AB-250 Transforming Contact Center Experiences with AI in Dynamics 365 Questions and Answers

Programmatic voice interaction requires a calling control surface rather than a configuration feature intended for chat, knowledge, or information governance. Dynamics 365 Contact Center's voice architecture is built on communication capabilities that expose calling APIs and SDK functionality for establishing, controlling, transferring, and extending calls. These APIs support scenarios where application logic must influence the active voice interaction rather than relying entirely on manually initiated representative actions.

Microsoft's current communications architecture exposes Calling SDK and Call Automation capabilities for programmable calling scenarios. Dynamics 365 Contact Center also supports consultation and transfer between service representatives and supervisors as part of the voice channel, which is the operational pattern required when the frontline interaction must involve another employee or back-office specialist.

A pre-chat survey applies to digital chat data collection. Knowledge-authoring functions provide resolution content but cannot manipulate a live voice call. Sensitivity labels provide information-protection classification and likewise have no voice-control function.

Therefore, the calling/voice APIs are the only option that provides the required programmable interaction control.

Study Guide reference: Implement channels - > Provision and manage voice channels - > Extend voice interaction and calling capabilities by using supported APIs/SDKs.

The automated voice interaction must be defined before downstream human-routing behavior can be finalized because the voice agent determines what customer input is collected, which requests can be resolved through self-service, and when escalation to a representative is required. The Fabrikam technical requirements explicitly require Copilot Studio voice agents, both spoken and keypad input, and the ability to carry collected information into the subsequent representative interaction.

Microsoft's Copilot Studio voice architecture supports voice-enabled agents that process speech and DTMF input, use NLU or generative orchestration, collect context, and transfer unresolved conversations to Contact Center. The voice agent is therefore part of the caller journey before the human-assignment stage.

After the interaction behavior and escalation conditions are understood, routing configuration can use the resulting context, intent, skills, and other classification information to determine the correct queue and representative. Configuring routing first would force routing logic to be created without knowing precisely which interaction outputs and escalation conditions the agent will generate.

Queue membership and representative assignment are later-stage work-distribution concerns and do not define the automated IVR experience.

Study Guide reference: Configure agents and AI capabilities - > Configure Copilot Studio voice agents, IVR interactions, context, and escalation.

Structured information collection in Copilot Studio begins by defining what type of information must be captured. Entities provide the semantic type for values such as numbers, dates, account identifiers, locations, or custom business concepts. Once the expected entity is known, a Question node prompts the customer and associates the response with the corresponding variable.

The captured value can then be reused later in the topic or made available more broadly through appropriately scoped variables. This is critical in the Fabrikam scenario because information collected by the voice agent must remain available for downstream processing and for human escalation. Microsoft includes variable management as an explicit AB-250 voice-agent objective and supports using topic/global variables to retain and reuse customer input during the conversational journey.

Creating variables before any value exists does not establish the required structured capture logic. Publishing before defining entities and questions reverses the correct development sequence. Disabling validation and relying entirely on free-text parsing also weakens reliability because the agent loses the structured entity recognition needed for predictable automation.

The technically sound sequence is therefore entity definition, Question-node capture, and persistence of the captured result for subsequent use.

Study Guide reference: Configure agents and AI capabilities - > Configure voice agents - > Manage variables in Copilot Studio.

For a custom portal, simply configuring Microsoft Entra ID as the portal identity provider does not by itself complete Dynamics 365 authenticated-chat integration. Contact Center must receive and validate an authentication token as part of the chat initialization flow.

Microsoft's documented custom-portal architecture requires a public/private key pair, a public-key endpoint, a server-side service that creates an RS256-signed JWT, and a JavaScript token-provider function. The authentication settings record in Contact Center references the public-key URL and client function, and the chat widget receives the signed JWT when the authenticated conversation starts.

Microsoft Entra ID can certainly participate in authenticating the portal user, but the scenario asks whether configuring Entra ID alone satisfies the Contact Center chat-authentication requirement. It does not. The authenticated identity must still be translated into the token flow understood and validated by the Dynamics 365 live-chat service.

This differs from Power Pages scenarios, where portions of the authentication integration are provided by the portal infrastructure. The question explicitly states that the organization operates a custom portal, making the token-based custom implementation relevant.

The supplied source also identifies B.

Study Guide reference: Implement channels - > Configure custom-portal authenticated chat and authentication-token integration.

Assign work only during scheduled working hours - > Add shift bookings for CSRs.

Prevent work assignment during breaks within a shift - > Set activity types to control availability during a shift.

Shift-based routing integrates representative schedules with unified routing so work is assigned only when the representative is actually scheduled to perform assignable work. Microsoft requires shift bookings to exist in Dynamics 365 before the routing engine can evaluate a representative's scheduled availability. The booking defines when that representative is expected to be working, allowing the assignment engine to exclude resources outside scheduled hours.

Within the shift itself, activity types distinguish assignable working periods from non-assignable activities. Microsoft explicitly notes that when a shift booking contains an activity type classified as non-assignable—for example, a lunch break or training period—the representative is not considered eligible for assignment during that interval. This provides the required control for breaks without modifying the overall shift or routing rules.

Channel-priority rules affect prioritization of work, not calendar eligibility. Staffing rules are used primarily for workforce forecasting and planning rather than determining whether an individual representative can receive a work item at a particular moment.

The answer area in the source confirms the combination of shift bookings and activity types.

Study Guide reference: Configure work distribution - > Configure shift-based routing and representative availability.

The requirement is to eliminate inconsistent access across the customer service representative, supervisor, and administrator personas while enforcing least privilege. Dataverse security roles are the correct control because they define the privileges users receive on tables, records, applications, and Contact Center capabilities. Dynamics 365 Contact Center supports administrator, supervisor, and representative personas, and Microsoft permits custom security roles to be mapped to these personas when the predefined roles do not precisely match organizational responsibilities.

The Fabrikam requirements explicitly state that access must be restricted according to user responsibilities, CSRs should access only records associated with the interactions they handle, supervisors require performance and scheduling visibility, and administrators require configuration access. Creating appropriately scoped Dataverse roles for each defined persona addresses these requirements directly.

Access teams primarily provide record-sharing semantics and do not establish the complete functional privilege model required here. Field-level security protects specific secured columns rather than defining an entire persona. Queue membership controls routing eligibility rather than application authorization.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Manage users - > Manage security roles, including custom security roles and personas.

Generative answers grounded in approved organizational knowledge allow a Copilot Studio agent to respond naturally to customer questions without requiring every possible utterance or FAQ variation to be manually represented as a deterministic topic. The agent retrieves relevant knowledge and uses that grounded material to construct a conversational response to the customer's request.

Microsoft's Contact Center architecture explicitly supports AI agents for autonomous self-service, and Copilot Studio generative orchestration can dynamically select knowledge, topics, or actions according to the customer's natural-language intent. This makes the approach substantially more flexible than a conventional fixed-menu IVR while still allowing the knowledge sources to constrain and ground generated responses.

The Fabrikam requirements require AI-driven handling of frequently asked questions and explicitly call for generative, natural conversational interactions. Therefore, a grounded Copilot Studio implementation directly addresses both requirements.

A fixed IVR menu may be appropriate where deterministic routing is required, but it cannot satisfy the generative requirement. An emailed FAQ is asynchronous and does not participate in the live interaction. A greeting message provides no question-answering capability.

Study Guide reference: Configure agents and AI capabilities - > Configure voice agents by using generative AI or classic orchestration; configure knowledge-grounded AI interactions.

Caller interacts from a noisy environment - > Numeric menu selection

Caller interacts from a quiet environment - > Speech transcription

For callers in noisy environments, DTMF-based numeric menu selection provides a deterministic input mechanism that is largely unaffected by background sound. Copilot Studio supports DTMF input specifically so callers can use the phone keypad instead of relying on spoken responses. Microsoft notes that DTMF can be configured on Question nodes and is well suited to deterministic IVR menu flows.

In a quiet environment, speech transcription provides a more natural interaction. Basic voice architecture processes the caller's speech through speech-to-text before the resulting text is evaluated by Copilot Studio's NLU or dialog logic. This supports natural spoken input while preserving the caller's intent and captured information for the automated interaction.

Entity extraction and intent classification are processing techniques applied after input is captured; they are not alternative physical input methods. Entity extraction identifies values such as account numbers or names, while intent classification determines what the caller is trying to accomplish.

The scenario also states that Fabrikam must support both naturally spoken input and keypad input, making this mapping directly consistent with its technical requirement.

Study Guide reference: Configure agents and AI capabilities - > Configure voice-enabled Copilot agents - > Configure speech and DTMF interaction behavior.

Standardize permissions for users in the same job function - > Add the user to a team that has the mapped role.

Enable persona-based access for the specific user - > Map each custom security role to a persona.

Security roles and personas solve related but different access-control requirements. When multiple users perform the same job function, assigning the relevant role through a common team provides a consistent permission baseline. Dataverse team membership allows members to inherit the privileges associated with the team’s security role, which simplifies centralized administration and helps prevent drift between users performing equivalent duties.

Persona mapping is used when a custom role must participate in Dynamics 365 Contact Center’s administrator, supervisor, or representative behavioral model. Microsoft defines these personas as archetypes and allows administrators to map custom security roles to the appropriate persona in Copilot Service admin center. Mapping the custom role ensures Contact Center recognizes the role in the correct functional context.

A critical implementation detail is that Microsoft’s current guidance requires roles used for role-to-persona mapping to be assigned directly to the relevant user rather than inherited only through team membership. Therefore, the persona mapping establishes the role/persona relationship, while the mapped role must also be assigned appropriately to the individual.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Manage security roles, including custom security roles and personas.

Reduce permissions to what is required for each CSR role - > Create a custom security role with required privileges only.

Allow non-administrators to assign roles to CSRs without full system access - > Create a custom role for delegated role assignment.

Dataverse security should be implemented according to least privilege. A CSR should receive only the table privileges, privilege depths, application permissions, and Contact Center-specific capabilities necessary for the assigned duties. Therefore, creating a custom security role with only the required privileges directly satisfies the requirement to reduce the excessive permissions inherited from broader default roles. Microsoft specifically supports creating and tailoring custom roles when predefined roles expose functionality beyond the intended job function.

Delegated role administration is a separate security problem. Microsoft documents that a non-System Administrator can assign security roles when a purpose-built custom administrative role grants the necessary privileges on the Security Role table. This provides controlled delegation without granting the broad System Administrator role. Microsoft also warns against simply copying System Administrator for delegated administration because that can create excessive privilege and elevation risks. The resulting design separates CSR operational access from delegated security administration and therefore conforms to least-privilege governance.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Manage users - > Manage security roles, including custom security roles and personas.

A quality-management and compliance process requires durable evidence of completed customer interactions. Voice call recording provides the actual audio record, allowing authorized reviewers to replay conversations for quality assessment, investigation, coaching, and compliance purposes. Microsoft supports administrators enabling call recording together with voice transcription and provides controls for accessing the resulting recordings and transcripts after conversations end.

Transcription creates a searchable textual representation of the voice interaction. When combined with appropriate retention controls, transcripts can be preserved for the organization's specified compliance period. Microsoft documents retention management for Contact Center transcripts and warns that recordings are subject to applicable consent and communications-recording laws.

Live sentiment analysis helps supervisors understand customer sentiment during or after an interaction but does not itself provide the retained conversation record. Real-time translation addresses language accessibility, while post-call surveys gather customer feedback. Neither substitutes for retaining the underlying communication.

Therefore, transcription with retention and voice call recording together provide the evidentiary artifacts needed for systematic quality review and compliant recordkeeping.

Study Guide reference: Implement channels - > Provision and manage the voice channel; configure recording and transcription capabilities.

Shift-based routing evaluates the activity type associated with a representative's shift booking to determine whether the representative is eligible for automatic assignment. Microsoft explicitly states that if the shift activity is configured as nonassignable, such as a lunch break or training activity, the representative is not considered for assignment during that interval.

The implementation therefore requires the organization to define appropriate shift activity types and mark the break activity so that it represents an unavailable/nonassignable period. When that break booking becomes active, unified routing excludes the CSR from assignment until the unavailable activity ends. This provides schedule-aware routing at the individual representative level and avoids unnecessarily altering queue behavior for other representatives who remain available.

Capacity profiles address how much concurrent workload a CSR can handle; they do not identify a scheduled break. Queue prioritization determines the order in which waiting work is processed. Staffing targets are planning and forecasting inputs. None of these establishes the break itself as a nonassignable period.

Microsoft's current shift-based routing documentation specifically uses lunch break as an example of a nonassignable shift activity, making activity-type configuration and unavailable status the technically appropriate combination.

Study Guide reference: Configure work distribution / Workforce Management - > Configure shift management and scheduling; configure shift-based routing.

Proactive chat is designed to initiate customer engagement before a visitor manually opens the chat widget. The first required capability is enabling Proactive chat on the relevant chat channel/widget. Microsoft exposes this setting in the chat-channel configuration and requires it to be turned on before proactive invitations can be presented.

The second component is the proactive trigger logic. Dynamics 365 can evaluate visitor behavior such as time spent on a page, page URL, journey information, and other configured conditions. When those conditions are satisfied, the system automatically displays a personalized invitation to start a chat. This directly addresses the scenario in which customers remain on product pages but otherwise leave without initiating a conversation.

Agent scripts are productivity guidance for representatives after an interaction has reached the service representative; they do not trigger a customer-facing invitation. Pre-chat surveys collect information after the customer elects to begin the interaction and therefore cannot initiate engagement with an inactive visitor.

The uploaded question records A and C as the correct actions.

Study Guide reference: Implement channels - > Configure chat - > Configure proactive chat and trigger conditions.

Unified routing uses queues to create controlled pools of representatives and route-to-queue rules to decide which pool should receive each incoming work item. Microsoft specifically notes that separate queues can be created for distinct lines of business or support functions and that queue membership determines the representatives available to receive work from that queue.

The appropriate design is therefore to create a queue for Service team 1 and another for Service team 2. Routing rules should direct general inquiries to team 1 and complex product issues to team 2. Unidentified work can either be explicitly routed to Service team 1 or handled through the workstream's fallback routing design where appropriate. Microsoft states that work for which no route-to-queue rule matches is routed to the workstream's fallback queue.

A shared queue would remove the required workload separation. Capacity profiles control concurrent workload but cannot distinguish inquiry categories. Session templates control the representative interface after assignment and have no role in determining which operational team receives a work item.

Study Guide reference: Configure work distribution - > Configure queues, queue membership, route-to-queue rules, and fallback routing.

Queue operating hours define the periods during which a queue is considered available to process work. Microsoft describes operating hours as the schedule during which the customer-support organization is active and explicitly supports associating operating-hours records with queues.

This is the correct level of control for the scenario because the requirement concerns whether the queue itself should accept and process work during a defined service period. By associating appropriate operating hours with each queue, the organization can ensure that routing behavior respects the actual staffed service window rather than simply allowing requests to accumulate until representatives eventually become available.

CSR working hours describe an individual's schedule and are useful for representative eligibility but do not establish the operating window of the entire queue. A traditional Customer Service calendar is often associated with SLA calculations and other service scheduling scenarios rather than the omnichannel queue-processing requirement described here. Channel operating hours may control availability of a channel such as a chat widget, but the scenario specifically states that several queues must process work only during defined periods.

Study Guide reference: Configure work distribution - > Configure queues - > Set up operating hours.

Copilot Studio Question nodes are designed to collect information from users and automatically store the resulting value in a variable. Microsoft explicitly states that the response captured by a Question node can be reused later in the same conversation. The variable can represent structured entity data, multiple-choice input, free-form responses, or values obtained through voice and supported DTMF interaction.

Topic variables are scoped to the topic in which they are created but can also be passed between topics when the interaction design requires reuse across a broader conversation path. For values that must be available throughout the complete session, the variable can additionally be promoted to global scope. Microsoft documents that global variables remain available across topics during the same user session.

A greeting message only outputs information and does not capture a reusable value. A scheduled report operates after the interaction and therefore cannot drive real-time conversational logic. A knowledge article is a static knowledge resource and cannot represent an individual caller's dynamically supplied value.

Study Guide reference: Configure agents and AI capabilities - > Configure Copilot Studio agents - > Configure questions, entities, variables, and conversational context.

1. Export the solution as managed.

2. Export configuration data separately.

3. Import the managed solution into the production environment.

Fabrikam requires governed, repeatable deployment, managed solutions in production, and supported migration of configuration data without manually rebuilding the environment. The solution therefore separates deployable solution components from configuration records that require configuration-data migration.

Microsoft recommends developing configuration in an unmanaged solution in the source environment, exporting that solution as managed, and importing the managed package into downstream environments. Direct editing of managed configuration in production is specifically discouraged because it introduces an active unmanaged layer that can override later managed updates and interfere with ALM servicing.

Contact Center also contains configuration records that may need migration separately. Microsoft documents the use of solutions together with the Configuration Migration tool for workstreams, queues, skills, capacity profiles, routing configuration, and channel-related records. Therefore, the deployment package is exported as managed, configuration data is exported through the supported migration process, and the managed solution is introduced into production rather than modifying production manually.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Create solutions; Configure ALM for Dynamics 365 Contact Center deployments.

Configure campaign contacts - > Connect to Customer Insights.

Configure behavior of the proactive dialer - > Configure journey-based engagement schedule.

Dynamics 365 Contact Center proactive engagement can integrate with Dynamics 365 Customer Insights - Journeys when customer outreach is driven through conversational journeys. Microsoft documents Customer Insights as a supported proactive-engagement intake/orchestration path and requires Customer Insights when journey-based customer outreach is used.

The customer audience or campaign-contact definition can therefore originate from Customer Insights, where segmentation, triggers, and customer-journey logic determine who should receive the outreach. This matches the case requirement to use customer data for personalized engagement without integrating the existing third-party marketing platform.

The dialer's operational behavior can then be coordinated with the journey-based engagement schedule, defining when the proactive outreach occurs according to the configured customer journey. Customer Insights triggers and journey configuration can initiate voice or messaging engagements when defined customer events or segment criteria are satisfied.

An outbound calling profile primarily governs representative-initiated calling, caller identity, and related voice settings; it does not define the campaign audience. SLA evaluation rules are unrelated to proactive dialing behavior. A generic Dataverse marketing list would also bypass the scenario's stated Customer Insights - Journeys direction.

Study Guide reference: Implement channels - > Configure proactive engagement; integrate Contact Center with Customer Insights - Journeys; configure outbound audience and engagement scheduling.

Update customers about high-latency operations - > Send Message Delay

Optimize audio clarity - > Sensitivity

Send Message Delay controls when a voice agent presents a latency message during a long-running operation. Microsoft defines this setting as the period the agent waits after a background operation begins before communicating the configured latency message. It is therefore the appropriate control when customers must be kept informed while an operation takes longer than expected.

Sensitivity governs how aggressively the voice agent distinguishes spoken input from environmental noise. Microsoft specifically documents that lower sensitivity is appropriate for noisy environments, public spaces, or hands-free scenarios, while higher sensitivity is more suitable for quiet environments and soft-spoken callers. Tuning sensitivity therefore directly affects practical speech clarity and recognition reliability.

Speech detection timeout and speech recognition timeout address different timing concerns. They determine how long the voice agent waits for speech to begin or how long it allows input after the caller starts speaking; they do not control latency notifications or the balance between speech and background noise.

Study Guide reference: Configure agents and AI capabilities - > Configure voice agents - > Configure voice optimization settings, latency messaging, and speech recognition behavior.

Streaming speech recognition and response generation during live calls - > Real-time Speech agent

Intent-based journey orchestration - > NLU in Copilot Studio

A real-time voice agent is designed for low-latency conversational processing in which spoken input and generated responses are handled continuously during the live interaction. Microsoft distinguishes real-time voice agents from basic voice agents: real-time agents use a speech-to-speech real-time model for natural, low-latency conversations, while basic voice agents typically convert speech to text, process the resulting text, and then synthesize speech.

Consequently, Real-time Speech agent is the appropriate component for streaming speech recognition and response generation during a live call. It addresses the requirement for fluid, interactive speech rather than merely supplying a standalone transcription service.

For intent-driven orchestration, Copilot Studio's Natural Language Understanding capability interprets customer language and determines which predefined topic or conversational path should execute. Microsoft documents NLU, NLU+, and classic orchestration as mechanisms for deterministic intent-driven dialog behavior, while generative orchestration dynamically chooses topics, actions, and knowledge.

Voice routing configuration operates at the Contact Center work-distribution layer rather than performing conversational intent understanding. Azure AI Speech provides speech services but does not itself represent the Copilot Studio orchestration component required here.

Study Guide reference: Configure agents and AI capabilities - > Configure voice agents, real-time voice, NLU, and IVR orchestration.

1. Initialize Channel Integration Framework context.

2. Run the API to get conversationid details.

3. Run an API call to send messages by using conversationid as a parameter.

The integration first needs an initialized Channel Integration Framework context because VanArsdel is operating an external CCaaS provider through the Dynamics 365 representative desktop. The active interaction must then be identified so the integration can target the correct conversation rather than relying on an unrelated session identifier.

Dynamics 365 exposes an Omnichannel JavaScript API containing getConversationId, which retrieves the unique identifier of the current conversation, and sendMessageToConversation, which sends a programmatic message to a conversation. Microsoft documents conversationId as the identifier used to target a particular conversation; when supplied, sendMessageToConversation directs the message to that interaction.

This distinction is important because a representative can operate multiple sessions. Conversation identity is the reliable interaction-level identifier required by the API. Supplying both sessionid and conversationid is not the documented signature for sendMessageToConversation. Therefore, after the framework context is established, retrieve the conversation identifier and use that identifier when issuing the message operation.

Study Guide reference: Configure the Dynamics 365 Contact Center representative experience - > Extend the representative experience by using APIs and productivity capabilities.

1. Register a Channel Integration Framework provider and host it as a web resource.

2. Use the Channel Integration Framework v2 to capture the interaction.

3. Call the Channel Integration Framework API to open an application tab.

VanArsdel is deliberately adopting the representative desktop before completing migration from its existing CCaaS platform. The technical requirements therefore call for Dynamics 365 Channel Integration Framework so the external communication provider can operate inside the Dynamics 365 model-driven representative experience.

The provider must first be registered so Dynamics 365 knows which channel provider implementation to load. The hosted web resource then functions as the provider application running inside the Channel Integration Framework environment. Channel Integration Framework 2.0 is appropriate because Microsoft designed it for integration of telephony channel providers with Dynamics 365 model-driven apps and exposes browser-based JavaScript APIs for communication-provider interactions.

After the interaction is available through the framework, Channel Integration Framework APIs can control the Dynamics workspace, including opening application content associated with the interaction. This allows the CSR to receive the interaction and expose corresponding CRM customer information without abandoning the standardized representative desktop.

Version 1 is therefore not appropriate for the required modern multisession Contact Center integration.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Configure and extend connectors for Dynamics 365 Contact Center; configure integration architecture.

The case requires supervisors to have both immediate operational visibility and retrospective performance insight. Real-time dashboards satisfy the first requirement by exposing current Contact Center conditions such as active interactions, queue backlog, representative availability, service levels, abandonment, and other time-sensitive KPIs. Microsoft describes real-time analytics as providing supervisors with continuously changing operational metrics so they can make immediate decisions during the working day.

Historical analytics provides the complementary long-term perspective. It allows supervisors and service managers to analyze previous conversation volumes, representative performance, handling patterns, channel trends, and other historical metrics. Microsoft distinguishes historical analytics as the appropriate capability for identifying patterns and supporting strategic improvement over time.

SLA metrics may be useful within service operations but do not provide the complete monitoring architecture required by the case. Supervisor actions such as monitor or transfer enable intervention in individual live conversations but do not supply the historical trend analysis explicitly required.

Together, real-time dashboards and historical analytics provide the operational and analytical coverage specified in the case study.

Study Guide reference: Manage analytics for Dynamics 365 Contact Center - > Configure real-time analytics, historical analytics, and supervisor monitoring capabilities.

Prevent Payment Card numbers from appearing in AI-generated transcripts and conversations - > Create and manage data masking rules.

Retain customer interactions in accordance with company policy - > Classify and set retention policies by using Microsoft Purview.

The two requirements address different stages of information governance. Sensitive information must first be protected within the interaction itself. Dynamics 365 Contact Center data-masking rules detect configured patterns such as credit-card numbers and replace the protected data with masked characters. Microsoft documents that text masked in a conversation message is also masked in the corresponding conversation transcript. Contact Center includes predefined masking patterns for credit cards and allows administrators to define additional rules.

The second requirement concerns lifecycle governance rather than runtime visibility. Retention policy determines how long customer interaction records must remain available according to organizational and regulatory policy. In the case study, VanArsdel specifically handles PII and payment information and must prevent sensitive information from being retained in transcripts and AI-generated summaries while maintaining compliant records.

Field-level security protects Dataverse columns but does not redact sensitive values appearing inside conversation text. Sensitive-data redaction is related, but the supplied requirement specifically maps payment-card protection to configurable masking rules. Retention and classification are therefore handled independently through Microsoft Purview governance controls.

Study Guide reference: Deploy Dynamics 365 Contact Center - > Configure data protection, security, governance, and Contact Center environments.