Automated quality management and conversation analysis.
Overview
Quality AI enables you to:
- Evaluate 100% of customer interactions.
- Identify coaching opportunities automatically.
- Monitor compliance and adherence.
- Drive continuous improvement with data.
How It Works
┌───────────────────────────────────────────────────────┐
│ Customer Interactions │
│ (Voice, Chat, Email, Social) │
└───────────────────────────┬───────────────────────────┘
│ ▼
┌───────────────────────────────────────────────────────┐
│ Quality AI Engine │
│ │
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ Speech/ │ │ Evaluation │ │ Insight │ │
│ │ Text │ │ Scoring │ │ Generation │ │
│ │ Analysis │ │ │ │ │ │
│ └─────────────┘ └─────────────┘ └─────────────┘ │
└───────────────────────────┬───────────────────────────┘
│ ┌──────────────────┼──────────────────┐
│ ▼ ▼ ▼
┌────────────────┐ ┌───────────────┐ ┌────────────────┐
│ Auto-Scores │ │ Coaching │ │ Compliance │
│ & Evaluations │ │ Assignments │ │ Monitoring │
└────────────────┘ └───────────────┘ └────────────────┘
Evaluation Criteria
Standard Criteria
Pre-built evaluation criteria available out of the box:
| Criteria | Description |
|---|
| Greeting | Proper introduction and identification |
| Empathy | Acknowledging customer emotions |
| Issue understanding | Correctly identifying the problem |
| Resolution | Providing accurate solution |
| Closing | Proper wrap-up and next steps |
| Compliance | Following required disclosures |
Custom Criteria
Create custom criteria to match your business needs:
Criteria: Product Knowledge
Description: Agent demonstrates accurate product knowledge
Weight: 15%
Scoring:
- 5: Excellent - Comprehensive, accurate information
- 4: Good - Mostly accurate with minor gaps
- 3: Acceptable - Basic knowledge demonstrated
- 2: Below expectations - Significant gaps
- 1: Unacceptable - Incorrect information provided
Auto-evaluate: true
Keywords:
positive: ["correct", "accurate", "helpful"]
negative: ["wrong", "incorrect", "misinformation"]
Evaluation Forms
Group criteria into evaluation forms and apply them to queues:
Form: Customer Service Standard
Criteria:
- greeting (10%)
- empathy (15%)
- issue_understanding (20%)
- resolution (30%)
- product_knowledge (15%)
- closing (10%)
Pass threshold: 80%
Apply to:
- queue: support
- channel: all
Auto-Scoring
How It Works
AI evaluates interactions in four steps:
- Speech/text analysis — Transcribe and analyze the conversation.
- Criteria matching — Map conversation content to evaluation criteria.
- Scoring — Assign scores based on evidence found.
- Confidence flagging — Flag low-confidence scores for human review.
Configuration
Auto-Scoring:
enabled: true
evaluation_rate: 100% # Evaluate all interactions
confidence_threshold: 0.8
human_review:
- low_confidence_scores
- failed_evaluations
- random_sample: 5%
Calibration
Keep scores consistent across evaluators:
- Select a calibration sample.
- Have multiple evaluators score the same interactions.
- Compare scores and discuss differences.
- Refine criteria definitions.
- Re-train the auto-scoring model.
Conversation Mining
Topic Analysis
Automatically identify and track conversation topics:
| Analysis | Description |
|---|
| Topic clustering | Group conversations by theme |
| Trend detection | Identify emerging topics |
| Sentiment by topic | Track sentiment for each topic |
| Volume tracking | Monitor topic frequency |
Root Cause Analysis
Identify what is driving quality issues:
Quality Issue: Low resolution scores in billing queue
Root Causes Identified:
├── 45% - Complex billing system navigation
├── 30% - Outdated knowledge articles
├── 15% - Missing escalation paths
└── 10% - Training gaps on new features
Insights Dashboard
Automatic insights surface:
- Top coaching opportunities
- Emerging quality trends
- Best performing agents and teams
- Areas needing attention
Coaching
Auto-Assignment
Trigger coaching automatically based on evaluation scores:
Coaching Rule: Resolution Improvement
Trigger:
criteria: resolution
score: < 3
count: 3 consecutive
Action:
assign_coaching: resolution_training
notify: supervisor
priority: high
Coaching Workflow
┌─────────────────────────────────────┐
│ [Quality Issue Detected] │
│ │ │
│ ▼ │
│ [Coaching Assigned to Supervisor] │
│ │ │
│ ▼ │
│ [Supervisor Reviews Evidence] │
│ │ │
│ ▼ │
│ [Coaching Session Scheduled] │
│ │ │
│ ▼ │
│ [Session Completed] │
│ │ │
│ ▼ │
│ [Follow-up Evaluation] │
└─────────────────────────────────────┘
Evidence Attachment
Each coaching assignment includes:
- Conversation transcript
- Audio recording
- Evaluation scorecard
- Specific timestamps and sections
- Comparison to best practices
Compliance Monitoring
Compliance Rules
Define rules to enforce regulatory requirements:
Compliance: PCI-DSS Card Handling
Rules:
- must_not_say: ["full card number", "CVV"]
- must_say: ["secure", "encrypted"]
- action_required: mask_card_data
Alert:
severity: critical
notify: compliance_team
Required Disclosures
Track mandatory script elements by interaction type:
| Disclosure | Required For |
|---|
| Recording notice | All calls |
| Rate disclosure | Financial products |
| Terms and conditions | New accounts |
| Privacy policy | Data collection |
Compliance Dashboard
Monitor compliance health at a glance:
- Compliance rate by disclosure type
- Violations by agent and team
- Trend analysis
- Alert history
Taxonomy Builder
Create Taxonomies
Organize quality categories into a structured hierarchy:
Taxonomy: Quality Categories
├── Communication
│ ├── Clarity
│ ├── Tone
│ └── Active listening
├── Knowledge
│ ├── Product
│ ├── Process
│ └── Policy
├── Problem Solving
│ ├── Issue identification
│ ├── Solution accuracy
│ └── Efficiency
└── Compliance
├── Disclosures
├── Data handling
└── Regulatory
Apply Taxonomies
Use taxonomies to structure:
- Evaluation forms
- Analytics categorization
- Coaching focus areas
- Reporting dimensions
Analytics
Quality Dashboards
| Dashboard | Metrics |
|---|
| Overview | Quality score trends, pass rates |
| Agent performance | Individual scores, improvement |
| Team comparison | Team-level benchmarking |
| Criteria analysis | Performance by criteria |
| Compliance | Compliance rates, violations |
Reports
Automated reports delivered on schedule:
- Daily quality summary
- Weekly team performance
- Monthly trend analysis
- Compliance audit reports
Setup Quality AI
Complete these steps in order to get Quality AI running.
1. Configure Permissions
- Go to User Management > Role Management > New Role > Other Modules. Learn more.
- Assign the Supervisor role or create custom roles with QM permissions. Learn more.
2. Set Up Contact Center
- Assign Supervisors and Auditors to the relevant queues so they can access the right interactions. Learn more.
3. Enable Features
- Enable Conversation Intelligence, Auto QA, and Bookmarks in Quality AI Settings. Learn more.
- Enable Answer and Utterance suggestions in GenAI Settings. Learn more.
4. Create Evaluation Metrics
- Choose a measurement type: By Question, Question Answer Pair, or Adherence (Static or Dynamic). Learn more.
- Create evaluation metrics. Learn more.
- Set the count type: Entire Conversation or Time Bound. Learn more.
5. Create Evaluation Forms
- Assign a name, description, channel, and pass score.
- Select metrics, assign weights, and link the form to queues.
- Learn more.
6. Analyze Interactions in Conversation Mining
- Use filters to review scored interactions. Learn more.
- Save filters to reuse them in audit assignments. Learn more.
7. Create Audit Allocations
- Assign interactions to auditors for manual evaluation. Learn more.
8. Run AI-Assisted Manual Audits
- Use AI-assisted audits for faster, more consistent scoring.
- Navigate interactions using adherence moments and violations.
- Learn more.
9. Monitor Performance
- Use the Dashboard to track individual QA progress and queue statistics. Learn more.
- Use the Conversation Intelligence Dashboard for contact center-wide performance trends. Learn more.
