Edit

Configure conversation orchestration by using natural language playbooks (preview)

[This article is prerelease documentation and is subject to change.]

Conversation orchestration keeps every conversation actively managed throughout the conversation lifecycle from initiation through resolution. Conversation orchestration monitors each conversation as conditions evolve, such as an increase in wait time, queues filling up, or service representatives going offline, and automatically runs the relevant action in accordance with your configured playbooks. Administrators define the logic by using natural-language playbooks, and conversation orchestration handles the execution. When you publish a playbook, the system converts your natural language playbooks into a structured runtime format.

In the preview release, conversation orchestration is available for voice and live chat channels only.

Important

Prerequisites

  • Specific licensing requirements apply to configure and use conversation orchestration in Dynamics 365 Contact Center. Learn more in Dynamics 365 Licensing Guide.
  • System administrator or Omnichannel administrator role.
  • The Power Platform Pay-as-you-go plan mandates the use of an Azure subscription that the system charges when the agent runs. Make sure that you set up consumption-based billing.
  • Unified routing enabled for your environment.
  • At least one queue and workstream is configured for voice or messaging channels.
  • A voice or messaging channel is configured.

Supported scenarios

The supported scenarios are as follows:

  • Dynamic prioritization:

    • Update priority based on transfer to queue
    • Escalate priority based on conversation wait time

  • Overflow handling:

    • Overflow based on support representative availability in the queue, sign in status, and queue operating hours.

Understand playbooks

Configure playbooks per scenario through guided templates that keep instructions focused and reliable. A playbook consists of a trigger event and a set of conditions and actions to run when the trigger event occurs.

  • Queues: The queues for which the playbook applies.
  • Trigger event: The event that initiates the playbook, such as a conversation waiting in queue or conversation transferred.
  • Channel: The channel that you select in the Queues dialog. The context variables that appear are based on the selected channel. If you update the channel in a draft playbook, then you must update the applicable context variables and the corresponding prompt.
  • Conditions: Business rules based on context variables, such as customer tier or country/region. You can define up to 10 conditions to configure the granularity. These conditions can have up to two context variables.
  • Actions: The outcomes when conditions are met, such as increasing priority or transferring to an overflow queue.

A playbook can have one of the following statuses.

Status Description
Draft The playbook is saved but not active. You can make changes freely.
Active The playbook is published and actively orchestrates conversations.

Manage playbooks

In Copilot Service admin center, go to Customer support > Conversation Orchestration (Preview). The Conversation orchestration (Preview) > New page displays available playbook templates organized by scenario.

Important

This feature is intended to help customer service managers or supervisors enhance their team’s performance and improve customer satisfaction. This feature is not intended for use in making—and should not be used to make—decisions that affect the employment of an employee or group of employees, including compensation, rewards, seniority, or other rights or entitlements. Customers are solely responsible for using Dynamics 365, this feature, and any associated feature or service in compliance with all applicable laws, including laws relating to accessing individual employee analytics and monitoring, recording, and storing communications with end users. This also includes adequately notifying end users that their communications with representatives may be monitored, recorded, or stored and, as required by applicable laws, obtaining consent from end users before using the feature with them. Customers are also encouraged to have a mechanism in place to inform their representatives that their communications with end users may be monitored, recorded, or stored.

Create a playbook

By default, the conversation orchestration page displays three popular prompts. Use Prompt gallery to view prompts for the different scenarios.

  1. On the Conversation Orchestration page, select Prompt gallery. The pop-up window displays prompt templates for all scenarios by default.

  2. In the drop-down field, select one of the following categories and then a prompt template that matches your scenario:

    • Dynamic prioritization:
      • Escalate priority based on wait time: For wait time-based dynamic or real-time prioritization.
      • Escalate priority based on transfer to queue: For transfer-based prioritization.
    • Overflow handling:
      • Configure overflow when queue goes out of operating hours: For out of operating hours overflow.
      • Configure overflow when all users are logged out: For all representatives in the queue are signed out overflow.
      • Configure overflow based on support representative availability in the queue: For representative availability overflow.

  3. In the playbook editor, accept the default name or enter a name in Playbook name.

  4. Select Queues > Edit to open the queues pane.

    1. Channel: Select Voice or Messaging.
    2. In Apply to, select how to apply the playbook.
      • All queues: Applies to all queues for the selected channel.
      • List of queues: Select specific queues from the list.
      • All queues except: Applies to all queues except the ones you exclude.
    3. Select Save.

  5. Optionally, add context variables to create conditions based on customer or conversation attributes:

    1. In the Add variables section, select Add customer attribute.
    2. Choose from available context variables such as CustomerTier, Is VIP Customer, Preferred Language, Customer Region, or Account Type.
    3. Provide a description for how the variable appears in the playbook.
    4. Select the values you want to use in your conditions.

  6. Configure conditions and actions based on your selected template. See Tips for this playbook on the edit playbook page to optimally configure the conditions.

  7. Select Save. The playbook is saved as a draft.

  8. Review your configuration for any validation warnings, and then select Publish to activate the playbook.

Note

Administrators configure natural language playbooks. The system uses a language model to convert them to a structured runtime format. The system uses the resulting runtime format to perform playbook actions. Because the system uses a language model to convert playbooks into a structured runtime format, the converted output might not fully capture the intended playbook logic. Any deviations can be identified only during runtime. Administrators should test and monitor runtime behavior and revise the playbook if orchestration doesn't work as intended.

Edit a playbook

  1. Go to the All playbooks tab in Conversation orchestration.
  2. Find the playbook you want to edit.
  3. Select the vertical ellipsis, and then select Edit.
  4. Make your changes.
  5. Do one of the following:
    • For draft playbooks: Select Save to save your changes, or select Publish when ready to activate.
    • For active playbooks: Select Save & publish to save and immediately publish your changes. The Save button is disabled for active playbooks to prevent unintended changes to live routing. If you cancel, the playbook reverts to the last published version.

Delete a playbook

  1. Go to the Playbook tab.
  2. Select the actions menu (...) for the playbook and then select Delete.
  3. Confirm the deletion.

Important

If you delete an active playbook, conversation orchestration stops for the corresponding queue immediately.

Validations

Conversation orchestration validates your playbook configuration and displays warnings for potential issues.

Validation Description
Required fields Mandatory fields are empty (priority score, time interval, action)
Range validation Values are outside allowed ranges (priority: 0–10,000; time: minimum 30 seconds)
Duplicate conditions Two conditions have identical variable-value combinations
Conflicting conditions Overlapping conditions might cause unexpected behavior
Cross-playbook conflicts Multiple active playbooks exist for the same queue and scenario

Multiple playbooks for the same queue

When you configure multiple playbooks that apply to the same queue, conversation orchestration handles them differently based on whether they target the same or different scenarios.

Same queue, same scenario (conflict)

You can't have multiple active playbooks for the same scenario and the same queue. This configuration creates a conflict that prevents publishing.

For example, if you have an active "Escalate priority based on wait time" playbook for Queue A, you can't publish another wait time-based priority playbook for Queue A - even if the time interval or priority values are different. Instead, edit the existing playbook to update its configuration.

When you attempt to publish a playbook that conflicts with an existing active playbook, the system displays an error message similar to the following one:

"Failed to publish policy: A playbook for "Escalate priority based on wait time" is already published for queue_name. You cannot publish a duplicate policy for the same queue(s). Please modify the existing published policy."

To resolve this conflict, do one of the following steps:

  • Modify the queue selection in your new playbook to exclude the conflicting queues
  • Edit or deactivate the existing active playbook before publishing your new one

You can save a playbook as a draft even if it conflicts with an existing active playbook.

Important

The system performs the cross-playbook validation only when you publish a playbook, not when you save it as a draft. The system checks your playbook against other published active playbooks only.

Same queue, different scenarios (Allowed)

You can configure different scenarios for the same queue. For example, a single queue can have both a "Dynamic Prioritization - Wait Time" playbook and an "Agent Availability Based Overflow" playbook active simultaneously.

When multiple playbooks with different scenarios are active for the same queue, they run based on the triggering event.

Example: How multiple playbooks for the same queue across different scenarios work

Consider a queue configured with two playbooks, each designed to handle a distinct scenario:

  • Dynamic prioritization playbook

    • Trigger: Wait time
    • Condition: Customer tier = Gold and wait time ≥ 30 seconds
    • Action: Increase priority by 1

  • Agent availability overflow playbook

    • Trigger: Immediate evaluation on queue entry
    • Condition: Customer tier = Gold and no agents are available
    • Action: Run overflow (for example, route to voicemail, transfer to another queue, or redirect to another number)

Although both playbooks target Gold-tier customers, they differ in their triggering condition—one is based on elapsed wait time, and the other on real-time agent availability.

How playbooks are run

Scenario 1: Agents are available initially

  1. A conversation enters the queue. Agents are available. The overflow condition isn't met, so no overflow action runs.
  2. The conversation remains unassigned for 30 seconds.
  3. The dynamic prioritization condition is now met, triggering a priority increase as defined.

Result: The system runs the dynamic prioritization playbook only.

Scenario 2: No agents available when work item enters the queue

  1. A conversation enters the queue. No agents are available. The overflow condition is met immediately. The overflow action is triggered (for example, redirect to voicemail or another queue).
  2. If the conversation:
    • Leaves the queue or is closed: No further actions occur.
    • Remains in the queue and is still waiting: On reaching 30 seconds, the action for dynamic prioritization playbook runs and increases the priority.
    • Overflow action runs first based on availability.
    • Dynamic prioritization runs only if the conversation remains in the queue long enough.

Things to consider

Issue Resolution
Playbook isn't routing conversations Verify the playbook status is Active, check that the correct queues are selected, and ensure the channel matches your conversation type.
Priority hasn't increased as expected Confirm the time interval setting, verify context variable values match your conditions, and check that the conversation is in a queue where the playbook is active. Also verify that the queue doesn't have custom prioritization rules configured.
Overflow isn't triggered Verify agent availability settings (presence, capacity, skills), check that the "no agents available" condition is being met, and ensure the playbook is active for the correct queue.
Dynamic prioritization isn't being applied to a queue The queue might have custom prioritization rules configured. Remove the custom prioritization configuration from the queue settings to enable dynamic prioritization.

View diagnostics using a custom query

To help you monitor and troubleshoot your playbooks, conversation orchestration provides diagnostics information for each conversation that is processed by an active playbook. You can view the diagnostics information by using a custom query. Learn more in Sample queries and dashboards.

Set up unified routing
Create and manage queues
Configure context variables
Dynamic prioritization scenarios
Overflow scenario
Conversation orchestration: Responsible AI FAQ

  • Last updated on