VIP Service Cases Breaching SLA in a Shared Work Queue
A financial-services CS application has 140 agents. Cases are created as Service cases with a Customer tier property: Standard, Gold, Platinum. Platinum cases carry an urgency of 10 and a 4-hour SLA; Gold 8 and 8 hours; Standard 5 and 24 hours. All new cases are routed by a single assignment shape configured as Route to workbasket to a workbasket named CSGeneral. Agents are organized into three work groups: General, Billing, Retention. The Platinum skill is assigned to 22 operators in Retention, but the workbasket is owned by General. Last quarter, 41% of Platinum cases breached the 4-hour SLA, and 28% were first worked by an agent without the Platinum skill. Agents say they pull from GetNextWork and see whatever is oldest. Management wants to keep a single queue for visibility.
1. Why does the current design fail to honor tier and skill, even though urgency and skills exist?
Read the worked answer
The case tier and urgency are not the same as assignment routing. The assignment shape routes to a workbasket, and GetNextWork selects the oldest available assignment in that workbasket according to assignment urgency and the operator's access. Because the assignment has no required skill, the Platinum skill on the operator record does not restrict or prioritize work. Because the workbasket is owned by General, the 22 Retention operators with Platinum skill cannot see the cases unless they also belong to that work group.
The case urgency of 10 may not even be copied to the assignment urgency; that must be set explicitly. The result is predictable: General agents without Platinum skill pull the oldest case, while skilled Retention agents never see Platinum work. A single workbasket provides visibility, but it defeats skill-based routing unless the routing itself uses skills and the queue is accessible to all qualified groups.
2. Propose a routing design that keeps one logical queue but meets the SLA, including skill and overflow handling.
Read the worked answer
Keep one logical queue but route to a skilled group instead of a workbasket. On the assignment, set required skill Platinum for Platinum tier, Gold for Gold, and use Route to skilled group so GetNextWork matches skills. Set assignment urgency explicitly from tier (Platinum 10, Gold 8, Standard 5) and put the SLA on the assignment so goal and deadline follow the routed assignment. Give all three work groups access to the queue, or route to a work group that contains the qualified agents.
For overflow, add a conditional: if no Platinum-qualified agent is available within 30 minutes, escalate to a Retention supervisor and allow Gold skill as fallback, with a separate 1-hour goal. Reporting can still show one queue by reporting on the case type and work queue, so the single-queue visibility requirement is met without forcing one workbasket. Test with an operator who has only Gold skill to confirm they cannot pull Platinum work before the escalation.
3. What configuration evidence would you inspect to confirm that urgency is actually driving assignment order?
Read the worked answer
Check three places. First, the assignment's own urgency property, not just the case property; if the assignment does not set urgency, GetNextWork may use a default and the tier urgency is irrelevant. Second, the SLA configuration: if the SLA is attached to the case, goal and deadline are calculated at case creation and do not automatically change when the assignment is routed or transferred. Third, the GetNextWork algorithm and work queue assignment order, because a queue can be configured to use FIFO rather than urgency.
Also inspect whether the assignment is routed to a workbasket or a skilled group, since only the latter can use skill matching. Use the assignment's Where Used and the work queue configuration to prove the path. A useful test is to create three cases in one queue with different tier urgencies and confirm the order an operator receives them; if the order is wrong, fix urgency propagation before changing skills.