Do the buyers in your company face a mountain of replenishment decisions every day? In the past several weeks I have worked with several firms that stocked more than 16,000 unique products in each of several warehouses. Their buyers seemed overwhelmed with the task of maintaining an adequate inventory of each of these items, most of which were not sold on a regular basis. At each company we worked to “tame the replenishment beast.” In this article we’ll look at how we decided to replenish items that are not sold or used on a regular basis.
The first step in the process of developing a set of replenishment ground rules for these products is to separate items sold on a regular basis from those products that experience sporadic sales. To do this we sort all stocked products in each facility in descending sequence based on the number of times the product is sold, regardless of quantity. This process is commonly referred to as ranking by hits. Here is a summary of the 16,348 items stocked in one of our customer’s warehouses:
| Number of Items | Number of Hits |
| 2,076 or 12.7% | 12 or more |
| 2,501 or 15.3% | 6-11 |
| 4,120 of 25.2% | 2-5 |
| 7,651 or 46.8% | Less than 2 |
Notice that only 2,076 or 12.7% of the items were sold, on average, at least once a month. The replenishment of these popular products should be micro-managed to maximize inventory turnover (i.e., the number of opportunities to earn a profit) while retaining a high level of customer service. Indeed most books and articles on inventory management (including ours) focus on maximizing the profitability of these items that customers request most often. Most if not all of the methods described in these publications involve a prediction of future demand based, at least in part, on a calculated average of past usage. Although the specific calculation may differ from method to method, most rely on the average or weighted average of the quantity sold or used over a specific period of time.
But can these same rules be used to replenish items that are sold less than once a month? In this case that’s 14,272 products or 87.3% of the stocked products in the warehouse! Can you determine proper stocking level of these products based on an average of past usage? Let’s look at an example. Consider an item with following usage history:
Ten pieces of the item were sold in December and another ten pieces were sold in March. The history displayed suggests that when customers order the product, they order 10 pieces. But any forecast demand formula based on an average (or weighted average) of past usage will calculate a forecast of future usage of less than ten pieces. To illustrate our point, we’ll apply two common demand forecast formulas to our usage history:
- The Six Month Rolling Average Method that averages the usage recorded over the past six months:
- (10+0+0+10+0+0) ÷ 6 = Forecast of 3.3 units for April.
This is well below the normal sales quantity of ten pieces.
- The Weighted Average Method that decreases the weight or emphasis of each month’s usage history over the previous five months in the average usage calculation:
| Month | Usage | Weight (Emphasis) |
Extension |
| March | 10 | 3.0 | 30 |
| February | 0 | 2.5 | 0 |
| January | 0 | 2.0 | 0 |
| December | 10 | 1.5 | 15 |
| November | 0 | 1.0 | 0 |
| Total | 10 | 45 |
The total extension of 45 pieces is divided by the total weight of 10 pieces resulting in a forecast of April’s demand of 4.5 units. Again this is well below the normal sales quantity of the product.
We could apply other forecast demand formulas, but the results will probably be the same. The demand forecast will be less than the normal sales quantity of 10 pieces, and as a result there will not be enough inventory on-hand to meet the customer’s needs.
An item experiences sporadic sales if its normal sales quantity is greater than the average quantity sold or used per month. In the example above, the normal sales quantity is 10 pieces while the average quantity sold per month is 3.3 pieces.
If an item with sporadic sales should remain as a stocked product (the subject of next month’ s article), its replenishment parameters normally cannot be determined using a forecast based on the average of past usage. A better way is to set minimum and maximum quantities based on the normal sales quantity. In the example above, you might set a minimum of 10 pieces and a maximum of 20 pieces. When the stock level of the product dropped down to 10 pieces (one normal sale quantity) a replenishment order would be issued for another normal sale quantity of the product. If due to the critical nature of the product (or to extended or inconsistent lead times) additional safety stock must be kept for the product, consider setting the minimum quantity to two normal sale quantities. On the other hand, if you are willing to risk a stock-out during the time it takes to order and receive a replenishment shipment set the minimum quantity to zero. In any case the minimum and maximum quantities for these items should be based on the normal sales quantity or the normal quantity used in an assembly or process.
How do you determine the normal sales quantity? The easiest way is to divide the total number of pieces sold or used over the past 12 months by the number of orders for the product received over the same time period. For example:
| Total Pieces Sold or Used Over the Past 12 Months | = 40 pieces |
| Number of Sales and Requisitions | = 4 pieces |
| Average Sale Quantity | = 10 pieces |
The average sale quantity often reflects the normal sale quantity. However its accuracy may be influenced by one or two unusual sales. A more accurate method of determining the normal sales quantity is to search transaction history for the mode in the transaction history of the product – that is, the quantity that is most often sold or used.
Although items with sporadic sales or usage do not (or should not) usually represent a large portion of your total inventory investment, stocking these items correctly is crucial to providing a high level of customer service. It does not make sense to stock these products unless you maintain the most commonly requested quantity in your warehouse. Next month, we’ll look at when it is advantageous to stock items with sporadic sales, and how the cost of the item and vendor package quantities play a part in stocking decisions.