How to Control Inventory: Reorder Points, Safety Stock, Cycle Counts

What does inventory control actually mean?

Inventory control is the everyday discipline of knowing what you hold, holding the right amount of the right things, and being able to prove both. It is narrower than supply chain management and more practical than inventory accounting. A stock controller in a 40-person distributor, a stores lead in a factory and a manager running a clinic store with expiry dates all do the same job: they decide when to order, how much to order, where stock lives, how it moves and how its records stay true.

A useful image comes from the Toyota Production System: stock is water in a river and problems are rocks on the riverbed. High water hides late suppliers, bad forecasts, picking errors and muddled units of measure. The aim is not simply less stock. It is to see the rocks, remove them one at a time, and let the level fall safely. Control is a smaller pile you can trust, not a bigger pile.

Stock costs money three ways: the cash tied up, the cost of holding it (space, insurance, handling, damage, expiry), and the cost of not having it when a customer or a production line needs it. Good control holds enough of the few items that matter and little of everything else.

Why should you fix the item master before anything else?

Every forecast, reorder point, count sheet and shelf label reads from the item master, the single list of everything you stock. If the same gloves exist under two codes, or welding wire is held in kilograms in one place and spools in another, no amount of counting will make the numbers agree, because the items themselves are not defined.

Give each item a short code that never changes and never carries meaning that might change, such as supplier or price. Write descriptions noun first, then size, then variant, then pack. Choose one stock unit per item and keep pack size as a separate field. Record cost, supplier, lead time, bin and status. Where goods already carry a GS1 global trade item number (GTIN), use it; your own shelf and bin labels can use any barcode your scanner reads.

When cleaning an existing list, export it first and keep the export, group probable duplicates, agree one stock unit per item, and never delete a record: set it to phase out or discontinued so the history stays. An AI assistant can propose duplicate groups quickly, but a person who handles the stock confirms every merge.

How do ABC and XYZ analysis tell you where to focus?

ABC analysis applies the Pareto idea, which the quality pioneer Joseph Juran called the vital few and the trivial many, to stock. Rank every item by annual usage value (units used in a year times unit cost). Items making up the first 80% of value are A, the next 15% are B and the rest are C. The cut-offs are a choice; 80 and 95 are the classic split.

XYZ analysis adds a second question: how predictable is demand? Divide the standard deviation of monthly demand by its average to get the coefficient of variation. Low values are X (steady), middling are Y, high are Z (lumpy, driven by projects or one-off orders). Together the two classes suggest a policy for each item.

• Read the A list with sales and purchasing: a cheap C item can be critical if a customer's machine cannot run without it
• Mark such items critical and give them A-class safety stock

How do you calculate a reorder point and safety stock?

The reorder point answers when to order. It is the stock you need to cover demand during the supplier's lead time, plus a cushion for surprises: average daily demand times lead time in days, plus safety stock. You order when available stock plus stock already on order falls to that level.

Safety stock is the cushion. The textbook version multiplies a z-factor, taken from the service level you want, by the standard deviation of demand over the lead time. A better version also includes lead time variability, because a supplier who delivers in anything from 8 to 12 days adds as much surprise as a customer whose orders wander. In words: z times the square root of (lead time times daily demand variance, plus the square of daily demand times lead time standard deviation).

A worked example. An item sells an average of 300 units a month over 25 working days, so daily demand is 12. The monthly standard deviation is 60, which scales to 12 a day (60 divided by the square root of 25). Lead time averages 10 days with a standard deviation of 2. For a 95% service level, z is 1.64. The variability term is the square root of (10 x 144 + (12 x 2) squared), the square root of 2,016, about 44.9. Safety stock is 1.64 x 44.9, about 74 units. The reorder point is 12 x 10 + 74 = 194 units.

Be honest about the assumptions: roughly normal demand, independent from day to day, with a stable average. That holds well enough for X and many Y items. For lumpy Z items the formula can be badly wrong, and judgement or ordering against firm orders does better.

How much should you order each time?

The economic order quantity, first described by the engineer Ford W. Harris in 1913, balances the cost of placing an order against the cost of holding what you ordered. The formula is the square root of (2 x annual demand x cost per order) divided by (unit cost x annual carrying rate). Continuing the example: annual demand of 3,600 units, a cost of 1,500 per order, a unit cost of 200 and a carrying rate of 22% give an EOQ of about 495 units.

EOQ assumes steady demand, a fixed cost per order, no discounts and no pack sizes. Real suppliers sell in cartons and set minimums. The useful insight is that the total cost curve is flat near the optimum, so rounding 495 up to a carton quantity of 500, or even a pallet of 600, costs very little. Bulk discounts on C items are a different story: a large order can tie up far more cash and space than the discount saves.

Not every item needs continuous review. Min-max sets min at the reorder point and max at min plus the order quantity. Periodic review tops items up to a target on a fixed cycle and needs extra safety stock to cover the review period. Two-bin systems suit cheap, steady items. Erratic items are often best ordered on demand.

How do lead times and suppliers affect stock levels?

A reorder point is only as good as the lead time behind it. Record the promised and actual delivery dates for every purchase order line, and update each item's lead time and its variability from that record every quarter rather than from the supplier's brochure. The on-time rate by supplier becomes one of your monthly numbers.

Peter Kraljic's 1983 Harvard Business Review article argued that purchasing should sort items by their importance to the business and the difficulty of the supply market. In stock terms: make cheap, easy items effortless to buy, secure supply of cheap but hard-to-get items, and build relationships and second sources for expensive, hard ones. Total cost of ownership is the companion idea: the cheapest unit price is rarely the cheapest item once late deliveries, rejects and chasing are added.

How does cycle counting improve inventory accuracy?

An annual stocktake counts everything at once. It may be required for your accounts and it is the right way to reset records nobody trusts, but as a control it is weak: a year of errors arrive in one weekend after the causes have gone cold. Cycle counting counts a few locations every working day or week, A items most often, so the whole stock is checked across the year and every variance is investigated while the trail is fresh.

The method matters more than the frequency. Print count sheets blind, without the system quantity, so the count is not a confirmation. Pause movements in the zone or list open documents. Compare in a quiet room, recount every line outside tolerance with a different person, then investigate: receipts, picks, transfers and returns since the last count, unit of measure and location. Only then raise an adjustment, approved by someone whose authority matches its value, and posted by someone other than the counter where staffing allows.

• Typical frequencies: A monthly, B quarterly, C twice a year; adjust to your risk
• Measure accuracy by location within tolerance, not by net value, because two errors that cancel are still two errors
• Never accept 'count error' as the cause of a variance; send it back for a recount and a look at the documents

Which inventory KPIs should you track each month?

Six numbers, read together once a month, tell you whether control is improving. Read them as a set: fill rate falling while stock value rises means the wrong items are in stock; good accuracy alongside customer complaints points at picking and labelling rather than counting.

Where do AI, valuation and safety fit in?

AI assistants and the forecasting or anomaly features in some inventory systems can spot seasonality, rank forecast error, draft supplier notices and flag unusual movements such as adjustments posted out of hours. They also produce confident wrong numbers. A workable rule: AI suggests, a person who knows the customers and the warehouse decides every order, reorder point change, adjustment and write-off, and any flag about a possible loss is treated as a question until the documents are checked.

Valuation is your accountant's job, but the warehouse supplies the inputs. Under the international standard IAS 2, inventories are measured at the lower of cost and net realisable value, with FIFO or weighted average cost for interchangeable items. Rules for your accounts and tax differ by country, so keep unit costs current, keep slow and damaged stock visible, and check your accountant before writing anything off.

Warehouses are also workplaces with vehicles, racking and heights. The UK Health and Safety Executive's provisional statistics for 2025/26 record 126 workers killed in work-related accidents in Great Britain, with 24 struck by a moving vehicle and 21 struck by a moving object. In the United States, OSHA's powered industrial truck standard requires operators to complete training and evaluation before they drive. A weekly housekeeping and safety walk keeps bins countable and aisles safe; check your national safety regulator and a qualified safety professional for the rules that apply to you.