100% call and chat coverage: How AI-powered quality insights transform agent coaching and contact center operations

For years, quality assurance (QA) in the contact center has been a necessary yet fundamentally flawed process that has created an unacceptable risk profile and limited the effectiveness of agent coaching.

The inherent gaps in the traditional QA model have long included:

• The 1% problem (lack of visibility): Due to high volumes and manual labor, QA analysts can review only a tiny fraction — typically 1–2% — of all customer interactions (e.g., calls, chats, emails and more). This left 98–99% of conversations unchecked, meaning critical compliance breaches or systemic service failures could go entirely unnoticed.
• Reactive, biased coaching: Agent coaching has been based on small, unrepresentative samples. Furthermore, human quality scoring is inherently prone to variability and unconscious bias, leading to inconsistent results and feedback that agents often perceive as subjective or punitive.
• Unquantified compliance risk: Without a comprehensive, objective review, we've lacked true visibility into regulatory adherence. This exposure represents a significant financial and reputational risk, costing upwards of millions of dollars per year in unnecessary credits, potential fines or lost business.

For TELUS, a leading Canadian telecom and parent company of TELUS Digital, the opportunity was clear: Move from a manual, sampling-based and subjective process to an objective, data-driven system that provides 100% visibility and enables proactive, personalized agent coaching.

Together, we built an AI-powered solution called Agent Quality Insights that:

• Delivers 100% coverage of calls and chats
• Gives supervisors more than 30% of their admin time back for direct, actionable coaching sessions
• Reduces spending on customer billing credits by 20% thanks to proactive service

Any enterprise can deploy Agent Quality Insights to cost-effectively modernize its contact center operations. Here’s how our solution uses AI to transform your agent coaching, compliance and more.

1. Using AI to unlock the value in unstructured contact center data

The vast majority of contact center value is locked within unstructured data — the actual words spoken and typed between agents and customers. Prior to the rise of advanced conversational AI, this data was functionally unusable for large-scale analysis. We relied on structured data, such as call duration and disposition codes, while ignoring the core content.

Our solution, Agent Quality Insights, is built on powerful machine learning (ML) and natural language processing (NLP) models. It delivers insights for agent coaching and more thorough analysis impossible to achieve manually.

• Unlocking every conversation: Agent Quality Insights first uses speech-to-text (STT) models and NLP to transcribe and analyze every single word. This transforms millions of minutes of unstructured audio and text into a massive, searchable and quantifiable dataset.
• Bias mitigation and contextual intent: Unlike manual scoring, where reviewer fatigue or personal interpretation can influence the outcome, an AI-based system can evaluate conversations in two complementary ways. Together, this hybrid approach delivers consistent, objective scoring at scale while still capturing the nuance of real customer conversations:
  • A rules-based layer that reliably checks for the presence or absence of required phrases and behaviors
  • An LLM-powered, contextual layer that understands intent, phrasing variations and conversation flow even when agents don’t use exact keywords

2. Defining the function and scope needed for complete quality coverage

We wanted our Agent Quality Insights tool to automatically analyze, score and report on every interaction, giving us instant access to complete quality coverage. Realizing that scope meant we’d need to build four key features:

• Interaction ingestion and enrichment
• Automated custom scorecards/rubrics
• Strategic topic modeling
• Multi-language support

Interaction ingestion and enrichment

Our agent insights solution processes 100% of calls and chats, transcribing and analyzing over 40 million minutes of conversation per month, while also enriching that data with key attributes to drive actionable insights.

Business value: Eliminates the critical ‘sampling gap,’ ensuring every customer interaction is reviewed for compliance and quality.

Automated custom scorecards/rubrics

We defined a custom scorecard (or scoring rubric) aligned with our business rules, including specific checks for script adherence, compliance and sentiment/customer experience.

Business value: AI automatically scores every conversation based on these objective criteria, providing consistent, scalable evaluations.

Strategic topic modeling

The insights engine groups conversations by topic, sentiment and effort score, enabling analysis across different functional areas (e.g., product, process, self-service).

Business value: Topic modeling reveals the true drivers of contact volume and pinpoints systemic customer frustration points that exist outside of agent control.

Multi-language support

Our engine also provides quality scoring and transcription for over 30 languages, ensuring consistency across multilingual operations.

Business value: Ensured equitable quality measurement and coaching for all agents, regardless of the language they serve.

The scope covers both voice and chat channels, providing a unified view of quality across the entire customer service footprint.

3. Redefining how contact center teams work together: The data-driven infinity loop

The shift to an AI-driven system for agent insights fundamentally redefined the QA and supervisory team roles at TELUS, moving them away from administrative tasks and toward high-value coaching and strategic input.

Our AI-powered system spurred this transformation through a repetitive data-driven process designed for continuous improvement:

1. Objective scoring (AI): The AI scores 100% of interactions against the custom rubric.
2. Targeted coaching (supervisors): Supervisors use the AI’s data to deliver personalized, objective coaching sessions focused on an agent’s specific, recurring gaps.
3. Feedback and improvement (QA/training): The aggregated data on top-failing rubric items (e.g. “75% of agents are missing the pricing disclosure”) directly re-informs the training curriculum and knowledge base, for both soft skills and hard skills. This crucial feedback is seamlessly fed into our AI-based training and simulation platform, Fuel iX™ Agent Trainer, allowing us to close the loop and ensure ongoing skill development.

Here’s how this data-driven infinity loop reshaped the responsibilities of TELUS’s QA team, contact center supervisors and agents.

Quality assurance (QA) team

Old way of working:

• 80% of time spent listening to and manually scoring calls

New way of working:

• 80% of time spent refining AI models, investigating systemic defects and performing root cause analysis on top-tagged issues identified by the AI

Supervisors

Old way of working:

• Focused on finding a few anecdotal calls to justify coaching
• Asking questions based on intuition and anecdotes

New way of working:

• Supervisors receive a daily dashboard of personalized coaching opportunities for each of their agents, highlighting areas for improvement with real-life examples from team members’ calls. Coaching is specific, outcome-focused and data-driven.
• They also leverage interaction data at scale to validate we’re asking the right questions and tracking behaviors that actually move key performance indicators (KPIs), such as sales, CX and repeats.

Agent experience

Old way of working:

• Coaching often perceived as subjective or punitive

New way of working:

• Coaching sessions are objective, focusing on specific skills and behaviors that the AI identified as missing across hundreds of their interactions (e.g., “Agent X needs coaching on empathy statements or the payment disclosure process”).

This shift in work allowed TELUS contact center supervisors to reallocate more than 30% of their administrative time back to direct, actionable coaching sessions.

Learn more about Agent Trainer here.

4. Realizing strategic value: Transforming the contact center into an insights engine

By analyzing 100% of customer conversations, Agent Quality Insights serves a purpose far beyond agent coaching. It turns the contact center into the organization’s most valuable source of voice of the customer (VoC) data, driving high-impact decisions across the entire business.

Product and upstream issue identification

Using keyword and volume analysis, Agent Quality Insights instantly flags unusual spikes in calls, mentioning things like “login failed” or “checkout error,” revealing that a recent product release was driving significant extraneous contact volume. This enables the product team to triage the issue within hours, preventing further widespread frustration.

Process optimization

Data analysis identifies key service bottlenecks, such as agents spending an inordinate amount of time navigating complex customer purchase histories. This insight drove the decision at TELUS to implement a GenAI-powered summary feature, projected to save minutes of average handle time (AHT) per interaction by eliminating unnecessary search time.

Self-service expansion

The insights engine can precisely quantify which intents are high-volume, repetitive and low-complexity. This data directly informed TELUS’s decision to prioritize pushing 60–80%+ of these calls to automated self-service channels, significantly reducing agent load and associated operational costs.

5. Measuring impact: Positive ROI to contact center ops, CX and compliance

The implementation of Agent Quality Insights at TELUS has driven transformative results across the organization, directly impacting risk, efficiency and customer experience (CX).

Coaching efficiency and operational savings

• 26% increase in coaching efficiency, allowing supervisors to manage more agents and deliver more impactful sessions

AI-powered coaching streamlines the two most time-consuming parts of the coaching cycle — preparation and documentation — by automatically pulling performance data into one place, generating recommendations and enabling joint SMART-plan creation directly in the tool.

CX improvement

• 20% reduction in spend on customer billing credits thanks to proactive service
• Improved retention, revenue protection and growth by giving leaders clear visibility into top-performing agent behaviors, surfacing missed save opportunities and enabling targeted coaching on discovery, articulating service value and tailoring offers

AI analyzes transcripts to identify the root causes driving calls and credits (e.g., self-serve failures, missed previous commitments, process gaps), allowing TELUS to address those causes proactively, improve the customer experience and reduce overall credit spending.

By analyzing contact center transcripts, TELUS measures the effectiveness of AI-powered retention offers targeted to at-risk customers.

Risk mitigation and compliance

• 16% reduction in fee waivers by reinforcing correct policies through targeted coaching

Using AI and data to review customer interactions, TELUS uncovered agents waiving fees or issuing credits outside of established guidelines.

By using AI to unlock the value of unstructured conversational data, Agent Quality Insights transformed TELUS’s QA function from a reactive, limited cost center into a powerful, objective, data-driven engine for compliance, risk management and agent skill development.

Ready to transform your agent coaching and contact center operations with AI?

If you’re ready to pull AI-powered insights directly from your own contact center data, reach out. We can help you understand what foundational capabilities you’ll need to support a data-driven, AI-powered agent experience driven by tools like Agent Quality Insights.

Learn more by exploring our AI-Powered Agent Tools.