20. The Characteristics Needed in a Forecast Program

This chapter is designed to list the sundry differences of many systems and their applicability. The information was created in collaboration with customers and suppliers. It was also accumulated as a wish list for future program changes in the system. When designing or buying a system, the following list can be used as a guide to what each system requires.

1. Many purchasing managers base their service level on different criteria. Some want to stock three or four weeks’ worth of stock as an average. This does not offer a quantitative view of how to reach the company goal. When trying to maintain a 97% service level with three weeks’ worth of stock on hand, this does not correlate. This is why the service-level metrics are important in a forecast system.

a. Service level by item. This enables the user to show and measure the service levels by item to include a hit list for the items which are at the worst service level. The information should be analyzed and developed further in collaborative department meetings. The only way to create solutions to existing problems is to know which items are causing them. Raising the service-level requirements on these items may solve the issue.

b. Service level by vendor. This is a good assessment technique for evaluating vendor performance. This can be used on a vendor scorecard when it is time to measure the vendor execution.

c. Service level by warehouse. If there is a large difference in the service levels among the various warehouses, there may be a logistics or personnel problem. This comparison should be readily available on a billing report. Many times this is also a metric used for training the trainer. Key people in the warehouses with the highest service levels and turns can train the personnel for new warehouses. The same personnel can share their ideas in semiannual management meetings.

d. Service level by department. The department service level is used as a barometer for future improvements. It is recognized that each department may vary from the next because of the uniqueness of the products carried. Comparing each department’s service level for improvement may not be the issue. Some departmental categories will historically carry a lower service level. It may be helpful to collaborate with the vendors involved to see whether they have better ideas for improvement. Internal improvement teams should meet with suppliers regularly.

e. Service level using the ABC. The final concept using the classification approach needs to be measured and input into the system. As discussed earlier, it is the weighted average of the volume, revenue, and profit models in calculating the k factor for the safety stock.

2. Scheduling inbound freight should be performed with the information from the forecast master file. The issues in this category revolve around logistics visibility. It is imperative to use an Advance Ship Notice (ASN) and software programs to match the delivery of a truck to a destination and time. The EDI 214 is also useful for freight scheduling from the carriers.

3. Seasonal profile forecasting is the concept that allows the computer to aggregate the seasonal patterns. When category management is used, lift by season for all the categories is outlined. This helps select the correct items for each season and lumps all like classes together. It is a good idea to lump products by their master codes. For instance, the National Hardware Retailer Association (NHRA) has categorized products by their type and use. These item categories can be further broken down to the end item. In the tools and drills department, comparisons by demand for a 12-volt variable-speed cordless drill versus cord model are optional. All vendors who make this type of drill will have the same NHRA code. When changing or adding vendor lines, all the suppliers are listed. The reference codes come in handy when looking for new items or seasonal patterns. This is a fast way to increase forecast accuracy. If the buyer can offer a rough estimate of the average sales for the item, it can be overlaid as a deseasonalized demand with the seasonality profile, which offers a better idea of how and when to buy the product.

4. Automatic reinitialization occurs when the computer automatically changes the smoothing rate and/or forecast models to minimize forecast errors. The system constantly checks for accuracy and determines whether the correct forecast model is being used. This is done by using demand filter and tracking signal limits. When the system sees problems, it will go back in and evaluate the current model for improvements. If the model needs corrected, it automatically performs the corrections. For example, around Father’s Day, certain items sell at higher rates over the weekend. The computer automatically notices the seasonality change and selects the correct model: horizontal, trend, seasonal, trend seasonal, or lumpy.

5. Adaptive forecasting is a system used for short-range and long-range forecasts. This is a term that denotes the capability to change the forecast constant as demand varies. In this case, the exponential constant is changed from something like .1 to .15 or .2, depending on the system suggestion. The beta constant for the safety stock also changes as the Volatility Index varies.

6. Modified EOQ occurs when software packages deviate from the conventional EOQ concept. This is accomplished through changes in the following areas:

a. Allowing quantity discounts to be entered into the calculations.

b. Making lead-time adjustments and considering how they will change the EOQ.

c. Making adjustments to allow EOQ to vary with seasonality.

d. Varying the EOQ to coincide with capacity.

e. Varying the EOQ to coincide with the minimum order multiple.

7. The use of regression analysis studies external changes and what their effect will be on the forecast. This is a useful technique in the evolution of demographic studies or categorizing items by region. This model is also used in the optimum pricing strategy and in promotional lift evaluations.

8. The management of correlation and data graphics services uses correlation analysis to tell how much of the variation in a forecast is explained by the independent variables. The data graphics services are used when management needs to see the activity of the item. In this case, the results are displayed graphically. This may be used for the following reasons:

a. Discovering exceptions in demand—What causes the demand to change?

b. Reviewing an early order strategy—What will happen if the merchandise is brought in early?

c. Reviewing critical items—Will the system will show the critical items and their performance?

d. Reviewing seasonal items—Is the best seasonal inventory maintained? What unique categories are influencing the demand?

e. Finding cross correlations—Is there a category of items that influence the movement of other items such as Market Basket Analysis?

9. Fixed or variable intervals are used for the review periods to evaluate a vendor on a fixed or variable interval. This usually depends on the purchasing policy or the relative importance of the vendor. A variable interval reviewing system is best used to minimize the inventory, and the number of orders and unpredictability of the vendor reviews is not as important. The fixed review system is in place when it is very important that a schedule of vendor reviews is kept in place.

10. Look-ahead options cover forecast systems that use advance planning to look ahead and anticipate the next move. This feature is used in the following areas:

a. Scheduling of production—What happens if production is delayed?

b. Scheduling of receiving—What happens upon waiting until the next review period?

c. Scheduling of demand planning—How does the receiving schedule change if the demands of subsidiary warehouses change?

11. Demand patterns will be able to track five ways: horizontal, trend, seasonal, trend seasonal, and lumpy.

12. Shelf life is used in the food and drug industry but is not limited to only these two industries. Some companies have used the concept of shelf life on very expensive inventory items.

13. The open-to-buy system is established with a commitment from the retailer to buy a fixed amount over time from the supplier. This amount can be changed as long as enough lead time is given before any significant changes are made in the consumption planning. This is called an open-to-buy orientation. It is used or modified for the following:

a. Minimum inventory allocation. This implies a long-term contract for purchasing a minimum amount of inventory over time to ensure service to customers. The minimum amount stands at the average inventory level. Any additional inventory or safety stock can be added later.

b. Item service level. In the open-to-buy system, it is best to use a service level per item rather than an overall service level by vendor. It’s necessary to get down to the micro level because purchasing out for six months is required. Any item overages on item inventories will be magnified. In any case, the item-service-level plans inventory using the ABCDE inventory analysis.

c. Grouped vendor allocation. It may be advisable to buy for all warehouses as one long-range purchase agreement to the supplier. This purchase can be drawn from over time, with the items sent to the appropriate warehouses. This option reduces demand volatility in the forecast.

14. Expediting systems recognize the need to expedite and request a faster delivery date. This would alert the system of a pending shortage if the vendor stays on its normal manufacturing lead-time consideration. The system will decrease the lead time and ask for a quicker delivery. An order would be generated if needed with a shorter lead time. The system can be modified to review the vendor only if absolutely necessary.

15. Capture lead times in a forecast system can increase accuracy in the forecast system. The more relevant the information, the likelier the system will increase its accuracy. The lead times should be exponentially smoothed over time to keep an average lead time per vendor, per distribution center by vendor, per item, or per item by distribution center.

16. Fixed order quantities can be used for discount buying, freight savings, or receiving and stocking optimization.

17. Min/Max strategy stands for Minimum/Maximum Strategy; it yields a bounded solution to purchasing when necessary. It is used when certain constraints in either buying or warehouse capacity are present.

18. Seasonal profiles can be used in the following areas:

a. Grouping like seasonal products. Using a profiler highlights areas where purchasing can be grouped together to buy in larger, consolidated shipments where volume discounts apply.

b. Determining the maximum period of sales. This can be used in grouping like products for freight savings. Knowing when and where the maximum shipment will occur offers additional leverage with the carriers for added discounts for volume shipments over time.

c. Profiling new items. A new item does not have any seasonal characteristics added to its forecast profile. A new item may actually be seasonal and the buyer will complete two important steps to ensure that all new items have the correct profile if needed.

i. The buyer will give an overall average for new items. This can be entered manually or approximated by a similar item with the same item classification code. The item classification code indicates that the item is the same as or similar to an existing item in inventory. It would make sense to use a similar demand average for similar items.

ii. There will be an overlay of the new item with the same seasonal profile or template as the similar item described previously. This helps in maintaining a better forecast seasonality on new demand.

19. Seasonal indexing is a relatively new concept for the evaluation of seasonal merchandise. It will aid in the following areas:

a. Predetermining a slight pre-buy. It uses seasonal base indexes that measure the amount of increase or decrease by period for demand. Any period with a base index above one will be considered as selling above the average.

b. Increasing or decreasing of safety stock. After seasonality is determined, increases or decreases in safety stocks based on the index values are implemented. Entering a low-demand season requires lowering the safety stocks in advance.

c. Increasing or decreasing the forecast. If you increase your forecast because of the base index, mentioned in step b, you would also increase your safety stock. If you lowered your forecast because of the base index, mentioned in step b, you would lower your safety stock.

20. The joint order system process allows for adding to an order to make the minimum shipment. The system looks ahead and analyzes whether the order results in a dip below the minimum service level. The order may not be accepted because it does not meet the vendor minimum weight requirements. If that occurs, the system will increase the order size to match the vendor minimum. In this case, the system determines the service-level threshold and looks at future demand.

21. Management information systems or queries can be incorporated into the following:

a. Vendor modeling—This is used in modeling for different inventory scenarios for optimum inventory and service-level strategies.

b. Vendor sundry reports—This involves offering many different reports that can be made by the user.

c. Lead-time simulation—This involves testing the lead-time changes of several vendors to see what effect they will have on the inventory.