THE STORES ACCURACY PROBLEM

Let's first look at what inaccurate stores records can do to the company. ·

 
• They lower customer service,
• cause late deliveries,
• and increase freight costs.

Another key area is the supply of material to the production line. 

• If we fail, we end up with shortages and waiting time. 
• Shortages lead to excesses because a shortage stops a product going out but all the material and labour already expended on the other parts for that job, now sit and cannot be shipped. 
• We have missed schedules and the productivity goes down

• We replenish our stock because it is needed or because we have used it, dependent upon the system we use
• In both cases, we are totally dependent upon that stock balance. 
• Either we will buy what we don't need, or we'll not buy what we do need. 

Have you actually tried to establish what your accuracy level is? The annual stock take is no help at all as it will tell you that you have either lost or gained a certain amount of money during the year. From a manufacturing point of view, that is totally unacceptable, because it is neat numbers which we are concerned with. We need the right pieces to be available when needed. 
 

Being involved in MRP implementations, I often ask companies what their accuracy level is. They usually answer "Not too bad" or "It could be better". We need to establish quite clearly what level of accuracy we are looking for. We should aim for 100% but we need a minimum of 95% accuracy if we want MRP to work, otherwise we will spend our life blaming the system for material not being available when it should have been. The one let out appears to be a little phrase which says: "Measured in terms of significant errors" which means under certain circumstances, we must allow a tolerance. 
 

What happens if we count and it is within tolerance, should we correct the stores balance or not? The answer is no, because the count is as good you can get and if you do adjust, you are open to a problem called creeping losses. (This is a loss experienced each time the item is counted but within the allowed tolerance, so no investigation is required.) When doing the count, the procedure is to allow the tolerance percentage against the record balance, and then establish whether it is a good or bad count. A bad count gets investigated, a good count is accepted and no adjustments are made. 

• when a shipment of material is received, the stock in store should be at its lowest, therefore this would be a good time to count it. 
• If you have sufficient stores staff, when you make an issue, not only count the quantity to issue but count the remaining stock.
• A very simple one is, when the stock balance reaches zero, or goes negative. It is easy to count if you do not have any.
• The last override is that when the system suggests you place an order, that might be a good time to go out and count.

• You need space, the store needs to be well organised,
• well laid out with sufficient equipment.
• Individual item identification is always a problem, but needs to be solved. Rubber stamping the part number on as the last operation in manufacture or identifying the containers as part of the receiving procedure are approaches used by some companies.
• A locked store is essential. While people are allowed to go into the stores and take what is needed, (I just want to borrow one for a minute) the balances will never be right. I often relate as an example, if you were made responsible for a store and any discrepancies at the month end you must pay for out of your salary, would you allow people to wander in and out of your store and take what they thought was necessary? It is very simple, it must operate like a bank. If they need something, they give you a document for it. The bank doesn't allow you into their vault, why should we allow manufacturing into the material manager's vault?
• We need to look at our documentation processing, as this is a common accuracy problem.
• The last point is, when the factory operates, the store must operate, which means that if the factory is on a night shift, so must the stores. Manufacturing will either go over, under or through, but they will get the material in the "interest" of the company. Don't ask them to keep records of what they take, it is not their job.

 
• poor form design which is difficult for the data capture operator to use,
• untrained stores staff who do not know the procedures to be followed
• and the most common of all problems, is poor documentation control.

Who is responsible for the accuracy of the store. It can only be the man who runs the store physically. He is the only one who can control who and what comes and goes within a store, but to do that we have got to give him the tools to do the job properly. If the data capture is done out of the store, the daily audit trail and the original documents should be returned for him to check. Add the target accuracy to his job description, then he will know it is important and he will be measured against it. When we have big stores with many pickers, it may be advantageous to sub-divide the store and then allocate people to each sub-store. This way you can make people responsible for what they can control and create competition between the sub-stores.
If we intend to beat the accuracy problem, we need to let everybody know what it is. We need to issue weekly and monthly summary reports, showing

 
• the number of items counted,
• the number of errors,
• the error percentage and
• the important one, the common cause of the errors.

A control group is where you take 50 - 100 items, a mixture of A's, B's, C's, large, small and get an industrial engineer to go into the stores and count them and reset the balances. He then recounts these same items every two weeks, and analyses the common causes of the errors. He can then look at methods and procedures to alleviate the problem. Only when that has been done does it justify counting the store, and starting cycle counting. 
 

 
• fix the store physically,
• train the people and 
• let it be known that it is a major issue.
• run a control group,
• find the common causes of the errors,
• fix the procedures and then introduce cycle counting.

February 1997