I was recently working at a large community hospital. The Supply Chain department was in a turnaround mode. They had a new supply chain leader who was addressing many of the issues. One contributing factor was a recent computer system change that was still causing problems. But there were other systemic issues. One element of my scope was to help them with their storeroom inventory. While the inventory was turning at a good rate of about 15 turns, there were still many problems.
For one, the last physical inventory showed a large variance. A second red flag was the excess/overflow stock stored in the hallway and on top of shelves. A detailed look at the individual item turnover uncovered an interesting fact: The overall rate was indeed 15, but most of the items were turning at very low rates, while a small number of items were turning excessively. One of the reasons that 15-18 turns is considered benchmark for a storeroom (see Back Talk, HPN, November 2006, www.hpnonline.com/inside/2006-11/0611-BackTalk.html) is the balance between the fiscal benefits of low inventory and the offsetting negatives of churn. But this inventory had both churn and higher-than-needed carrying costs for most items.
It quickly became clear that one reason for the unusual turnover pattern was the method of ordering. Staff had decided against the best practice: Allowing the computer program to reorder based on min/max levels or other algorithms. Instead, staff was walking around the shelves daily and ordering stock based on their perception of what was needed.
Computer-generated reordering is almost always superior to manual — as long as inventory accuracy is generally good. When levels are set correctly, items are reordered only when needed, and the quantities are both sufficient and appropriate for the storage area. Stockouts due to usage variations are rare, and transaction costs from excessive orders and overnight shipping are low. Not only is the inventory better managed, the staff time needed for the reorder function is a small fraction of the time taken for manual reordering.
In our case, moving to computer-generated from manual ordering needed a three-pronged approach. The first is bringing the inventory accuracy to an acceptable level. The second is establishing appropriate reorder points and reorder quantities. The third is having the right amount of space available for each product.
Inventory accuracy requires several factors. Of course, staff must be careful to pick the right items in the right quantities. However, if other processes are working well, staff error is rarely a significant contributor to inaccuracy. In our case timely transaction input was poor. Items were picked but the issues were not input until later — sometimes as much as three days later. Even receipts were a problem. They were using advance ship notices to receive items (another best practice). But the receipt was being processed when the invoice was generated — well in advance of the goods being on site. So the inventory quantities included stock that had not been received and, over a weekend, might not be received for days. Even the stock that was received might not be shelved for hours or days.
Another best practice that helps to maintain inventory accuracy is a robust cycle count program (see Back Talk, HPN, January 2009, www.hpnonline.com/inside/2009-01/0901-BackTalk.html). But you can only do a cycle count when there is a time period where the physical count and the computer count should agree. Because of the timing issues noted earlier, there was never a time when everything was up to date. Hence, cycle counting was not possible.
Fixing inventory accuracy was the top priority. Three changes were implemented: Those pulling stock became responsible for inputting the issued amounts directly after pulling, advance ship notices were turned off and items were received into the system only when they were physically on site, and workflow was changed so goods from the prime distributor were put away before PAR tasks were started. These changes also allowed the start of cycle counting.
As the accuracy problem was on the way to being rectified, the other two fixes could be addressed. A detailed review of usage and order ship time was conducted. Appropriate reorder points and max quantities were calculated. At a minimum, reorder points were set so that sufficient quantities would be on hand (without overordering) to satisfy needs over a three-day weekend. This process also identified many low/no use items that were targeted for deletion. Once new max quantities were established the final step was to assure there was sufficient storage space in each location to contain the max quantity. Some spaces expanded to accommodate larger max quantities. Other spaces were reduced so that only enough space was allocated for the max.
With all this work completed, computer-generated ordering could begin. At first the order was checked in detail. Staff still walked around the shelves to double check the computer computations. If staff questioned a computer order or saw an item not being ordered that looked off, they would check both the inventory accuracy and the min/max. Adjustments were made as needed. After a short time, staff began to trust the computer generated order and reviews became more cursory.
The benefits of this exercise are now clear. Staff has time for more productive activities, accuracy is good and checked frequently, and the halls of clear of excess. Computer-generated inventory reordering does work. If you are not doing it yet, the time to change has come.