Demand forecast
Demand forecast projects your stock forward. The screen puts it plainly: “What your stock is likely to do over the next N days, from your own paid orders.”
Two choices shape it: the History window it learns from, and the Plan ahead horizon it projects into.
Reading a row
Section titled “Reading a row”| Column | What it is |
|---|---|
| In stock | What you have now. |
| Sold | How much went in the history window. |
| Per day | That, as a daily rate. |
| Forecast | What the horizon is likely to consume. |
| Cover | How long what you have will last. |
| What to do | The suggestion. |
Cover is the column to read first. In stock tells you a quantity; cover tells you a deadline — and a deadline is what a purchasing decision actually needs.
The four suggestions
Section titled “The four suggestions”- Reorder — cover runs out inside the horizon.
- Watch — close enough to matter.
- Well stocked — comfortable.
- No signal — not enough sales to say anything.
🔑 “No signal” is the honest one, and it is not a bug. A product that sold twice in ninety days has no rate worth projecting. The screen says so instead of manufacturing a confident number from two data points, and that restraint is worth more to you than a forecast you cannot trust.
It is built from paid orders
Section titled “It is built from paid orders”Not carts, not pending payments — paid orders. That is the right basis for buying stock, and it explains two things you might otherwise find odd:
- A launch day of unpaid pre-orders moves nothing.
- If nothing sold, the screen says “No paid orders in the last N days, so there is nothing to forecast from yet” rather than showing zeros.
There is a second empty case worth knowing about, because it looks like a bug and is not: if everything that sold belongs to variants that no longer exist, the screen says so. Your history was real; the things it was about have been deleted.
A forecast assumes the future resembles the past
Section titled “A forecast assumes the future resembles the past”The arithmetic cannot know about your Black Friday, your supplier’s holiday, or the influencer who is about to mention you. It extrapolates the window you chose.
The practical habit: choose the history window to match the future you expect. A ninety-day window through a quiet season will under-forecast a busy one.
What this page does not say
Section titled “What this page does not say”- It does not order anything. Every suggestion is a suggestion; purchasing happens with your supplier, not here.
- It does not account for lead time. “Reorder” tells you cover is running out, not that ordering today arrives in time. Only you know how long your supplier takes.
- It does not say which history windows or horizons are available, or whether they can be saved as a default. Neither has been checked against the current screen.