For most retailers a fifth of the year's orders arrive in six weeks. The warehouse that copes easily in May is the one that fails in late November, and by then it is too late to change anything but the apology. The work that decides peak is done in August and September.

Forecast by line, not by total

A forecast of «three times normal» is no help to the person laying out the shelves. What matters is which 200 lines will make up most of the orders. Give those the positions nearest the packing benches, at waist height, and move the slow lines upstairs.

Get the stock in early

Inbound containers and outbound orders compete for the same doors and the same people. We ask peak customers to land their stock by the middle of October, so that November is spent picking and not unloading. A container that arrives in peak week waits its turn like everything else.

Decide the cut-off and keep it

An order cut-off of 16:00 that holds every day is worth more than one of 18:00 that fails twice a week. Publish it on the checkout page and let the warehouse plan its last truck around it.

• Agree packaging sizes in advance; three box sizes cover most ranges
• Plan the returns bench as carefully as the packing bench: January is its peak
• Share the marketing calendar with the warehouse, since a promotion is a shift pattern

Peak is not a surprise. It is on the calendar on the same days every year.

Space that shrinks again

The reason to share a warehouse is January. Stock that needed 3,000 pallet positions in November needs 1,000 by February, and a shared site lets the bill follow the stock. Ask any provider how the charge falls after peak, not only how the space grows before it.