How inventory accuracy is calculated

After a stock count the most common question is: "How accurate are the balances in our warehouse?" The answer should be a single percentage. Inventory accuracy gives you that figure, and you do not need complicated software to calculate it — count results and a few agreements are enough. The key is to define those agreements clearly from the start.

The basic formula

Inventory accuracy = (number of positions that matched the records ÷ number of positions counted) × 100.

A position is one catalogue unit, that is, one SKU. Example: at a small wholesale warehouse, 400 positions were counted. In 372 of them the balance in 1C was the same as on the shelf; 28 showed a difference. Accuracy = 372 ÷ 400 × 100 = 93 percent. Put differently, roughly one position in every fourteen or fifteen differs from the records.

The divisor is the positions actually counted, not the whole catalogue. If 400 out of 5,000 SKUs were counted, the 93 percent applies to those 400 only. To draw conclusions about the rest, the sample must be random, not made up of convenient shelves near the entrance.

What "matched" means

This is the most important agreement, because the percentage depends on it. There are two ways:

• Strict: a position counts as matched only if the records and reality are exactly equal. Right for medicines, electronics and other expensive goods.
• A tolerance: for example two percent or three pieces. For loose and weighed goods — flour, sugar, cement, nails — this is practical, because spillage and scale error produce a natural difference.

Example: 1C shows 500 kg of rice and the count found 497 kg. Under the strict rule that is a mismatch. With a one percent tolerance (5 kg) the position matches. Whichever rule you choose, write it down and apply it the same way every time, otherwise results from different months cannot be compared.

Accuracy by value

Counting positions does not show the money: a cheap nail that is one piece short and a phone that is three pieces short are both just "mismatched". That is why a second measure is needed.

Accuracy by value = 100 − (sum of absolute differences ÷ book value of stock × 100).

Example: book stock is 480 million so'm. Shortages total 9 million and surpluses total 5.4 million so'm. The absolute sum is 9 + 5.4 = 14.4 million. 14.4 ÷ 480 = 3 percent, so accuracy is 97 percent.

Note: do not net surpluses against shortages. Otherwise you get 9 − 5.4 = 3.6 million, only 0.75 percent, and accuracy looks like 99.25 percent. In reality one item is short and another is over — similar SKUs were swapped — and netting gives false comfort.

Calculating it in Excel

Build a table: SKU in column A, book balance in B, count result in C. In column D mark a matching position with 1 and a non-matching one with 0. With a two percent tolerance the formula is IF(ABS(C2−B2) ≤ 0.02 × B2, 1, 0). For the strict method: IF(C2=B2, 1, 0). Then accuracy = SUM(D2:D401) ÷ COUNT(C2:C401). The separator in formulas (comma or semicolon) depends on your Excel settings.

In the same table add column E and calculate the value of each difference: difference × unit cost. Then both accuracy figures sit on one sheet.

Common mistakes

• Movement was not stopped during the count: goods counted in the morning were shipped in the afternoon, and a difference "appears".
• Counting only fast-moving goods: the result is unreliable for the whole warehouse.
• Dropping zero balances: this is where "zero in the records, stock on the shelf" cases hide most often.
• Not recording the cause: there is a percentage but no basis for correction.
• Ignoring location: the item exists but sits in another bin, so picking will not find it anyway.

Measure accuracy regularly, not once: expensive and fast-selling goods can be counted often, the rest less often.

Conclusion

Inventory accuracy is a simple ratio, but to be trustworthy it needs an agreed rule for "matched", a proper sample and a second measure by value. When the percentage is read together with its causes, it becomes a practical decision. CountLab helps work out accuracy from stock-count results, both by position and by value, and show the causes of differences by category and zone.