Every stock decision reduces to two questions: when should I order, and how much should I order? Answered by feel, they produce the two failures small businesses oscillate between — running out of a component mid-production, and discovering EUR 4,000 of material that has not moved in a year.
Both questions have well-established answers. They need a handful of numbers you already have and one genuine business decision about how often you are willing to run out.
When to order: the reorder point
The reorder point is the stock level that triggers a purchase order. It has to cover everything you will consume between placing the order and the delivery arriving, plus a buffer for the days that do not go to plan.
Reorder point = (average daily demand x lead time in days) + safety stock
The first term is lead time demand. If you use 40 units a day and your supplier takes 10 days, you will consume 400 units while waiting. Order with 400 in stock and you arrive at zero exactly as the delivery lands — assuming demand is exactly average and the supplier is exactly on time, which is the assumption safety stock exists to relax.
Get lead time from your own purchase history, not the supplier's website. Measure from the date you place the order to the date goods are usable — not received. If material needs inspection, curing, or decanting before it can enter production, that time is part of the lead time.
How much buffer: three ways to size safety stock
1. Fixed days of cover
Safety stock = average daily demand x days of cover
Pick a number of days — commonly 7 to 14 — and hold that much. Crude, but transparent and much better than nothing. Reasonable for low-value items where the cost of over-holding is trivial.
2. Maximum minus average
Safety stock = (max daily demand x max lead time) - (average daily demand x average lead time)
This sizes the buffer to the worst case you have actually seen. It needs no statistics, only records. It tends to over-hold, because it assumes peak demand and worst lead time coincide, which they occasionally do and usually do not.
3. Service level based
The proper version. You choose the probability of not stocking out during a replenishment cycle, and the formula sizes the buffer for it, accounting for variability in both demand and lead time.
Safety stock = Z x √( lead time x demand variance + average demand² x lead time variance )
Where Z is the service factor for your chosen service level:
| Service level | Z | Expected stockouts per 20 cycles |
|---|---|---|
| 85% | 1.04 | 3 |
| 90% | 1.28 | 2 |
| 95% | 1.65 | 1 |
| 97.5% | 1.96 | 0.5 |
| 99% | 2.33 | 0.2 |
Service level is a commercial decision, not a technical one. Going from 95% to 99% raises Z by 41%, and safety stock with it. That is justified for the component that halts your whole production line, and wasteful for a decorative ribbon with three alternative suppliers. Set it per item, not once for the business.
A worked example
A workshop consumes a moulded component. From twelve months of records:
- Average daily demand: 40 units
- Standard deviation of daily demand: 12 units
- Average lead time: 10 days
- Standard deviation of lead time: 2 days
- Target service level: 95%, so Z = 1.65
Safety stock:
| Step | Calculation | Result |
|---|---|---|
| Demand variability term | 10 x 12² = 10 x 144 | 1,440 |
| Lead time variability term | 40² x 2² = 1,600 x 4 | 6,400 |
| Combined | √(1,440 + 6,400) = √7,840 | 88.5 |
| Safety stock | 1.65 x 88.5 | 146 units |
| Lead time demand | 40 x 10 | 400 units |
| Reorder point | 400 + 146 | 546 units |
The instructive part is the split. Lead time variability contributes 6,400 of the 7,840 — 82% of the total. Demand variability contributes 18%.
So the highest-leverage action here is not forecasting demand better. It is making the supplier more predictable. Cutting the lead time standard deviation from 2 days to 1 drops safety stock from 146 to about 87 units, a 40% reduction in buffer inventory from one supplier conversation. Run this decomposition before investing in forecasting; for most small businesses the answer points at the supplier.
How much to order: economic order quantity
Ordering frequently keeps inventory low but multiplies ordering cost — the admin, the goods-in handling, the per-delivery carriage. Ordering rarely does the opposite. EOQ finds the balance.
EOQ = √( 2 x annual demand x cost per order / annual holding cost per unit )
Continuing the example: annual demand 10,000 units, EUR 45 to place and receive an order, EUR 2.40 to hold a unit for a year.
EOQ = √(2 x 10,000 x 45 / 2.40) = √375,000 = 612 units
So: order 612 units whenever stock falls to 546. That is roughly 16 orders a year, an order about every 15 working days.
Holding cost is the input people get wrong. It is not just warehouse space. Include the cost of capital tied up, insurance, obsolescence risk, shrinkage, and handling — typically 18-30% of unit value annually for a small business. Understate it and EOQ tells you to order far too much.
EOQ is also gloriously insensitive: because of the square root, being 25% off on the order quantity raises total cost by only about 2.5%. Do not agonise over precision. Use it to land in the right order of magnitude, then adjust for reality:
- Supplier minimums that exceed EOQ — take the minimum, and reconsider the supplier if it is wildly larger.
- Price breaks — compare the discount against the extra holding cost over the period the excess sits there.
- Shelf life — never order more than you will consume within the usable life, whatever EOQ says.
- Pack sizes — round to whole cases or pallets rather than splitting.
- Cash — EOQ ignores your bank balance. A cash-constrained business rationally orders smaller and more often, accepting higher total cost to keep working capital free.
Do not do this for every item
Classify first. In most inventories about 20% of items account for around 80% of value, and those are where the effort pays.
| Class | Share of value | Approach |
|---|---|---|
| A | ~80% | Service-level safety stock, EOQ, review monthly, count often |
| B | ~15% | Simple safety stock, review quarterly |
| C | ~5% | Fixed days of cover, order in bulk, count annually |
Running the full statistical treatment on C-class screws is a hobby, not stock control. Getting it right on the six A-class items that stop production is the whole return.
Keeping the numbers alive
Reorder points decay. Demand shifts with seasons and product mix, suppliers change, and a point calculated in January quietly stops being right by June. Review A-class items monthly and recalculate whenever average demand moves more than about 20% or a supplier's lead time changes.
This is where the calculation being done in a spreadsheet becomes the limiting factor. The formulas need current demand history, current lead times, and current stock on hand, and a manual file has none of those without someone maintaining it. When purchase receipts, production issues, and sales all move stock through one ledger, demand rates and lead times are a by-product of ordinary work, and the reorder point can simply raise a flag when it is crossed. Deciding what to hold is the judgement; noticing you have crossed the line should not require any.
One caution worth stating plainly: these formulas assume roughly stable, roughly normal demand. They are poor guidance for a brand new product with no history, a highly seasonal line, or a promotional spike. For those, plan from the sales forecast and the production plan instead, and let the statistical approach take over once you have real history.