Reorder points from your real sales (no crystal ball required)

Every stockout has the same postmortem: “we should have ordered sooner.” The fix isn’t a forecasting oracle - it’s a reorder point: the stock level at which, if you order today, the delivery arrives before you sell out. Boring, mechanical, and worth more to most stores than any AI forecast.

The mechanics

A reorder point is three numbers multiplied and added:

reorder point = (daily sales velocity x supplier lead time) + safety stock

Sell 4 a day, supplier takes 10 days, keep 12 as a buffer: reorder at 52. When stock touches 52, you order - and the new stock lands as the buffer starts to bite. That’s the whole idea. The craft is in the inputs.

Input one: velocity (and the trap inside it)

Daily velocity sounds simple - units sold divided by days. The trap: days you were out of stock count as zero-sale days, and they drag the average down precisely for your best sellers.

Sell 6 a day for three weeks, then sit out of stock for a week, and a naive 28-day average says you sell 4.5 a day. Set your reorder point from that and you’ll order less, stock out again, and watch the average sink further - a doom loop where the more you run out, the less the system thinks you need.

The fix is to compute velocity from days you could actually sell. That’s how Stockroom does it: out-of-stock days are excluded from the denominator, so a stockout doesn’t teach the system that demand fell.

Input two: lead time (measured, not remembered)

Ask anyone their supplier’s lead time and you’ll hear the promise (“about two weeks”). Your receiving history knows the truth - the gap between ordering and receiving, order after order. Use the measured number, and when a supplier’s real lead time is 19 days against a promised 14, price that honesty into your buffer. Stockroom surfaces observed lead times per supplier from your actual POs for exactly this reason.

Input three: safety stock (a decision, not a formula)

Safety stock is where judgment belongs. Textbook formulas exist, but a working shortcut is to hold a few extra days of velocity - more for bestsellers you can’t afford to miss, for volatile sellers, and for unreliable suppliers; less (or none) for slow movers where cash matters more than availability. The important part is that it’s your call per product, not a global percentage something imposed on you.

Two constraints the math must respect

Real purchasing has rules the arithmetic doesn’t know:

• Pack sizes - suppliers sell in cases of 12, so a suggestion of 17 units is really 24.
• Minimum order quantities - below the MOQ, there is no order.

Suggestions that ignore these get “corrected” by hand every time, and hand-corrected systems stop being trusted. Set them once per product and let every suggestion arrive already rounded to reality.

Explainability is the feature

Whatever tool computes your levels, insist on seeing the working. A reorder point you can’t decompose - this velocity, times this lead time, plus this buffer - is a number you’ll override on instinct within a month. Stockroom attaches an explain dialog to every suggestion for exactly this reason, and every suggested level is yours to accept, edit or ignore.

No machine-learning mystique, no black box - measured velocity, measured lead times, and buffers you chose. That’s a reordering system you’ll still be using in a year.

If you’re setting this up for the first time: start with your top 20 products by revenue, accept the suggested levels, and let low-stock alerts catch the rest. You can tune the long tail as it earns attention.