An annual support-triage review should answer two questions: does work reliably reach the right owner, and can the team detect and recover when it does not? Use a fixed review period, representative cases, and the same metric definitions as the prior review. The goal is not to generate a large dashboard; it is to find rules, queues, and obligations that no longer match reality.
Confirm scope and ownership
List every intake channel: email, form, chat, phone, social, marketplace, monitoring alert, and internal escalation. For each, name the authoritative system, default queue, business owner, technical owner, hours of coverage, and fallback when automation fails. Retire undocumented side channels or bring them into the operating guide.
Inventory queues and their purpose. Microsoft’s unified routing queue guidance describes queues as collections of cases or conversations distributed to representatives using assignment methods. A queue should exist because it changes ownership, skill, priority, access, or service obligation—not merely because it was created years ago.
Check membership and least privilege. Remove departed users through the approved account process, verify specialist access, and confirm that sensitive cases are not visible to a broader queue than intended. Record who may change routing rules and who approves emergency overrides.
Review demand and backlog with stable definitions
Choose a representative period and record filters, channels, queues, status rules, and time zone. Compare incoming volume, assigned and unassigned work, longest wait, transfer rate, reopenings, abandonment where relevant, and the age distribution of open requests. Microsoft’s record-routing queue report includes queue priority, assigned and unassigned work, longest wait, and agents online.
Use averages carefully. Microsoft’s queue dashboard guidance also describes incoming and engaged sessions, average wait, transfer rate, rejected or timed-out sessions, and handle-time measures. Averages can hide one stranded urgent case, so review the oldest items and a sample of priority cases individually.
Reconcile open work. Every active request should have a valid owner, next action, and understandable priority. Sample closed work for premature closure, duplicate replies, unnecessary transfers, and repeated reopening. Keep customer content in its access-controlled system; use identifiers in the review record.
Challenge classification and priority rules
List categories, labels, priorities, service commitments, and escalation rules. For each rule, identify the evidence that it still helps. Merge near-duplicate categories, but preserve distinctions required for safety, security, privacy, legal handling, payment issues, account access, or contractual service.
Test ambiguous cases. A useful default route should receive work that no classifier understands. Verify urgent signals cannot be downgraded by a missing optional field and that customers are not required to diagnose their own technical category before help begins.
Review AI or automation separately. Record its purpose, input, decision boundary, human override, failure mode, monitoring, and rollback. Compare a labeled sample with human decisions and inspect errors by customer impact, not only aggregate accuracy. Do not allow a model-generated summary to replace the source conversation for consequential decisions.
Audit changes and dependencies
Review routing, queue, priority, workstream, channel, and service-level changes made during the year. Microsoft’s Customer Service audit guidance lists queues, routing rules, prioritization rules, workstreams, channels, and SLA-related configuration among auditable areas. Confirm the organization’s actual audit retention and access rather than assuming defaults.
Inventory integrations, webhooks, mailboxes, bots, identity providers, and reporting exports. Name an owner and renewal or review date for each. Test one harmless synthetic request through every critical intake path and verify delivery, assignment, notification, reply, and closure.
Rehearse failure and recovery
Run a tabletop or controlled exercise: stop one routing rule, make one integration unavailable, or use a test environment to create an unassigned request. Confirm the alert, manual intake path, authoritative queue, communications owner, and rollback steps. Account for every synthetic item after recovery.
Review backups or exports according to the platform’s supported process. A successful export is not automatically a restorable service. Record scope limits, file-link dependencies, retention, and who is authorized to access the material.
Refresh the operating guide and decisions
Update a short guide covering intake, priority, ownership, transfer, escalation, privacy, outage fallback, and incident reporting. Give synthetic cases to a new or rotating team member and observe where the guide fails. Training confusion is evidence about the workflow.
Finish with a decision log: queues to retain or merge, rules to change, access actions, monitoring gaps, tested fallbacks, unresolved risks, owners, due dates, and approval. Preserve the baseline so next year's review can compare like with like.
The annual review passes when supported work is accounted for, priority and access rules match current obligations, automation is bounded and reversible, and the team can explain how an unassigned or misrouted request will be found. A clean dashboard alone is not enough.
Sources
- Create and manage queues for unified routing — Microsoft Learn
- View and understand real-time analytics for record routing in the queue report — Microsoft Learn, June 26, 2026
- Queue dashboard — Microsoft Learn
- Enable tables for Customer Service audit logs — Microsoft Learn