Counting Widgets, Wasting Time

What are the purposes of doing a physical inventory?

Those that still do the practice believe it provides assurance of the recorded assets. Some believe that it makes sure that the items' count is correct. Others do it only because the financial auditors require it.

If the inventory count is incorrect, there's only three main reasons:

• the physical inventory count was inaccurate (and this happens)
• the process of executing inventory transactions has problems, like timeliness, accuracy of receipts/sales/usage, unit of measure confusion, accounting for waste and other normal losses, etc.
• intentional shrinkage (theft).

Implementing a physical inventory doesn't fix any of those problems. Therefore, within a few hours or a day, the inventory count is incorrect again. Small inaccuracies build up over time and quantity of transactions till the percentage of error looks tremendous. For example, suppose I use 100,000 widgets over 1,000 transactions, and my usage is incorrect by 1%. If I take a count around transaction #900, the system record would show I should have 10,000 widgets left. In reality, I'll have anywhere from 9,100 to 10,900--an error rate of 9% and well outside acceptable physical count limits.

Financial auditors need reassurance that the overall inventory dollars are correct. QA professionals have, for almost 100 years, shown that random sampling is a good test of a hypothesis that the product lot meets the specifications. It would be very easy to set up a random sample of items to count, check their aggregate costs against the recorded values and determine if the hypothesis stands that the dollars are accurate within the bounds of confidence limits. If the sample is accurate, then the processes and security is in good shape to state that the balance sheet record of inventory assets is accurate. Auditors must realize that 100% confidence is impossible. With a random sample, the risk of saying it's good (and it's really bad) is a factor in determining the sampling plan, and this can be minimized.

If management is not doing it's job by improving the inventory transaction processes, a physical inventory doesn't solve the problem or do any more to encourage the right behavior then a random sample. The subtle punishment in a random sample is a complete count, if too many items deviate from their accepted counts. If the system is tough shape, then the organization will be forced to count the whole mess--and be shut down longer while this occurs. A properly done sampling plan is nearly impossible to "game" and not find problems if rampant problems exist.