Seasonal restocking: order against the season opening, not the empty shelf

Souvenir retail runs on a clock that most sourcing calendars ignore. The shelf empties in a burst, and the instinct is to reorder when you see it empty. By then the season that would have sold the stock is over, and the goods arrive to sit in a stockroom for most of a year.
The fix is not ordering more. It is ordering against the season’s opening date rather than against the shelf’s current state.
Work backwards from the date the season opens
The chain that has to fit before the first visitor arrives is longer than the production time, and production is usually the smallest piece of it:
- Artwork adaptation and approval rounds
- Tooling, if any design in the order is new
- Pre-production sample, shipped, reviewed and approved
- Production
- Inspection
- Freight, customs clearance and inland delivery
- The retailer’s own receiving and merchandising window, which is real and is never zero
Ask your supplier for each of those as a separate figure rather than one lead time, because a single number hides which stage will actually slip. Approval round-trips and freight are the two that move most, and neither of them is inside the factory.
The three orders a season actually needs
The opening order
Placed deep in the off-season, against a date rather than against demand. This is the order that carries new designs and new tooling, because it is the only one with enough runway for a sampling cycle.
The in-season top-up
Placed early enough in the season that it can land while the season is still running. This is where the reorder minimum stops being a technicality: if a single design cannot be reordered on its own at a sensible quantity, the top-up is impossible and the site simply runs out. That interaction is the whole argument in the mix-and-match post and it is where ranges split too thin come apart.
Top-ups should be existing designs on existing tooling only. A season is not the time to introduce a design that still needs a sample.
Next season’s opening order
Placed while the current season is still running, because that is when you have the only data worth ordering on — actual sell-through by design and by site. Wait until the season closes and you have lost the runway; order before it starts and you are guessing.
What the sell-through data should change
Two decisions, and they are different:
Which designs to repeat. Straightforward: repeat what sold, drop what did not. The trap is judging a design by total units when it was only ever on one rack. Compare rate of sale per site, not totals.
Which destinations to add. A form that sold well at three sites is a candidate to roll out to more, and on a shared form that costs a print change rather than tooling. This is the moment the plate-change economics pay for themselves — you are extending something with till data behind it, not gambling on a new shape.
The two failure modes
Ordering on the empty shelf. Feels responsive, arrives late, and turns working capital into stock that waits out the off-season. The shelf being empty is information about the last season, not an instruction about this one.
Over-ordering the opening. The overcorrection. A season’s worth of stock ordered in one go removes the top-up risk and replaces it with a year of held inventory on designs that may not repeat. Splitting into an opening plus a planned top-up costs a little more per piece and is usually the cheaper outcome.
What to lock in before the first order
- The stage-by-stage timeline, not a single lead time
- The reorder minimum on existing tooling, which decides whether a mid-season top-up is even available to you
- Whether tooling is retained between seasons, and for how long — a range that runs annually needs the tools to survive the gap
- The cut-off date after which an order can no longer land before your season opens, stated by the supplier rather than assumed by you
That last one is worth asking for in writing every year. It moves with freight conditions and factory calendars, and the version you remember from last season is not necessarily the version that applies to this one.