Dynamics 365 Customer Service — Complete Guide
Case Management · Omnichannel · Knowledge Base · AI Copilot · Unified Routing · SLA · Scenarios · Cheat Sheet
Table of Contents
- Core Concepts — Basics
- Case Management — Deep Dive
- Omnichannel for Customer Service
- Knowledge Management & AI Copilot
- Unified Routing & Skills
- SLA & Entitlements
- Analytics & Reporting
- Customisation & Integration
- Scenario-Based Questions
- Cheat Sheet — Quick Reference
1. Core Concepts — Basics
What is Dynamics 365 Customer Service and what does it do?
Dynamics 365 Customer Service is a Microsoft customer relationship and service management application built on Dataverse. It enables organisations to manage customer support operations across multiple channels.
Core capabilities:
- Case management: create, track, route, escalate, and resolve customer issues
- Omnichannel: handle conversations across chat, email, voice, SMS, social media in a unified agent workspace
- Knowledge management: author, publish, and surface relevant knowledge articles to agents and customers
- SLAs: define and enforce response/resolution time commitments with automatic escalations
- AI and Copilot: agent assist, sentiment analysis, case summarisation, and smart routing
- Customer Service workspace: unified agent desktop combining all channels and case management
What are the main tables (entities) in Dynamics 365 Customer Service?
| Table | Description |
|---|---|
| Case (incident) | Central entity — represents a customer issue or service request |
| Account | Business/organisation customer |
| Contact | Individual customer |
| Queue | Work item bucket — cases and activities routed here for agents |
| Queue Item | Links a work item (case, email, chat) to a queue |
| Knowledge Article | Published support article for agents and customers |
| Entitlement | Defines support terms — number of cases, response SLA |
| SLA | Defines KPIs (first response time, resolution time) and breach actions |
| Activity | Email, Phone Call, Task, Chat, Voice — interactions linked to a case |
What is the Customer Service workspace vs Customer Service Hub?
Customer Service Hub: traditional model-driven app for case management. Provides clean interface for managing cases, knowledge articles, and queues. Does NOT include Omnichannel channels (chat, voice, SMS).
Customer Service workspace: modern unified agent desktop. Includes all Customer Service Hub features PLUS Omnichannel channels — live chat, voice, SMS, social messaging — all in a single tabbed multisession interface.
Tip: For any new D365 Customer Service implementation, Customer Service workspace is the recommended agent interface. Customer Service Hub is used by administrators and supervisors for configuration and oversight.
What are the D365 Customer Service licensing tiers?
| Licence | Key Features |
|---|---|
| Customer Service Professional | Core case management, knowledge base, SLAs, basic routing. No Omnichannel. |
| Customer Service Enterprise | All Professional + advanced routing (Unified Routing), AI features, Copilot, analytics |
| Omnichannel add-on | Live chat, SMS, social messaging, voice — on top of Enterprise |
| Copilot for Service | AI assistant for agents — case summarisation, response drafting, knowledge search |
Warning: Omnichannel is a separate add-on — not included in base Customer Service Enterprise. Voice channel requires an additional Azure Communication Services resource and licensing.
2. Case Management — Deep Dive
What is the Case lifecycle and what are the key statuses?
Case Status (statecode):
Active (0) → Case is open and being worked
Resolved (1) → Case has been resolved — resolution logged
Cancelled (2) → Case cancelled (duplicate, invalid, customer withdrew)
Status Reasons under Active:
In Progress → agent is actively working the case
On Hold → waiting for customer or third party
Waiting for Details → waiting for customer to provide more info
Researching → agent is investigating
Status Reasons under Resolved:
Problem Solved → issue fixed
Information Provided → customer needed guidance
Cancelled → not a valid issue
Case Lifecycle:
[Created] → [Assigned/Routed to Queue] → [In Progress]
→ [On Hold if awaiting info] → [Resolved] → [Closed after X days]
Key rules:
→ SLA KPIs start when case is created
→ SLA pauses when Status Reason = On Hold (configurable)
→ Resolution records time spent, resolution type, and description
→ Closed cases cannot be re-opened — create new case or follow-up
What are Queues in D365 Customer Service?
Queues are work item buckets holding cases and activities waiting to be picked up by agents.
Queue types:
| Type | Visibility | Use Case |
|---|---|---|
| Public | All agents with access | General team queues |
| Private | Specific users/teams only | Specialist team queues |
| Personal | Each user's own queue | Individual workload |
Queue workflow:
Case created → assigned to Queue (manually or routing rule)
Agent opens Queue → sees all items → clicks "Pick" to take ownership
OR: supervisor assigns directly to agent (bypasses queue pick)
Queue Item states:
Queued → waiting to be picked up by an agent
In Progress → assigned to and being worked by an agent
Tip: In Omnichannel, queues are extended to handle live conversations (chat, voice, SMS) in addition to cases — the same queue concept applies across all work types.
What is the Parent-Child Case relationship?
Parent-Child cases allow one complex case to be broken into sub-cases worked by different teams in parallel.
Use case: Customer reports complete system failure
Parent case: "System Outage — Customer X" (Account Manager)
Child case 1: "Network investigation" → Network team queue
Child case 2: "Database connectivity" → DBA team queue
Child case 3: "Application error" → Dev team queue
Status sync options:
→ Auto-resolve parent when ALL child cases are resolved
→ Or: manual — parent managed independently
SLA: each child case has its own SLA clock
Customer view: sees one parent case — internal complexity hidden
What are the Case subject tree and categories?
The Subject tree is a hierarchical classification structure for cases — enabling consistent categorisation, routing, and analytics.
Subject tree example:
Products
└── Product A
├── Technical Issues
│ ├── Installation
│ └── Performance
└── Billing Issues
├── Invoice Query
└── Refund Request
Services
└── Professional Services
└── Onboarding
Benefits:
→ Consistent case classification across teams
→ Routing rules: IF subject = 'Billing Issues' → Finance queue
→ Analytics: case volume and resolution time by topic
→ Knowledge article association: link articles to subject areas
3. Omnichannel for Customer Service
What is Omnichannel and what channels does it support?
Omnichannel for Customer Service is the real-time customer engagement layer routing customer conversations from multiple digital channels to agents in a unified desktop.
| Channel | Description |
|---|---|
| Live Chat | Web chat widget embedded on websites or portals |
| Voice | Inbound/outbound phone calls via Azure Communication Services |
| SMS | Two-way text messaging |
| Microsoft Teams | Internal escalation from Teams to D365 agent |
| Enhanced email routing within agent workspace | |
| Facebook Messenger | Social messaging |
| Messaging via Business API | |
| Twitter/X, WeChat, LINE, Telegram | Via third-party providers |
| Custom channel | Direct Line API integration for any custom channel |
Tip: All channels surface in the same Customer Service workspace — agents see one unified inbox regardless of channel.
What is the Omnichannel agent workspace and its key features?
- Active Conversation screen: central workspace — customer conversation alongside customer summary (name, account, open cases, recent history)
- Multisession: agents handle multiple conversations simultaneously via tabbed sessions
- Customer 360 summary: contact details, account info, recent cases, sentiment score, conversation history
- Communication panel: channel-specific interaction (chat transcript, call controls, email compose)
- Productivity pane: agent assist — knowledge search, similar cases, AI-suggested responses, Copilot
- Presence management: agent sets availability (Available, Busy, Do Not Disturb, Away) — routing only assigns to available agents
- Supervisor dashboard: real-time monitoring of agent workloads, queue depths, sentiment, SLA compliance
What is agent capacity in Omnichannel?
Agent capacity defines the maximum workload an agent can handle simultaneously. Each work type consumes capacity units.
Example capacity configuration:
Agent total capacity: 100 units
Work item costs:
Live Chat: 20 units per conversation
Voice call: 50 units (prevents multi-tasking during calls)
SMS: 10 units per conversation
Case: 5 units per case
Concurrent handling:
→ 1 voice (50) + 2 SMS (20) + 2 cases (10) = 100 (full)
→ 5 chats (100) = 100 (full)
→ 4 chats (80) + 2 SMS (20) = 100 (full)
When capacity = 0: no new assignments
When presence = Busy: capacity treated as 0 for new work
Tip: Configure voice capacity close to total agent capacity to prevent agents receiving chat while on a call.
How does the Copilot Studio bot-to-human handoff work in Omnichannel?
Customer journey:
1. Customer starts chat on website
2. Copilot Studio bot handles conversation:
→ Answers FAQs from knowledge base
→ Collects customer details (name, account, issue type)
→ Resolves simple issues (password reset, account balance)
3. Bot cannot resolve → escalates to D365 Omnichannel
Context passed to agent:
→ Customer name, email, account ID (collected by bot)
→ Issue category (determined by bot intent recognition)
→ Full bot conversation transcript
→ Bot escalation reason ("Cannot resolve: Complex billing issue")
D365 uses context to:
→ Pre-populate case fields (customer, subject, description)
→ Route to correct queue based on bot-determined category
→ Show agent a summary of the bot interaction
Result: customer never repeats themselves — agent continues from
exactly where the bot left off
Tip: Bot-to-human escalation with full context handoff is one of the most commonly asked D365 CS architecture scenarios in senior .
4. Knowledge Management & AI Copilot
What is the Knowledge Base lifecycle?
Knowledge Article statuses:
Draft → In Review → Approved → Published → Expired
Draft: Author creates article — title, content, keywords, related products
In Review: Submitted to knowledge manager for review
Approved: Reviewed and approved — not yet visible to external users
Published: Live — visible to agents AND customers on self-service portal
Scheduled: Auto-publishes on a future date
Expired: Past expiry date — auto-unpublished, needs renewal
Key features:
→ Versioning: each edit = new version, previous versions preserved
→ Translation: articles translated into multiple languages
→ Feedback: agents/customers rate articles (helpful/not helpful)
→ Analytics: views, likes, dislikes, cases linked to article
→ Internal vs external: internal-only (agents) or external (portal)
→ Keywords: improve search relevance
→ Related articles: link related content for agent context
How does Copilot for Customer Service work?
Copilot is an AI assistant in the agent productivity pane powered by Azure OpenAI.
| Capability | Description |
|---|---|
| Ask a question | Searches knowledge base + similar cases → synthesised answer with citations |
| Draft a response | Generates contextually appropriate reply based on conversation + KB |
| Case summary | One-click AI summary: customer issue, steps taken, current status |
| Conversation summary | Auto-generates end-of-conversation notes for case records |
| Suggested articles | Surfaces most relevant KB articles based on case context |
Tip: Copilot for Customer Service measurably reduces Average Handling Time (AHT) and improves First Contact Resolution (FCR). Know its capabilities for any AI-focused question.
What is real-time sentiment analysis?
Real-time sentiment analysis uses Azure AI to analyse customer messages in live chat or voice conversations and display a sentiment score — updated as the conversation progresses.
Sentiment levels: Positive, Slightly Positive, Neutral, Slightly Negative, Negative
How it's used:
- Supervisor sees sentiment for ALL active conversations in the Supervisor Dashboard
- Agent sees their own conversation's sentiment in the conversation panel
- When sentiment drops to Negative: supervisor can be alerted and may intervene
- Sentiment stored on conversation record — available for historical analysis
5. Unified Routing & Skills
What is Unified Routing and how does it work?
Unified Routing is the intelligent work distribution engine routing both cases (record routing) and Omnichannel conversations to the most appropriate queue and agent.
Unified Routing pipeline:
Work item arrives (case created OR conversation started)
↓
[Classification] — AI or rule-based
→ Set: priority, skills required, category, sentiment, language
↓
[Route to Queue] — routing rules
Rule: IF language = 'French' → France Support Queue
Rule: IF priority = 'Critical' → Tier3 Escalation Queue
Rule: IF topic = 'Billing' → Finance Queue
Rule: IF entitlement_tier = 'Gold' → Gold Priority Queue
↓
[Assignment] — assignment method
1. Highest capacity → assign to agent with most available capacity
2. Round robin → distribute evenly across available agents
3. Least active → agent with fewest active sessions
4. Skills-based → match required skills to agent skill profile
Tip: Unified Routing replaces both the older basic routing rules for cases AND handles real-time Omnichannel conversation routing in a single engine. Always use Unified Routing for new implementations.
What are Skills and how do you implement skills-based routing?
Skills setup:
1. Define skill types: Language, Product Knowledge, Technical Level
2. Create skills: French (10), SAP ERP (8), Network+ (9), Level2 (7)
3. Assign skills to agents with proficiency (1-10):
Maria: French(10), SAP ERP(8), Level2(7)
John: English(10), Network+(9), Level1(5)
Skills classification on work item:
AI-based: NLP analyses case content → auto-assigns required skills
→ Detects language → requires Language skill
→ Detects "SAP" keyword → requires SAP ERP skill
Rule-based: IF product = 'SAP' THEN require SAP ERP skill ≥ 7
Assignment with skills:
→ Exact match: agent has ALL required skills at required proficiency
→ Closest match: if no exact, find closest skill profile
→ Skill relaxation: after 5 min, lower proficiency requirement
after 10 min, relax to any available agent
Tip: Skill relaxation is critical to prevent long wait times. Balance quality (exact skill match) with responsiveness (relaxation over time).
What are the Omnichannel Supervisor capabilities?
The Supervisor Dashboard provides real-time visibility into:
| View | Metrics |
|---|---|
| Conversations | Active conversations, status, channel, sentiment, duration, agent |
| Agents | Agent presence, active conversations, capacity utilised |
| Queues | Queue depth, average wait time, longest waiting item |
| Ongoing conversations | Ability to monitor, assign, transfer, or join conversations |
Supervisor actions:
- Monitor: silently observe a conversation (agent unaware)
- Barge-in: join an active conversation (agent and customer see supervisor join)
- Transfer: move a conversation from one agent to another
- Assign: pull a queued item directly to a specific agent
6. SLA & Entitlements
What are SLAs in D365 and what are the two types?
Standard SLA: basic SLA with a single KPI. When breached, a warning flag shows on the case. No automatic actions.
Enhanced SLA (recommended): supports multiple KPIs, pause/resume, and automatic breach actions.
Enhanced SLA example — "Gold Customer SLA":
KPI 1: First Response By
Warning: 30 minutes → action: email agent's manager
Failure: 1 hour → action: escalate to Tier2, email customer
KPI 2: Resolve By
Warning: 6 hours → action: alert supervisor
Failure: 8 hours → action: escalate to manager, notify account team
Pause condition: Status Reason = "Waiting for Details"
Resume: when Status Reason changes back to "In Progress"
SLA timer on case form:
→ Countdown timer: time remaining until breach
→ Amber indicator: warning threshold reached
→ Red indicator: SLA breached
→ Paused indicator: case on hold
Warning: Always use Enhanced SLA for production. Standard SLA lacks pause/resume and automated breach actions. Enhanced SLA is the current standard.
What are Entitlements?
Entitlements define the support terms for a customer — how many cases they are entitled to, applicable SLA, and which channels they can use.
Entitlement structure:
Entitlement → linked to Account or Contact
Allotment type: Number of cases (e.g., 10 per year)
OR Hours (e.g., 40 hours per year)
Total terms: 10
Remaining terms: auto-decremented when case is created
Start/end date: validity period
Associated SLA: Gold SLA (applied automatically to cases)
Channels: Phone + Chat + Email (Gold) or Email only (Bronze)
When case created for customer with entitlement:
→ D365 auto-associates the entitlement to the case
→ Remaining terms decremented by 1
→ When terms = 0: agent sees warning — entitlement exhausted
→ Supervisor can override and continue or reject new case
Tip: Entitlements are the bridge between sales contracts and service delivery — they ensure customers receive exactly what they paid for. Agents see entitlement status directly on the case form.
7. Analytics & Reporting
What analytics are available in D365 Customer Service?
Built-in Power BI dashboards:
| Dashboard | Key Metrics |
|---|---|
| Summary | Total cases, AHT, FCR rate, CSAT, SLA compliance |
| Agent | Cases handled, avg resolution time, CSAT, SLA compliance per agent |
| Queue | Avg wait time, queue depth, abandon rate, longest waiting item |
| Omnichannel | Conversations by channel, abandonment, avg conversation time, sentiment trends |
| Knowledge | Article views, likes/dislikes, articles linked to resolved cases |
| Bot | Self-service deflection rate, escalation rate, bot CSAT |
Key KPIs:
CSAT → Customer satisfaction score (post-resolution survey rating)
FCR → First Contact Resolution % (resolved without re-open or follow-up)
AHT → Average Handling Time (total agent time per case/conversation)
ASA → Average Speed of Answer (arrival to agent assignment time)
Abandon Rate → % customers who left queue before being answered
SLA Compliance → % cases resolved within SLA target
What is Customer Service Insights?
Customer Service Insights provides AI-driven analytics identifying topics driving case volume and opportunities to reduce it.
- Topic clustering: AI groups similar cases into topics automatically — no manual categorisation
- Impact analysis: topics with highest volume, longest resolution time, or worst CSAT
- Emerging issues: alerts when a new topic suddenly spikes in volume — early warning for product issues
- Deflection opportunities: identifies topics where knowledge articles or chatbot could deflect live agent contacts
8. Customisation & Integration
What are the key customisation points in D365 Customer Service?
| Layer | Customisation |
|---|---|
| Forms | Customise Case, Contact, Account forms — add/remove fields, sections, tabs |
| Views | Configure list views for queues, cases, knowledge articles |
| Business rules | Field show/hide/set/validate logic (client-side or server-side) |
| Business process flows | Stage-based guidance for agents working complex cases |
| Plugins | C# server-side logic on case create/update (e.g., auto-assign, validation) |
| Power Automate flows | Automated actions triggered by D365 events (SLA breach, case status change) |
| PCF controls | Custom UI components on case forms |
| API | Dataverse Web API for external system integration |
How do you integrate D365 Customer Service with external systems?
Common integration patterns:
- Dataverse Web API: CRUD operations on cases, contacts, accounts from external systems
- Power Automate: event-driven integration — when case created → notify Slack/Teams, update ERP
- Webhooks: real-time push notifications to external systems on D365 events
- Azure Service Bus: reliable async messaging for high-volume case creation from external portals
- Embedded experience: embed D365 widgets (chat, case history) in external portals via iFrame or JS API
- CTI (Computer Telephony Integration): connect third-party phone systems to Omnichannel via Channel Integration Framework (CIF)
What is the Channel Integration Framework (CIF)?
CIF is a JavaScript API framework that allows third-party telephony and communication providers to integrate with the D365 Customer Service workspace. It provides:
- CIF v1: Single-session — one communication provider in the side panel
- CIF v2: Multi-session — multiple providers, integrates with Omnichannel routing
Use cases:
- Connect existing Avaya, Cisco, Genesys, or Five9 phone systems to D365
- Surface caller information (screen pop) in D365 before agent answers
- Enable click-to-call from D365 records
- Log call activities automatically to case records
9. Scenario-Based Questions
Scenario: Design a D365 Customer Service solution for a bank with Gold/Silver/Bronze customer tiers.
Architecture:
-
Entitlements per tier:
- Gold: unlimited cases, phone + chat + email, 1h first response / 4h resolution
- Silver: 20 cases/year, chat + email, 4h first response / 24h resolution
- Bronze: 5 cases/year, email only, 24h first response / 72h resolution
-
Enhanced SLAs: three SLAs with appropriate KPIs and breach actions. Auto-associate via entitlement.
-
Queues: Gold Priority Queue, Silver Queue, Bronze Queue — segregated so Gold is never affected by Bronze volume.
-
Unified Routing rule: check entitlement tier on case creation → route to tier-appropriate queue → assign with highest capacity method.
-
Channels: Gold gets Omnichannel (chat + phone). Silver/Bronze get email only.
-
Agent skills: senior agents tagged "Gold" skill → assigned to Gold queue. Junior agents handle Bronze.
-
Reporting: SLA compliance by tier. Escalation alerts when Gold SLA approaching breach.
Scenario: A case is stuck in queue for 2 hours without being picked up. How do you handle this?
Prevention (proactive):
- SLA breach actions: configure "First Response By" KPI — at 1h warning: email queue manager. At 2h breach: auto-escalate to supervisor, change priority to Critical.
- Power Automate escalation: trigger on SLA KPI breach → Teams notification to supervisor with case link and customer tier
- Supervisor real-time dashboard: "Longest Wait" column — supervisors see queue depth and waiting time, can manually assign
Reactive (when it happens):
- Supervisor opens case from dashboard → assigns directly to available agent
- Or: supervisor uses "Assign" from queue view to push to specific agent
- Post-incident: review routing rules and agent capacity allocation for the queue
Scenario: How do you implement seamless chat-to-voice escalation in Omnichannel?
Requirement: Customer on live chat → issue is complex → needs a phone call without losing chat context.
- Agent initiates: in chat session → clicks "Consult/Transfer" → selects Voice → enters customer's phone number
- D365 initiates outbound call via Azure Communication Services
- Multisession merge: voice opens in a new tab. Original chat transcript remains visible alongside.
- Context preserved: agent has full chat history during the voice call — customer never repeats
- Single case record: both chat and voice activities linked to the same case — unified timeline
- Transfer to specialist: if escalating further, all context (chat transcript + call notes) transfers with the conversation
Key value: customer never repeats themselves. Context travels with the conversation across channel escalations.
Scenario: How do you measure and improve First Contact Resolution (FCR) rate?
Measuring FCR in D365:
FCR = cases resolved in first contact without re-open
Power BI calculation:
FCR = (Cases resolved without follow-up) / (Total resolved cases) × 100
Track in D365:
→ Cases with no follow-up case linked (no re-open)
→ Cases where resolvedon = first contact date
→ Monitor re-opened cases (cases resolved then re-activated)
Improving FCR:
- Knowledge base quality: Knowledge Analytics → find articles rated unhelpful or linked to re-opened cases → improve or replace
- Copilot for agents: AI-suggested responses and case summaries reduce resolution errors that cause re-opens
- Skills-based routing: route complex issues to specialist agents first time — avoid unnecessary tier 1 escalations
- Customer Service Insights: identify topics with worst FCR rates → target for KB article creation or agent training
- Root cause tracking: add "Reason for re-open" field on re-opened cases → analyse patterns (wrong resolution, incomplete fix)
Scenario: How do you configure D365 Customer Service to handle a sudden spike in cases during a product outage?
- Increase agent capacity: temporarily raise agent capacity units during the spike — allows agents to handle more simultaneous conversations
- Deploy Copilot Studio bot: configure a bot to handle the most common outage-related queries (status updates, workarounds) — deflect volume from live agents
- Create outage knowledge article: publish immediately with known issue, workaround, and estimated resolution time — agents share this rather than typing individual responses
- Bulk case creation: if customers are proactively notified, pre-create cases for affected accounts with status "Researching" — no customer needs to call
- Copilot case summary: agents use Copilot to rapidly understand incoming cases and apply the known resolution faster
- Supervisor dashboard: monitor queue depths and agent sentiment — intervene early before queues back up significantly
- Post-incident: Customer Service Insights will cluster outage cases into a topic — use for retrospective analysis and knowledge base improvement
10. Cheat Sheet — Quick Reference
Case Status Quick Reference
statecode (Status):
0 = Active 1 = Resolved 2 = Cancelled
statuscode (Status Reason) — Active:
1 = In Progress 2 = On Hold 3 = Waiting for Details 4 = Researching
statuscode (Status Reason) — Resolved:
5 = Problem Solved 1000 = Information Provided 6 = Cancelled
SLA behaviour:
→ SLA starts: on case creation
→ SLA pauses: when statuscode = 'On Hold' or 'Waiting for Details'
→ SLA resumes: when statuscode returns to 'In Progress'
→ SLA stops: when statecode = Resolved or Cancelled
Unified Routing Quick Reference
Components:
Workstream → defines channel + routing mode (push/pick)
Routing rules → classify and assign to queues
Queue → holds work items
Assignment method → how agents are selected from queue
Assignment methods:
Highest capacity → most available capacity units
Round robin → even distribution
Least active → fewest active sessions
Skills-based → skill + proficiency match
Skill relaxation:
After X minutes: lower required proficiency
After Y minutes: accept any available agent in queue
SLA vs Entitlement
SLA (Service Level Agreement):
→ WHAT response/resolution time is committed
→ Defines KPIs, warning thresholds, breach actions
→ Linked to: Entitlement or applied globally
→ Timers shown on: case form
Entitlement:
→ HOW MANY cases/hours a customer is entitled to
→ Tracks remaining support terms
→ Defines: channels available, associated SLA
→ Linked to: Account or Contact
→ Consumed when: case is created
They work TOGETHER:
Entitlement → defines TERMS + specifies which SLA
SLA → defines the TIME commitments for those terms
Omnichannel Channel Architecture
Customer channels:
Website → Live Chat widget → Chat workstream → Queue → Agent
Phone → ACS PSTN → Voice workstream → Queue → Agent
SMS → ACS/TeleSign → SMS workstream → Queue → Agent
WhatsApp → WhatsApp Business API → Messaging workstream → Queue → Agent
Facebook → Facebook App → Social workstream → Queue → Agent
Custom → Direct Line API → Custom workstream → Queue → Agent
All conversations surface in:
→ Customer Service workspace (agent side)
→ Omnichannel Supervisor Dashboard (supervisor side)
Context passed with every conversation:
→ Customer identity (if authenticated)
→ Bot transcript (if escalated from bot)
→ Pre-chat survey responses
→ Routing context variables
Key KPIs Reference
CSAT = Customer Satisfaction Score
(post-interaction survey, scale 1–5 or 1–10)
FCR = First Contact Resolution Rate
(resolved in first contact / total cases × 100)
AHT = Average Handling Time
(total agent time: hold + talk + wrap-up)
ASA = Average Speed of Answer
(queue arrival → agent pick-up time)
ABN = Abandon Rate
(customers who left queue before answered / total arrivals × 100)
SLA% = SLA Compliance Rate
(cases resolved within SLA / total cases × 100)
Deflection Rate = cases resolved by bot / total contacts × 100
NPS = Net Promoter Score (would you recommend? 0-10)
Top 10 Tips
- Case = incident table — the central entity in D365 CS. Know its key fields: subject, description, priority (1=High, 2=Normal, 3=Low), status, status reason, customer (polymorphic: Account or Contact), owner, SLA.
- Enhanced SLA over Standard — Enhanced SLA supports multiple KPIs, pause/resume, and automatic breach actions. Standard SLA is legacy. Always recommend Enhanced for production.
- Unified Routing replaces basic routing — Unified Routing handles both cases (record routing) and Omnichannel conversations in one engine. Know the pipeline: Classification → Route to Queue → Assignment.
- Omnichannel add-on is separate — chat, voice, SMS are not in base Customer Service Enterprise licence. Always confirm add-on requirements when asked about Omnichannel architecture.
- Bot-to-human handoff with context — the key architecture pattern: Copilot Studio bot handles first contact → escalates to Omnichannel with full conversation context → agent continues seamlessly.
- Agent capacity controls concurrency — voice typically consumes 50+ units to prevent multi-channel conflicts. Know how to configure and why.
- Skill relaxation prevents queue backup — gradually lower skill proficiency requirements over time to prevent long waits when exact-match agents are busy.
- Entitlements connect contracts to service — they auto-associate SLAs to cases and track remaining support terms. Know the allotment types (cases vs hours).
- Copilot reduces AHT and improves FCR — the AI capabilities (ask a question, draft response, case summary) are the primary D365 CS differentiators in 2025. Know all five capabilities.
- Customer Service Insights = proactive improvement — topic clustering identifies volume drivers automatically. Deflection opportunities guide knowledge base priorities. This is the data-driven continuous improvement story.
No comments:
Post a Comment