers	All associated numbers with labels

Phone Number Matching

When a call comes from a known phone number:

1. Voice QA automatically links it to the client
2. Interaction history is updated
3. You can track quality trends per client

---

Understanding Evaluations

When a call is evaluated, Voice QA produces comprehensive results.

Evaluation Process

1. Transcription — Voice converted to searchable text
2. Category Matching — AI identifies relevant categories
3. Rule Evaluation — Each applicable rule is scored
4. Score Calculation — Total percentage computed
5. Insights Generated — Summary, intent, and recommendations

What Categories Are Applied

For each call, the AI applies:

• General Category — Always included
• Matching Specific Categories — Based on conversation content

Example: A sales inquiry about a billing issue would trigger:

• General ✓
• Sales ✓
• Billing ✓

---

Scoring System Explained

How Scores Are Calculated

The final score is a percentage representing points earned vs. points possible.

Score = (Points Earned / Points Possible) × 100%

Fixed Scoring Example

All rules on 0-10 scale:

Rule	Score	Max
Greeting	9	10
Identity Verification	10	10
Issue Resolution	8	10
Closing	7	10
Total	34	40

Final Score: 85% (34/40)

Weighted Scoring Example

Rules with custom scales:

Rule	Score	Max
Greeting (medium)	2	3
Identity Verification (high)	5	5
Issue Resolution (critical)	8	10
Closing (low)	2	2
Total	17	20

Final Score: 85% (17/20)

Score Interpretation

Score	Level	Meaning
90-100%	Excellent	Consistently follows best practices
75-89%	Good	Minor areas for improvement
60-74%	Average	Several areas need attention
Below 60%	Needs Improvement	Requires coaching

---

Reading Evaluation Results

Each evaluated conversation provides detailed insights.

Overview Tab

Field	Description
Score	Percentage achieved (e.g., 86%/100%)
Satisfaction	Customer satisfaction rating (0-10)
Sentiment	Positive, Neutral, or Negative
Intent	Short summary of what the customer wanted
Recommendations	Suggestions for improvement
Transcript	Full text of the conversation

Evaluation Tab (Rule-by-Rule)

Switch to the Evaluation tab to see detailed rule analysis.

For each rule, you'll see:

Section	Description
Score	Points earned vs. maximum (e.g., 8/10)
Evidence	Direct quotes from the conversation
Analysis	AI explanation of why this score was given
Improvement Suggestion	Specific advice for doing better

Example Rule Evaluation

Rule: "Agent verified customer identity before accessing account"

Field	Content
Score	5/5 ✓
Evidence	"Could you please confirm your date of birth and the last four digits of your account number?"
Analysis	Agent properly verified identity using two authentication factors before proceeding with account access.
Suggestion	None needed — excellent adherence to security protocol.

---

Best Practices

For Creating Rules

• ✅ Be specific and measurable
• ✅ Use objective criteria
• ✅ Align with business goals
• ✅ Keep to 10-15 rules per category
• ❌ Avoid vague terms like "was friendly"
• ❌ Don't create overlapping rules

For Writing Category Descriptions

• ✅ Explain when the category applies
• ✅ Give specific examples
• ✅ Use clear, unambiguous language
• ❌ Don't use single-word descriptions
• ❌ Don't assume context

For Managers

• Review trends, not just individual scores
• Use rule-by-rule data to identify training needs
• Celebrate high performers
• Update rules as standards evolve

For Agents

• Review feedback after each evaluation
• Pay attention to Evidence quotes
• Apply Improvement Suggestions
• Track your own progress over time

---

Troubleshooting

Calls Not Appearing

Check	Solution
Integration enabled?	Verify sync is on in Integration tab
Correct department?	Check department mapping
Date range?	Adjust filters

Low Scores Unexpected

Issue	Solution
Rules unclear	Rewrite rules with specific criteria
Category mismatch	Improve category descriptions
Wrong call type	Check Incoming/Outgoing settings

Wrong Categories Applied

Issue	Solution
Vague description	Make category description more specific
Overlapping categories	Differentiate category triggers

Audio Quality Issues

Issue	Solution
Transcription errors	Check original audio quality
Missing segments	Verify complete file upload

---

Need Help?

• Email: [email protected]
• Documentation: Voice QA Overview
• PBX Integration: Contact support for custom PBX configuration

---

Related Documentation

• Voice QA Overview — Module introduction
• Departments — Department configuration
• Conversations — Managing recordings
• Staff — Agent performance
• Clients — Customer profiles
• Integration — Phone system setup