Stock record accuracy and the cost of mismatches
From a sample count it calculates record accuracy by lines and by value, the net balance of mismatches and estimates the lost revenue from “phantom” stock: an item is on record but not on the shelf.
At a glance
- What question does it answer?
- How far stock records match reality and what the mismatches cost.
- What data is needed?
- Result of a sample count: lines checked and accurate.
- How to read the result?
- Compare accuracy by lines with the value error: many lines with small errors point to receiving and sizes; few lines but large sums mean shortages and theft. Lost revenue shows the cost of phantom stock: the system does not replenish what is not on the shelf. Example on the default values: Accuracy by lines — 70% · goal 95%; Value error of the records — 9.4%; Net balance of mismatches — −RUB 225,000; Share of lines with phantom stock — 9.5%.
Why and how to reason
If goods are in the records but not on the shelf, they do not sell: the records “deceive” purchasing, sales staff and the online storefront. Checking the entire warehouse is expensive, so a sample reconciliation is done and estimated from the sample.
Line accuracy = accurate / checked. Value error = (surpluses + shortages) / book value: it accounts for the size of discrepancies, not only their number. Balance = surpluses - shortages. Phantom stock is lines where records show stock and the shelf has zero: lost revenue = such lines x pace x days to detection / 7 x price.
Example. Of 400 lines 280 are accurate: accuracy 70%. Surpluses 85, shortages RUB 310 k on book value RUB 4.2 m: error 9.4%, balance -RUB 225 k. Phantom lines 38 (9.5%). Pace 1.8 a week, detection after 21 days: lost 38 x 1.8 x 3 x 4,000 = RUB 820.8 k. The estimate for the entire stock per cycle is about RUB 13.3 m.
When to use it
It fits after a stocktake or sample reconciliation to assess the scale of the problem and make the case for addressing it. Suitable for stores and warehouses where balances are used for ordering. Assumption: the sample is representative. If only easy items were checked, the estimate is understated. It does not explain causes, which needs a breakdown (receiving, theft, mix-ups). Especially important for stores with automatic ordering and an online storefront.
How to read the result
Line accuracy of 70% means every third item has an error; for stores with automatic ordering that is serious. The 9.4% value error and negative balance speak of shortages. Lost revenue (RUB 0.82 m on the sample, RUB 13.3 m on the whole stock) is the price of phantoms, which is usually higher than the losses themselves. Decision: set up regular cycle counts, fix receiving and transfer processes, detect zeros on the shelf quickly. Determine which groups of goods give the most discrepancies: often new arrivals and items with many sizes. Focus checks on them. Track how accuracy changes after a process change.
Common mistakes
Typical trap: checking lines that were already checked; not counting surpluses and shortages together; computing lost revenue at full price when goods are discounted; taking days to detection “by opinion”. Also: extrapolating a sample to the whole stock without checking. Also: counting a line as accurate when the discrepancy is within a tolerance that was not agreed.
What data is needed?
- Result of a sample count: lines checked and accurate.
- Book value of the sample, surpluses and shortages at cost.
- Optional: lines “on record, zero on the shelf”, pace, days until found, price.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Lines checked | lines | 400 | A line: a style’s size and color in one location. |
| Of which accurate | lines | 280 | Record and count matched or differed within your tolerance. |
| Book value of the lines checked | ₽ | 4,200,000 | At cost. Needed for the value-based accuracy. |
| Surpluses (count above record) | ₽ | 85,000 | At cost. Empty: zero. |
| Shortages (count below record) | ₽ | 310,000 | At cost. Empty: zero. |
| Lines “on record, zero on the shelf” | lines | 38 | How many lines are on record but have zero on the shelf. Needed for the cost of phantom stock. Empty: 0. |
| Average sales pace of a line | pcs a week | 1.8 | The row’s average sales pace a week in periods when the item was in stock: from the till; lost sales from a record mismatch are counted from it. |
| Days until found | days | 21 | How long on average an item “exists” on record without selling. |
| Price excl. VAT | ₽ | 4,000 | Unit price excl. VAT: for the lost revenue on phantom lines. |
| Unit cost | ₽ | 2,000 | Unit cost excl. VAT: it values the mismatch in money; empty means the margin is not calculated. |
| All lines on record | lines | 6,500 | To estimate the loss across all stock. Valid if the sample is random. |
| Target accuracy by lines | % | 95 | The target record accuracy by line in percent: the share of lines where the records matched the count. The target is yours, there is no single standard in Russia; the example value is illustrative. |
How it is calculated
| Step | Formula |
|---|---|
| Accuracy by lines | accurate lines ÷ lines checked |
| Value error | (surpluses + shortages) ÷ book value |
| Lost revenue | lines with zero on the shelf × pace × days ÷ 7 × price |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Accuracy by lines
- 70 % · goal 95 %
- Value error of the records
- 9.4 %(surpluses + shortages) ÷ book value. Value accuracy = 100% minus this figure.
- Net balance of mismatches
- −225,000 ₽Surpluses minus shortages; a negative balance is a shortage.
- Share of lines with phantom stock
- 9.5 %
- Lost revenue in the sample
- 820,800 ₽ · margin 410,400 ₽
- Estimate for all stock per cycle
- 13.3 m ₽Revenue not earned while phantom lines wait to be found; recurs with each count cycle.
There are more errors by lines than by value: many small mismatches (sizes, colours). Look for receiving errors and mis-sorting.
Accuracy is below the goal: count fast-moving and expensive items (ABC class A) more often, and after receiving deliveries and stocktakes.
For labelled goods the stock on record and the stock of codes in the labelling system are two different records: reconcile them with each other. Check the procedure for taking codes not scanned at the sale out of circulation, and the liability, with the labelling operator and an accountant against the current rules.
Nuances and limits
- Do not count only the warehouse total: separate sizes and colors.
- Treat goods in transit, in fitting rooms and in returns alike in records and the count.
- The sample must be random, otherwise carrying it over to all stock is wrong.