You need to input the following values for a sku:
- EOQ or Normal Order Quantity (in Day's Supply): This is the Economic Order Quantity in day’s supply, or the average quantity you normally order – also expressed in day’s supply.
- Review time: This is the normal replenishment order interval, also in days.
- Lead Time: This is the average lead time, in days. Note that you need to include the full time span – from the time you “freeze” the inventory history to begin the purchasing process until the time a purchase order is received, processed, and the goods are stocked and ready for sale. Many distributors use one day on the front end to process suggested purchase orders based on demand and on-hand data from last night. And, many distributors require one or more days to receive, stock, and “post” receipts. You need to count all that in the lead time.
- Lead Time Variability: Lead time variability arises from two sources. First, there can be some variability in the number of days to the first receipt for a purchase order. This can arise from varying transit times, or varying times in your own company to receive and process shipments. Second, unless a vendor ships every order at 100% order fill, there will be additional lead time variability at the sku level. In our auto parts business, sku level lead time variability was the greatest on the mid-range items (ranked by popularity). Vendors typically give the best order fill (and therefore less lead time variability) on the fastest moving items – where they may have continuous production line manufacturing processes in place. Vendors typically give relatively poor order fill (and high lead time variability) on the slowest moving items. That typically doesn’t matter much, as it still might not “round up” to a higher safety stock level on these very erratic slow movers. But, in the mid –range items, vendors typically have “job lot” manufacturing processes in place, and if they run out of an item, they may be out of stock for an extended period – until their next production batch is scheduled. If your software does not track actual lead time and lead time variability, try using at least 25% for this input variable. It will add a few days of safety stock, and you likely need it.
- Demand Variability (as the Mean Absolute Deviation % of the Forecast): Most forecasting packages will report some level of demand history variability. If your software reports the Mean Absolute Deviation, use it (as a % of the forecast) as the input. If your system reports the Standard Deviation, sigma, you can use approximately 0.8 times the Standard Deviation (expressed as a % of the forecast) as an estimate for the Mean Absolute Deviation.
- Periodicity: This is the number of history keeping periods per year in your forecasting system. If you forecast monthly, use 12 here. If you forecast weekly, use 52. Etc. This matters because it “scales” the Demand Variability. Suppose you have an item that sells an average of 100 units per week. A 30% Mean Absolute Deviation on a weekly forecast is ±30 units. But, that same item has an average demand of just over 400 per month. If you forecast monthly, a 30% MAD is a deviation of ±120 units. So, an item with MAD at 30% of average demand requires a lot more safety stock if that deviation is based on monthly forecasts vs. weekly forecasts.
Input all these values, and tab out of the input field for the calculator to produce the chart. The chart shows the number of days of safety stock for various service level goals from 85% to 99.5%.