Compute each location's need, not a network average
The core figure is days of cover: how long the current stock lasts at the current rate of sale. Everything else is ranked by it, and the order quantity is derived from it.
What is wrong today
- Orders are placed “as usual”, unrelated to a location's actual cover.
- Write-offs and returns land in sales statistics and inflate the next order.
- Slow movers drop out of the calculation while sitting on the shelf.
How it works
Write-offs, returns and transfers are excluded. Not a hypothetical: in one network a write-off document counted as a sale, so the system kept reordering yesterday's bin contents.
One supplier delivers in three days, another in two weeks — the need follows that, not a network-wide norm.
Pack multiples and minimum lots are the supplier's constraint. If the lot exceeds a week of demand, that is visible before the order goes out.
What it looks like in use
Sorted by cover: what runs out first sits on top, not what costs more.
| Item | Location | Daily sales | On hand | Days of cover | State |
|---|---|---|---|---|---|
| Drinking water 0.5 l | Location A | 34 | 410 | 12 days | fine |
| Drinking water 0.5 l | Location B | 51 | 96 | 2 days | will not make it |
| Wet wipes, 60 pcs | Location C | 8 | 640 | 80 days | overstock |
| Toothpaste 100 ml | Location D | 17 | 205 | 12 days | fine |
An anonymised example: locations are lettered and the dataset is shared across every screen on this site.
What it is built on
- Dead stock and the cash in it are computed per location today — the same calculation, minus the expiry date.
- Transfers between locations already work: the mechanism is ready, what is missing is the trigger to fire it in time.
- Write-offs are already excluded from demand — otherwise an expiry product would reorder exactly what it just wrote off.
Who stays in control
Any line can be edited by hand, and the edit is stored with its author and timestamp. A large order can require explicit confirmation — then it never leaves on its own.
How it is measured
By comparing the system's calculation with what a person would have ordered over a parallel period: where it matched, where it did not, and why. Then by cover and the share of lines needing no manual edit.
Show us one critical process. We will show how it runs here.
We look at your cycle: how an order is assembled today, who decides, where time leaks and what the system takes over.