A new location with 6,000 products has a planning problem that no forecast tool solves out of the box: there is no history. Zero sales, zero seasonality, zero signal. And yet on opening day, the shelves must be full enough to sell and lean enough not to drown the launch in dead stock.
Here is the structured model I used to turn that blank page into supplier-ready purchase orders.
You cannot forecast from nothing, so you do not. You pick the most comparable existing location and use its sales as the demand skeleton. Then you adjust honestly for what will differ: catchment size, customer profile, competition, expected ramp-up. The starting demand is deliberately an assumption, written down, so it can be corrected fast once real sales arrive.
Uncertainty is not equal across the assortment. Fresh products punish overstock through expiry. Long-life top sellers punish understock through lost sales and a bad first impression. So buffers are set per category: tighter where shelf life is short, more generous where a stockout in week one damages the launch. One global safety percentage is the most common and most expensive shortcut.
A demand number is not an order. Between the two sit constraints that break naive plans:
Launch planning is a master data stress test. The model has to validate product status, so paused and delisted items do not get ordered. It has to check existing purchase orders, so nothing is bought twice. Every SKU that fails validation goes to an exception list for a human decision instead of silently becoming a wrong order.
The output that matters is not a spreadsheet of target quantities. It is automated purchase orders, grouped by supplier, with correct case packs and delivery dates, plus a short exception list for manual review. That is the difference between an analysis and an operating system: the system ends in an action someone can execute.
The whole model stands on one principle: separate assumptions from mechanics. Assumptions, like reference demand and ramp-up, will be wrong and get corrected. Mechanics, like case packs, shelf life and duplicate prevention, must be right on day one.
If you are opening a location, resist the two default failure modes: copying another location's stock list one to one, or letting every category manager guess independently. Build one model with explicit assumptions, category buffers, hard constraints, validation, and automated output. Then let the first four weeks of real sales correct the assumptions, which is exactly what they are for.
Build one model with explicit assumptions, category buffers, hard constraints, validation and automated output, then let the first four weeks of real sales correct the assumptions.
The Cash Scan turns your own data into four documents in seven days: where your working capital is sitting, which SKUs are costing you sales, and your planned media budget translated into required units by week. $1,500, credited in full against the Sprint, and findings in seven days or you do not pay.
See what the Cash Scan includes Free calculator first