Test Count
A test count is an independent count of selected items performed by an auditor to evaluate whether the client's own counting procedure is producing reliable results. The auditor counts the items personally and compares the result with what the client's team recorded for the same items. It tests the process rather than the population, so it supports a conclusion about the whole count rather than replacing it.
What Is a Test Count?
An auditor attending a physical count is not there to count the inventory. The counting is management's responsibility and management's staff perform it; the auditor is there to form a view on whether the procedure being followed is capable of producing a reliable result. Selecting items and counting them personally is how that view is obtained, and the evidence is how that result compares with whatever the entity's own staff wrote down against those same items.
The procedure is deliberately two-directional. Items are selected from the count sheets and located physically, which tests whether recorded quantities exist, and items are selected from the floor and traced to the sheets, which tests whether everything present was actually counted. A one-directional test cannot detect stock that was simply missed. Where differences emerge, the significance is not the individual error but what it says about the process, and a pattern of differences will usually lead to the count being extended or repeated rather than adjusted.
What Test Count Means at a Physical Count
For the attending auditor this is the day's core work, and it is deliberately limited.
- Observing before counting anything, watching how sheets are issued, how areas are allocated and whether anybody is checking.
- Selecting items personally rather than accepting a list, weighted toward value and toward anything awkward to reach.
- Counting each one without reference to what the sheet says, then comparing afterwards, since seeing the expected figure first shapes what a person counts.
- Running the same exercise in reverse from the shelf to the paperwork, which is what finds areas the teams skipped.
- Recording every difference individually, because the conclusion being formed is about whether the procedure works rather than about the items themselves.
How Test Count Works in Practice
- The auditor is present while the entity's team works, and begins by watching how the exercise is being run rather than by counting.
- Entries are drawn from the finished sheets and the corresponding goods located, testing whether the recorded figures describe anything real.
- The exercise is then run the other way, lifting goods off the rack and hunting for them in the paperwork, which exposes anything skipped entirely.
- Each chosen item is counted by the auditor personally, and that figure is set beside whatever the entity's own team wrote against it, with every difference noted separately.
- The conclusion drawn is about the procedure rather than the population. A clean result supports reliance on the wider count; a pattern of differences leads to the count being extended, repeated, or reported as unreliable, rather than to the individual items simply being corrected.
Test Count: A Worked Example
| Auditor sample | Book | Counted | Variance | Extrapolated? |
|---|---|---|---|---|
| High-value items, 40 lines | Rs 3,10,00,000 | Rs 3,08,40,000 | -0.52% | No, 100% of stratum |
| Random sample, 60 lines | Rs 90,00,000 | Rs 88,20,000 | -2.00% | Yes, to Rs 4.20 crore |
| Projected shortfall on the untested balance | - | - | - | Rs 8,40,000 |
An auditor attending a Jaipur count at a jewellery manufacturer does not recount the floor. The client's team counts; the auditor independently checks a sample and forms a view on the whole.
Two strata behave differently and are therefore treated differently. Every high-value line is checked, so its 0.52% variance is a fact about that stratum and is not projected anywhere. The random sample is a sample in the statistical sense, so its 2% variance is extrapolated across the untested balance to give a projected shortfall of Rs 8.40 lakh. That projection, not the sample result, is what the auditor weighs against materiality. A test count that finds a variance the client cannot explain generally leads to a full recount rather than a larger sample.
Common Mistakes With Test Count
The purpose of the procedure is misread more often than the procedure is performed badly.
- Expecting the auditor to count the inventory, when counting belongs to the entity and the auditor is judging whether its method holds up.
- Selecting only accessible, well-labelled items, which are systematically the ones least likely to be wrong and tell you nothing about the rest.
- Running the check in one direction only, so items that were never written onto a sheet at all remain invisible.
- Treating a difference as an item to be adjusted rather than as evidence about the process. The individual error matters far less than what it says about the count as a whole.
- Continuing with the count after a pattern of differences has emerged, when the sensible response is to stop and recount.
Need Help With Test Count?
Terminology takes you only as far as the question. Where the counting procedure itself needs testing rather than the stock, what follows is attendance, testing and a reconciliation, described under inventory audit. Share the site list and the records, and the effort can be sized from them.