Inventory control
Inventory control is not one forecast: bind actions to checkpoints
An inventory action should be a reversible decision: how much to open or protect, for how long, and which data ends it.
One checkpoint, one question
Use D-28 to test whether the cumulative difference is only an early-sales effect, D-14 for recent pickup and remaining demand, and D-7 for cancellations, groups, connections, and remaining higher-fare demand. Record curve version and sellable capacity each time.
| Checkpoint | Reference | Example | Difference |
|---|---|---|---|
| D-28 | 72 | 76 | +4 |
| D-14 | 116 | 121 | +5 |
| D-7 | 148 | 154 | +6 |
Protecting 18 seats is not locking 18 seats
Assume 121 bookings at D-14 and a current final forecast of 181. The exercise temporarily protects 18 seats for later demand with a 48-hour review. Eighteen is a control parameter, not a promise of 18 higher-fare sales.
At review, use both cumulative bookings and the latest three-day pickup. If pickup trails reference, a competitor adds capacity, or a group cancels, re-estimate remaining demand instead of mechanically holding the protection level.
Write the stop condition
The action record needs quantity, cabin, start time, next review, withdrawal trigger, and owner. A temporary action without a stop condition can become a permanent setting that nobody revisits.
When the old curve stops being comparable
After changes in aircraft, frequency, departure time, distribution, or route rules, an old curve may no longer be comparable. Inventory control needs current capacity and commercial information, not automatic execution from history.