The encyclopedia · Engineering & Operations · Operational decision · 1991
L.L. Bean staffed its call center by economics, saving $9–10M a year
A queueing model set trunks, agents and queue positions to balance service against cost, not hit a target.
L.L. Bean
The solution
L.L. Bean, a large telemarketer and mail-order catalog house, had set telemarketing resources by service-level criteria. The resulting staffing looked expensive, but nobody had priced the tradeoff between waiting customers and idle agents.
The OR team built an economic-optimization model using queueing theory: an expected total-cost objective balancing trunk, agent and queue costs against the revenue lost when callers abandon, with retrials modeled explicitly.
The economically optimal deployment differed significantly from the old service-level plan and saved an estimated $9–10 million a year. Management accepted the approach, which now sets trunks, agent schedules and maximum queue positions.
Why it worked
- Abandonment and retrial behavior were modeled, not assumed
- Economic optimum replaced an arbitrary service target
- The same model set trunks, agents and queue caps together
What can be applied
A service-level target is a proxy; model the real tradeoff between waiting cost and staffing cost, and let the optimum set the target.
Aftermath
The 1991 Interfaces paper documented the system as an Edelman competition finalist and became a classic early application of queueing economics to call centers.
Sources
- Allocating Telecommunications Resources at L. L. Bean, Inc.
- Allocating Telecommunications Resources at L. L. Bean, Inc. (abstract)
spotted an error? The archive wants to know.