Configure quality and coaching
The Quality and Coaching features help organizations consistently monitor, measure, and improve customer conversations using AI-powered evaluations and real-time coaching. With a centralized quality library and flexible evaluation plans, supervisors can spot issues early, coach representatives at the right moment, and ensure conversations meet standards.
Note
For Dynamics 365 Contact Center, Quality and Coaching evaluate conversations (chat, messaging, and voice interactions). Case evaluation applies to Dynamics 365 Customer Service only.
Your administrator configures and enables Quality and Coaching, including connection references and data sources. Once it's running, your role as a supervisor is to define what quality looks like in your contact center, shape which conversations get evaluated, and act on the results.
To access Quality and Coaching settings, go to Copilot Service admin center > Customer support > Quality and Coaching.
Build your quality library
The quality library is the foundation of your quality program. It contains the reusable quality indicators and guardrails that power your evaluation plans. Build the library first so your criteria are consistent and reusable across multiple plans.
Quality indicators
Quality indicators represent measurable dimensions of conversation quality—for example, empathy, greeting etiquette, resolution accuracy, or handling of sensitive information. Each indicator contains one or more questions that the AI uses to score conversations. Scored indicators contribute to an overall quality score, and you can reuse indicators across multiple evaluation plans.
To create a quality indicator, go to Quality and Coaching > Quality library > Quality indicators, then select Add new or choose an existing indicator to edit. You can also copy a built-in indicator to edit and reuse.
When building an indicator, add one or more questions and configure each with:
| Field | Description |
|---|---|
| Answer type | Choose Yes/No, Multiple choice, or Choose from list |
| Scores | Assign a score to each answer option |
| AI instructions | Optional guidance to improve evaluation accuracy |
Common quality indicator categories include:
| Category | What it measures |
|---|---|
| Greeting and closing adherence | Whether the representative followed the required opening and closing script |
| Resolution accuracy | Whether the customer's issue was resolved correctly |
| Empathy and tone | Whether the representative acknowledged the customer's frustration appropriately |
| Required disclosures | Whether mandatory statements (such as call recording notices) were delivered |
After saving, activate the indicator so it can be selected in evaluation plans.
Guardrails
Guardrails define the allowed and prohibited behaviors in a conversation—such as providing financial advice or handling vulnerable customers. The system evaluates guardrails for violations rather than numeric scores.
To create a guardrail, go to Quality library > Guardrails, then select Add new or choose an existing guardrail to edit.
Key fields when configuring a guardrail:
| Field | Description |
|---|---|
| Priority | Select Low, Medium, or High to indicate urgency when a violation is detected |
| Nudge text | Optional message surfaced to the representative when a violation occurs |
| Suggest next response | Enable this field to allow AI-generated guidance to be sent to the representative when a violation is detected |
After saving, activate the guardrail so it can be used in evaluation plans.
Create evaluation plans
An evaluation plan combines quality criteria from your library with conditions, frequency settings, and actions, defining what gets evaluated, when, and what happens with the results.
To create an evaluation plan, go to Quality and Coaching > Evaluation plans, then select Add new.
Key decisions when building a plan:
| Setting | Description |
|---|---|
| Quality criteria | Select one or more quality indicators and guardrails from the quality library to include in this plan. |
| Frequency | Choose when evaluations run: In real-time (as the conversation happens) or On conversation close. |
| Conditions | Scope evaluations to specific conversations by filtering on queue, workstream, channel, or other conversation attributes. Use conditions to focus real-time evaluation on high-value queues or workstreams. |
| Actions | For scored quality indicators, define Critical, Warning, and Normal score ranges. For each range, choose whether to notify a supervisor, send a coaching nudge to the representative, or both. Nudges appear as real-time messages during the conversation. |
When the configuration is complete, select Save, then activate the plan to begin evaluating conversations.
Tip
Build your quality library first so indicators and guardrails are reusable across multiple plans. Start with conservative actions and expand notifications or nudges based on operational results.
Review evaluations and act on insights
Evaluations appear in the Quality and Coaching workspace after each conversation is processed. Each evaluation includes:
| Component | What it contains |
|---|---|
| Interaction summary | An AI-generated summary of the conversation transcript. |
| Score | A rating based on your quality criteria, indicating whether the interaction met the threshold. |
| Recommendations | Suggested improvements tied to specific criteria that scored poorly. |
Conversations that fall below the threshold trigger the actions you defined in the plan, such as supervisor notifications or coaching nudges. Use flagged interactions as the starting point for coaching conversations with representatives.
The evaluation system's recommendations focus on process and behavior improvements, not employment decisions. Inform representatives that their conversations are evaluated, obtain any required consent, and align with your organization's local employment and privacy policies.
With a well-designed quality library and targeted evaluation plans, Quality and Coaching shifts your quality program from reactive sampling to consistent, systematic oversight—with built-in coaching at the moment it matters most.