Replenishment is the process of reordering stock so you can keep selling without over-buying — deciding what to order, how much, from which supplier, and when. It sounds like admin; done across thousands of products with MOQs and a finite budget, it’s the hardest maths problem in the building.
Replenishment, in plain English
Stock goes out of the door; something has to decide what comes back in. That something — a buyer with a spreadsheet, an ERP reorder module, or an optimisation engine — is your replenishment process, and its quality decides both your service level and how much cash is buried in the warehouse.
The main methods, honestly compared
- Reorder point (continuous review): each product has a trigger level; stock falls to it, you order. Simple and responsive — but the trigger is only as good as the demand and lead-time numbers behind it, and most were set years ago.
- Periodic review: every week or month, review everything and top up to a target. Fits supplier order cycles nicely; between reviews you're blind, so the targets need bigger buffers.
- Min/max: order back up to max when stock hits min. Easy to run, crude in effect — the interesting question was always "what should min and max be?", and static answers age badly.
- Forecast-driven replenishment: order what the demand forecast says you'll need, when you'll need it, with buffers sized from forecast accuracy. The most precise method — and only as good as the forecasts, which is why forecast transparency matters.
What separates good replenishment from guesswork
- It works per product, not per policy. A best-seller and a long-tail line shouldn't share one rule.
- It respects real-world constraints. MOQs, pack sizes, supplier order cycles and container fills aren't footnotes — a recommendation you can't place is noise.
- It answers the budget question. The real weekly problem isn't "what does each product need?" — it's "I have £X; what's the best way to spend it across everything?" Most tools dodge this entirely.
Optimal Chain is a forecast-driven replenishment engine with the constraints built in: every product gets its own forecast (18 methods compete, you see the scores), buffers sized from real forecast accuracy, and — the part we haven't found elsewhere in this bracket — you give it your actual purchasing budget and it plans the orders to the penny, every quantity a valid MOQ multiple, the most urgent revenue protected first.