Skip to content
Stock · 5 min read

How to count stock so the figure can be trusted

Why the counted figure differs from the recorded one, how to run the count, and what to do with the difference once you find it.

The short answer

A stock count compares what the books say with what is on the shelf, and the difference between them is a fact, not a fault. Stop movement before counting, count in the unit the item is stored in rather than sold in, record the counted figure before looking at the recorded one, and settle the difference with an adjustment whose reason is written down. Skip any of those and the resulting number is not one you can build on.

The steps

  1. Stop movement, even for an hour

    Counting while selling and receiving produces a figure that belongs to no moment: an item counted, sold, then counted again. Freeze receipts and issues during the count, or count outside trading hours. If neither is possible, log every movement that happens during the count separately so it can be added or subtracted afterwards.

  2. Count in the storage unit

    If an item is stored in cases and sold in units, count the cases and convert afterwards. Counting in the selling unit creates multiplication errors: a 24-unit case counted as 20 units means somebody opened it without recording it, and that information is lost if you count in units from the start.

  3. Count blind: do not show the recorded figure

    The count sheet does not carry the expected balance. The eye looks for confirmation: somebody who sees "48" printed counts to 48 and stops. A blind count surfaces the differences that a pre-filled sheet hides, and that is the practical difference between a count that finds something and a count that confirms itself.

  4. Explain the difference before you post it

    Differences have known causes: damage, theft, a receiving error, an issue that was never recorded, or two similar items mixed up. An adjustment posted with no reason hides the problem and brings it back next month. Write the reason on the adjustment itself, and the count becomes an operating tool rather than an annual ritual.

  5. Count the items that matter more than once a year

    One full annual count is exhausting and arrives too late. Cycle counting is better: high-value or fast-moving items monthly, the rest quarterly or twice a year. Leakage is then found within weeks rather than a year later, and the stock figures stay usable all year round.

Common questions

How is the counted quantity valued?
At cost, not at selling price. The costing method — weighted average or first in, first out — is chosen once and applied consistently, because changing it makes a comparison between two years meaningless.
What if the difference is very large?
Recount before adjusting. Large differences are usually a unit-of-measure mistake, an item counted twice, or a location nobody visited — not a real loss. Once confirmed, post the adjustment with its full reason.

Terms used in this guide

More guides

All guides