…well nothing really, but it serves as a great analogy for the sequential testing method for counting inventory.
There are approximately 100,000 hairs on the average human head. Although this number differs slightly depending on color variants (blondes tend to have a little bit more, redheads, a little less), in all cases it’s a huge number.
Hence counting each hair individually would be quite a hairy task (pun intended).
How about a little mind game? Let’s assume we have a guy, who somehow claims to know the exact number of hairs on his head and even the position of each individual hair follicle. He also claims that baldness is not a problem for him as his daily hair loss does not exceed the human average for people without a genetic disposition for baldness. Is it possible to prove him right or wrong without counting each hair on his head? As a condition let’s assume that the guy put all his hair inventory data into a multidimensional table, so we have a snap-shot of the current situation.
Let’s assume you have the superpower of nearly microscopic vision and can switch rapidly from one random follicle to another in just one second. Do you think the task is still impossible?
I think I could prove him right or wrong in less than one minute. And no, I am not cheating!
The sequential testing process is a mathematical method used in inventory sampling and could help me do this job. Using this statistical algorithm, I only need to look at as few as 30 random hair follicles on the head and sequentially test them against the data in the table. If 30 healthy follicles are present, the guy is telling the truth. If not, baldness may be right around the corner. Not convinced? Maybe I need to explain sequential testing in more detail. You may be able to understand the method better when it is described in the context of a warehouse.
Although warehouse managers probably never have a reason to start counting their hairs, at some point they will need to know how much inventory is actually stored in the warehouse. Given that book data alone does not typically satisfy auditors, inventory-taking is an unavoidable task that the majority of businesses must complete. However, for both businesses and warehouse managers, taking a full count of inventory in a warehouse with a six-digit stock number is a daunting task.
Using the sequential testing method of inventory sampling, the accuracy of book data, even in very large warehouses, can be proven (or disproven) with minimum effort. During the ongoing inventory process, it checks whether the figures in the books are accurate, and whether they can be used as a basis for financial reporting. This is done by drawing random samples, which are tested sequentially.
After each sample is counted, the number of “erroneous” samples is compared to previously calculated acceptance and rejection boundaries. So after each single step in the counting sequence, you can say whether your inventory is accepted, rejected or indifferent (which means more samples must be drawn in order to reach a decision). The figure below illustrates the decision process:
If you have highly accurate book figures, the counting sequence can reach the acceptance boundary with a minimum of 30 samples, regardless of the size of the warehouse. So it doesn’t matter if you have 10,000, 100,000 or even 1,000,000 positions on hand.
As you can see in the diagram above, the acceptance and rejection boundaries for the whole inventory rise cumulatively for each “erroneous” test sample. If you have inaccurate book data and you are also very unlucky, you might get stuck in the indifference area forever, resulting in the full inventory count you wanted to avoid.
Although sequential testing is the most efficient method of counting inventory, as you can probably now imagine, it is only suitable for warehouses with very precise stock management. It is essential that the inventory differences are smaller than 1%, while the number of single differences permitted is also limited. Generally, only automated warehouses meet these strict requirements. But if you have an inventory accuracy of 99% (regardless of the process you use to reach that figure) this just might be the best method for you to test the accuracy of your book data.
During my experience of inventory sampling, I have seen several large companies in Germany use the sequential test and some of them actually managed to complete their inventory-taking with the minimum number of samples (30). And even while this sounds implausible to people who are not that well versed in mathematics, auditors approve of this method.
30 out of 100,000 is less than the average daily hair loss. You won’t even notice.
Still not convinced? We will be holding a webinar on the topic on August 29th: “Inventory Sampling: Stocktaking made easy.” We will present the various methods used during the sampling procedure, including the sequential test. Click here to read more and sign-up!