Auditing Inventory: Observation Procedures and Common Pitfalls

8/11/2026

The most common misunderstanding among first-year staff sent to an inventory count is about whose job it is.

You are not counting the inventory. The client counts the inventory. You are evaluating whether the client's count procedures produce a reliable result — and testing enough of it to conclude on that.

Which changes what you do on the floor. An auditor who spends the day counting boxes has generated a small amount of evidence about a few items. One who watches how the count is being performed, tests in both directions, and notices that two teams are counting the same aisle has evidence about the whole population.

Before the Count — Where Most of the Value Is

Obtain and read the client's count instructions in advance. The single most common failure in this area is arriving without having read them, or without them existing. Read for: how areas are assigned, how teams are composed, whether a second count occurs, how tags or sheets are controlled, how movement is handled during the count, and who resolves differences.

If the instructions are inadequate, say so before the count, when it can be fixed. Afterward it is a finding rather than a correction.

Select the locations to attend on materiality and risk, not on convenience or travel cost. New locations, locations with prior differences, locations with weak controls, and locations holding high-value or easily moved goods rank above the warehouse nearest the office. Document why you chose what you chose.

Understand the inventory itself. What it physically is, how it is measured, whether identification requires expertise, and whether any of it is held for others or held elsewhere.

Review the prior year — the differences found, the adjustments made, and the recurring problems.

Arrange the cut-off — that receiving and shipping will be halted during the count, or that movements will be documented and controlled. This is the arrangement most often left vague and it undermines everything if it fails.

On the Floor

Observe the count being performed before doing anything else. Are the instructions being followed? Are teams counting independently or copying? Is anyone counting an area twice? Are the counters people who know what they are looking at?

Make your test counts in both directions, and this is the point juniors most often miss:

From the count sheet to the floor tests that recorded items exist and are correctly quantified.

From the floor to the count sheet tests completeness — that items physically present were actually recorded. This is the direction that finds omissions and the direction that gets skipped, because it is harder: you have to select items in the warehouse yourself rather than working down a list.

Both are necessary. A count tested only in the first direction cannot detect an entire pallet that nobody wrote down.

Control the count records. Record the tag or sheet numbers issued, used, unused, and voided; account for all of them; and keep a copy of the sheets covering your test counts — signed or otherwise identified — so that the totals cannot be changed afterward without your knowledge. Retaining your own record of what you tested is the difference between a test count and an anecdote.

Watch the cut-off. Note the last receiving and shipping document numbers, observe any movement that does occur, and check that goods received during the count are segregated and handled per the instructions.

Look for what only the floor tells you. Dusty, damaged, obsolete, rusted, or oddly located stock; items in a corner with no movement; returns awaiting disposition; and anything the client's records will not reveal. This observation feeds the valuation assertion and you cannot make it from a spreadsheet.

Note the physical conditions — how goods are stacked, whether quantities can be verified without unstacking, whether the client's measurement method is reliable, and whether restricted areas exist.

Ownership Is a Separate Question From Existence

The classic error, and it runs in both directions.

Goods on hand that the client does not own. Consignment inventory received from a supplier, customer-owned material being processed, goods received but not yet accepted. Counting them and including them overstates inventory.

Goods the client owns that are not on hand. Consignment inventory placed with customers or distributors, goods stored at a third-party warehouse, and goods in transit — where the shipping terms determine ownership at the period end, and where the client's cut-off procedures frequently do not reflect them.

Two practical procedures: ask specifically about consignment in both directions rather than waiting for it to be volunteered, and examine the terms on shipments near the period end rather than accepting the client's classification.

For inventory held by third parties, confirmation with the custodian is the usual starting point — and where the amount is material or the custodian's reliability is uncertain, that alone may not be sufficient, which the standards address directly. Verify what is required.

Work in Process Is the Hard Part

Quantity is only half of a WIP conclusion, and the other half is harder.

Stage of completion requires understanding the production process, and it is where estimates and management judgment enter.

Cost accumulation — materials, labour, and applied overhead — determines the value, and the applied overhead is where errors and manipulation concentrate.

Standard costs and variances need examination: standards not updated for actual costs distort inventory, and the disposition of variances between inventory and cost of sales is a judgment with a direct earnings effect.

Which produces the general rule for a manufacturer: the count addresses existence and completeness; the costing work addresses valuation, and it is usually the larger risk.

Interim Counts and the Roll-Forward

Where the client counts at a date other than the period end, the count itself is not the conclusion — the roll-forward is.

Its reliability depends entirely on the controls over movements between the count date and the period end: whether perpetual records are maintained, whether receipts and shipments are recorded completely, and whether the client reconciles perpetual records to physical counts routinely.

Two practical points: test the roll-forward transactions, not just the arithmetic; and where the movement controls are weak, an interim count does not support a period-end conclusion no matter how well the count itself was performed.

Valuation Is Not the Count

Worth its own section because a clean observation gets mistaken for a clean inventory conclusion.

A perfect count tells you the quantities. It tells you nothing about the amount.

The separate work: cost determination and whether the costing method is applied consistently; the lower-of-cost-or-market or net realizable value assessment; excess and obsolete reserves tested against actual usage and sales history rather than against a formula nobody has revisited; the treatment of freight, duty, and capitalized overhead; and — for entities using it — the layer computations under a last-in-first-out method.

The reserve is where auditor and client most often disagree, and the useful test is behavioural rather than formulaic: what has actually happened to slow-moving items historically — were they sold at a discount, written off, or returned — and does the reserve reflect that history.

Technical grounding sits in the audit training courses catalog and the internal auditing training courses listing, with analytical context in analyzing financial statements.

The Fraud Dimension

Inventory is a classic vehicle for overstatement, because it is physical, dispersed, and hard to verify completely. Per our post on fraud investigation techniques, the indicators worth noting:

Pressure on the count schedule — a client who wants the count moved, shortened, or performed at fewer locations than planned.

Restricted access to an area, or an explanation for why a location cannot be visited.

Inventory at unusual locations, or held by a related party.

Large adjustments between the count and the recorded amount, particularly ones that resolve conveniently.

Consignment claims that cannot be documented.

Count sheets altered after the count, which is why retaining your own copies matters.

A margin trend inconsistent with the inventory movement, which is an analytical signal rather than a floor observation.

None of these establishes anything by itself. Each is a reason to extend procedures. Fuller method is covered in the fraud and forensic accounting catalog and the Certificate in Forensic Accounting.

A Note on Remote Observation

Video or remote observation became common and remains in use, and the honest assessment is that it is weaker evidence than attendance.

What you lose: the ability to select items yourself rather than accepting where the camera is pointed; the incidental observation of condition, dust, and location; the ability to move freely and look at what was not offered; and the deterrent effect of being physically present.

Where it is used, it needs compensating attention: direct the camera yourself, insist on selecting locations and items, verify the location is what it is claimed to be, and document the limitations honestly. Where inventory is material and risk is elevated, remote observation is generally not an adequate substitute — and the standards' provisions on impracticability are narrower than convenience.

Documentation

What the file must contain: the count instructions reviewed; locations selected and why; the count date and personnel; your test counts in both directions with items, quantities, and results; the tag or sheet control record; copies of the sheets you tested; cut-off document numbers; observations on condition and obsolescence; consignment and third-party inquiries and results; differences noted and their resolution; and your conclusion on whether the client's count procedures were adequate — which is the actual point of attending.

Where Inventory Observations Go Wrong

Counting the inventory instead of evaluating the client's count
Arriving without having read the count instructions
Not raising inadequate instructions until after the count
Selecting locations on convenience rather than materiality and risk
Testing only from the sheet to the floor, missing completeness entirely
Not retaining copies of the sheets tested, so totals can change
Tags and sheets unaccounted for — issued, used, unused, voided
A vague cut-off arrangement, with movement uncontrolled during the count
Not recording the last receiving and shipping document numbers
Missing obsolescence and condition observations, which only the floor provides
Not asking about consignment in both directions
Accepting the client's classification of goods in transit without checking terms
Relying on a third-party confirmation alone where the amount is material
Treating a clean count as a clean inventory conclusion, when valuation is separate
Reserves tested against a formula rather than against actual disposal history
Standard costs not updated, and variance disposition unexamined
An interim count accepted where movement controls do not support the roll-forward
Testing only the arithmetic of the roll-forward rather than the transactions
Remote observation used where inventory is material and risk elevated
Accepting the camera's view rather than directing it

The summary for the senior planning this: read the count instructions before you go and fix them if they are weak, choose locations on risk rather than distance, and test in both directions — because the floor-to-sheet direction is the one that finds what nobody wrote down and the one your team will skip. Then remember that everything you did establishes quantities, and that cost, ownership, and obsolescence are separate conclusions requiring separate work.

Frequently Asked Questions

Is the auditor supposed to count the inventory?

No. The client counts the inventory; the auditor observes and evaluates whether the client's count procedures produce a reliable result, and performs enough test counts to conclude on it. An auditor who spends the day counting boxes has evidence about a few items rather than about the population.

Why must test counts run in both directions?

Because they test different assertions. Sheet-to-floor tests that recorded items exist in the stated quantities; floor-to-sheet tests completeness — that items physically present were actually recorded. The second direction requires selecting items in the warehouse yourself, which is why it gets skipped, and without it an entire unrecorded pallet cannot be detected.

What ownership problems arise at an inventory count?

Both directions. Goods on hand the client does not own — supplier consignment, customer material being processed, goods received but not accepted — overstate inventory if counted. Goods owned but not on hand — consignment placed with customers, third-party warehouse stock, and goods in transit where shipping terms govern ownership — are omitted if the cut-off does not reflect them.

Does a clean count mean the inventory balance is right?

No. The count establishes quantities and says nothing about the amount. Cost determination and consistency of method, lower-of-cost-or-market or net realizable value, excess and obsolete reserves, capitalized freight and overhead, and layer computations are all separate conclusions — and for a manufacturer the costing work is usually the larger risk than the count.

When is an interim count sufficient?

Only when the controls over movements between the count date and the period end support the roll-forward — perpetual records maintained, receipts and shipments recorded completely, and routine reconciliation of perpetual records to physical counts. The roll-forward transactions should be tested, not just the arithmetic, and weak movement controls mean the interim count does not support a period-end conclusion however well it was performed.

Is remote observation an adequate substitute for attending?

It is weaker evidence. You lose the ability to select items rather than accept where the camera points, the incidental observation of condition and location, freedom to look at what was not offered, and the deterrent effect of presence. Where it is used, direct the camera and item selection yourself, verify the location is what it is claimed to be, and document the limitations — and where inventory is material and risk elevated it is generally not sufficient.