Stock Control for Retail Chains: The Right Stock in Every Store

Stock control in a chain of stores means the right stock in each store: not too little, not too much, and a stock record that matches what is on the shelf. Each of the three goes wrong in its own way and has its own arithmetic. This page puts them in one order for an operations or buying lead: one question per section, the rule that answers it in a few sentences, and the full guide behind it.

Stock control in a chain is a store-level job

A chain's stock report adds every store together. The total can look healthy while one store's shelf is empty and another holds a year of a product nobody buys there. So every check below runs per product and per store; the chain total is where these problems hide, not where they are found.

The questions follow the order in which the problems cost money: the empty shelf first, then the orders that refill it, then the stock that does not move, then the counts that say whether any of the figures can be trusted.

1. Where are the shelves empty?

Start with zero stock on a product that sold in that store in the last few weeks, and price each line by what the product takes on a day it sells, so the list is read from the top. Then look for the empty shelves the record hides: units on the record and a run of days without a sale, longer than the product's pace allows. In a study of a European drugstore chain, records showing more than the shelf held were the more frequent error in every store¹. Two more checks complete the list: a product that sells everywhere in the chain except one store, and a line with fewer days of supply than days to the next delivery. The out-of-stock detection guide sets out all four.

2. Why sales history lies after a stock-out

The till records y = min(D, I): the demand on a day the shelf holds out, and only the stock on a day it does not. Averaging those days in pulls the sales pace and its spread down; the forecast learns the shelf instead of the customer, the next order is smaller and the shelf empties sooner. People do the same by hand: in a supermarket chain with automated ordering, store managers cut the proposed order more often after a stock-out, which made the next one more likely². The rule is to leave empty-shelf days out of the average, or treat them as a floor. The guide to why empty-shelf days are not zero demand shows how the error feeds itself.

3. How much buffer to hold

An order has to last the lead time L plus the wait R until the supplier's next order day. Safety stock covers two uncertainties over those days, the spread of daily demand σ and the spread of the lead time σ_L: safety stock = z·√((L + R)σ² + μ²σ_L²), where μ is daily sales and z sets the service level. One buffer factor for every supplier is too thin for a nearby supplier ordered once a week, and an erratic supplier can double the buffer for the same product. Measure lead times from your own receipts, because planned lead times tend to understate the real ones³. The safety stock guide works through three supplier patterns.

4. When to order, and how much

A days-of-cover rule such as "below seven days, order up to fourteen" ignores the lead time, the wait to the next order day and how erratic demand and deliveries are. The periodic-review formula replaces it: on each order day, bring the inventory position (stock on hand plus stock on order) up to S = μ(L + R) + safety stock. It needs four inputs: sales pace without empty-shelf days, lead time from your own receipts, each supplier's order days, and a service level per product class. The reorder point guide works it through for a nearby supplier and an importer.

5. Who decides the order: rules, people or AI

Quantities, timing and the arithmetic of cover belong to rules; supplier choice, large or unusual orders and demand the history has not seen belong to people. Overrides are neither good nor bad on average: cutting an order right after a stock-out tends to cause the next one, while raising it ahead of an event the system cannot see tends to help. Software that prepares orders should earn authority in steps, advice first and drafts for approval next, and its effect should be measured against comparable stores where it only advises. The guide to AI store replenishment covers the field evidence. In marql, the AI supply employee is a pilot, and every order or transfer it prepares is a draft that a person approves.

6. Stock that does not move

Dead, slow and excess stock need different actions, so sort before acting. Days of cover, stock on hand divided by average daily sales over the last 90 days, per product and per store, separates them: dead stock has not sold for long enough that the silence is conclusive, slow stock sells but would last months or years, and excess stock sells well but sits above its target. Count the shelf before marking anything down, because a product nobody can find also shows no sales. Then move stock to stores where it sells, stop reordering slow and excess lines, and return or mark down only what is dead everywhere. The dead stock guide has the four checks.

7. How fast stock moves overall

Stock turnover = cost of sales ÷ average stock at cost, and days of stock = 365 ÷ turnover. Both sides must be at cost, because stock is carried at cost⁴; dividing sales by stock at cost overstates the turn. A chain's turnover is a weighted total, so read it store by store, and never read a high turn without checking the shelf: empty shelves raise it too. The stock turnover guide has a calculator and category benchmarks.

8. Counting and shrinkage

Every figure above rests on the stock record, and the record drifts from the shelf. Retailers put their shrink down to external theft, their own employees, and process and control failures⁵. A stock take counts what is there and corrects the record; the difference written off is shrinkage. Counting goes further when the data picks the lines: in the same drugstore study, counting the products a zero-sales model ranked highest found many of the hidden empty shelves while covering only a small share of the range¹. The stock take guide covers the count itself.

9. When the books and the shelf disagree

A total shortage figure says almost nothing on its own. Split it by location, period, shift and product category, then test each explanation against the shape it would leave: a system-wide calculation error spreads across sites, shifts and categories, while loss from misconduct concentrates in compact, valuable items on particular shifts. The breakdown points the investigation; documents, counts and conversations prove it. The essay on how to investigate an inventory shortage follows one case through.

10. The KPIs to watch

Four measures from this path belong in the regular review, each by store: availability, from the empty-shelf checks; days of cover, with the value of dead and slow stock; stock turnover; and shrinkage from the last counts. Sell-through rate, the share of the stock received that sold in the period, shows how a range or a delivery is moving. The guide to retail KPIs sets out the wider list, and the sell-through rate guide the formula.

One routine for the whole chain

The pieces fit into one routine, run at different speeds. Check the empty shelf first and every day: it loses sales today, and an empty day left in the history makes the next order too small. Orders follow on each supplier's order day, stock that does not move once a week, turnover and the KPIs once a month, and the counts underneath all of it.

The order matters as much as the checks. An order built on a sales history full of empty days is too small however good its formula, and a dead-stock list built on a wrong record marks down stock that was never there. Fix the shelf, then the order, then the excess, and count where the data says the record is weakest.