Omnichannel Contact Center for a Retail Bank
Contoso Bank's retail contact center handles 2.1 million contacts a year with 450 agents in two sites. Customers reach the bank by phone, secure email, web chat, SMS, Facebook Messenger, and WhatsApp. Each channel has its own queue, and supervisors cannot see workload across them. Average handle time is 9 minutes 20 seconds; 38% of contacts escalate to a second agent because the first agent cannot find the right policy or product information. Knowledge lives in 14 SharePoint sites and three PDF libraries. CSAT is 72, and the bank's regulator requires a full interaction history for complaints. The bank already uses Dynamics 365 Sales for mortgage advisers but has no service platform. It is evaluating Dynamics 365 Customer Service and asks what an omnichannel deployment would add, whether Copilot can reduce handle time, and what should remain in core banking systems.
1. Explain how Dynamics 365 Customer Service and its omnichannel capabilities would work together, and which capabilities address the bank's specific symptoms.
Read the worked answer
Customer Service is the case platform. It stores the customer, the issue, the interaction, entitlements, queues, and SLAs. A complaint becomes a case with a timeline, so the regulator's requirement for a full interaction history is met by the case record rather than by stitching together channel logs.
Omnichannel adds the unified work distribution layer. It brings voice, chat, SMS, social, and messaging into one agent desktop and one supervisor view. The six separate queues become queues defined by skill and priority, not by channel. Supervisors get live workload and agent presence across channels. This directly addresses the symptom of invisible cross-channel workload.
Knowledge management addresses the 38% escalation rate. A searchable knowledge base with versioning, feedback, and article linking to cases gives agents one source instead of 14 SharePoint sites and three PDF libraries. The mechanism is faster retrieval and consistent answers, which reduces transfers to second agents.
Unified customer profile addresses context switching. The agent sees the customer's cases, recent interactions, and relevant Sales records such as a mortgage application. That reduces repeat questions and improves first-contact resolution.
Copilot and agent assist can address handle time, but they depend on the knowledge base and case data being clean. They should be introduced after the knowledge consolidation, not as a substitute for it.
2. Where can Copilot realistically help reduce average handle time, and what limits should the bank set before enabling it?
Read the worked answer
Copilot in Customer Service can help in three concrete places. Case summaries reduce after-call work by summarizing the conversation and case timeline. Draft responses give the agent a starting point grounded in knowledge articles, which shortens searching and typing. Knowledge suggestions surface relevant articles during the conversation. Each acts on the agent's current task, so the benefit is measured in seconds per contact and in reduced after-call work.
The limits matter more than the feature list. Copilot output must be reviewed by the agent; it is not an autonomous adviser. The bank should restrict it from giving regulated advice or making credit decisions, and should log when a draft was used. Grounding data must be governed: if the knowledge base contains outdated policy PDFs, Copilot will summarize outdated policy confidently. The bank should also define which conversation data can be used and retained, and test for biased or unsafe responses before production.
A realistic target is a reduction in after-call work and search time, not a guaranteed reduction in talk time. The bank should pilot on one queue, compare handle time and quality scores against a control group, and keep human escalation paths unchanged. Without those controls, Copilot becomes a compliance risk rather than a productivity tool.
3. Which requirements in the scenario should not be solved by Customer Service or other customer engagement apps? Justify each exclusion.
Read the worked answer
Several requirements belong in core banking or specialized systems. Account balances, transactions, payments, card controls, and loan servicing are system-of-record functions. Customer engagement apps can display a summary through integration, but they must not become the ledger. Duplicating balances in Dataverse creates reconciliation risk and regulatory exposure.
Fraud adjudication and anti-money-laundering casework often require specialized investigation tooling, evidence handling, and regulatory reporting. A general case record can capture the customer interaction, but the investigation workflow and reporting should stay in the approved system.
Regulatory reporting and records retention should be met through the bank's records management and archiving controls. Customer Service can hold interaction history, but retention schedules, legal hold, and disposal must follow the bank's compliance platform.
Credit decisioning is a regulated decision process, not a service case. It may be initiated from a customer interaction, but the decision engine and model governance stay in the core or risk system.
The judgement is to use Customer Service for the interaction, orchestration, and service history, and integrate to systems of record for the facts. Trying to rebuild core banking in a customer engagement app is the failure mode.