A capacity number is not a recovery plan.
Queues recover only when effective output remains above incoming demand long enough to reach a defined control threshold.
Teams often begin with theoretical capacity: stations, labor hours, system slots, or vendor commitments. That is a useful ceiling, but it ignores utilization loss, rework, aged-work complexity, downtime, and the volatility of incoming demand.
A date becomes credible only after those losses are visible. Otherwise, the “recovery plan” is a target divided by an output rate the operation has never demonstrated.
Every factor should be observable, explicitly assumed, or labeled for validation.
Three equations turn a queue into a decision.
Net backlog burn
Effective throughput − weekly inbound. If the result is zero or negative, the queue will not recover under the current assumptions.
Weeks to threshold
(Current backlog − control threshold) ÷ net burn. Use ceiling rounding when commitments are expressed in whole weeks.
Required throughput
Weekly inbound + backlog gap ÷ target weeks. The difference from effective throughput is the capacity gap.
Pressure-test the promise before publishing the date.
- Define the queue.Confirm what enters, exits, ages, reopens, or bypasses the backlog measure.
- Choose the control threshold.“Zero” is rarely required. Define the stable level the operating system can manage.
- Use demonstrated output.Reconcile rated capacity with realized utilization, yield, downtime, and skill constraints.
- Separate temporary and structural capacity.Overtime or contractors can bridge a window; they should not disguise an unstable base system.
- Stress demand and execution.Test what happens when inbound or effective throughput misses the central assumption.
- Name the decision owner.Assign who can fund capacity, reset scope, control intake, or change the target date.