9. Promotional Forecast System

Promotional forecasting is the Achilles’ heel of any system. Promoting an item may require selling in one month the amount normally sold in one year. In the case of Do it Best Corp., one-third of the revenue of the stock sales was produced by promotions. Sale merchandise generally has the poorest service levels and tends to have higher levels of stock left after a promotion. This process is in need of improvement.

To solve the problem, it is necessary to implement a promotional forecast system that uses exogenous variables, or variables outside the system. An excellent statistical package such as SAS is necessary for this process. The forecast is made with multiple regressions to add value to the understanding of the dynamics of the promotion. The system begins with a number of variables that need to be described.

Promotional lift is also called price elasticity to demand. It sets the level of sales depending on the price of the promotion. For instance, dropping the price by 10% may increase sales by 40%. This is a lift of 4. The correlation of price-to-demand has a number of factors that need to be analyzed. The forecast is different by warehouse and by vendor. The reason for this is that the centers are in different areas of the United States and demographics tend to play out in this scenario. The concept of buying power index (BPI) does influence the buying in many areas. This figure can be gained by Polk or Clarita’s.

The promotional lift by item should be stored on a promotional database. The database will have a summary of all promotions and their associated lift, their margin, and the month promoted. It is very important to keep the promotions and regular demand in two different files. Do not add the promotional data to the regular usage sales file, or the demand data will be corrupted.

For example, if the same sale is not run next year and the promotional sales data is added into the regular demand data, purchasing will buy according to old information. The sales usage file shows inflated sales that will no longer be in effect because the promotion is outdated. When the data is stored properly, the purchasing manager can use the information to plan future promotions based on which items sold the best and at what price.

The dynamics of the optimal price for maximum profit can also be evaluated. The concept of cannibalization of gross margins must also be evaluated. This involves the idea that people will pay the regular price regardless of the promotion that has lowered the profit margin. Another consideration is the cross-correlation to other items. The concept of market basket analysis occurs when certain items are promoted, affecting the sales of other items.

Cross-correlation of promotion to stock sales addresses the fact that not all the customers take the promotional sale circulars. They tend to buy at the everyday low price or the stock price. The promotion increases the visibility of the product, and even the merchant not selling the item on promotion experiences an increase in sales. This amount of increase of regular sale items must be quantified. If the merchant thinks ahead, he could buy the merchandise on the promotional price and sell it at the regular price. This factor is known as PSD, or Promotion to Stock Demand.

It is necessary to have an indication of what the promotional sales are compared to regular stock sales. This is known as Promotion to Regular Usage (PRU). Not being aware of this information can be very dangerous. For example, running a promotion when the promotional forecast system says it is necessary to bring in a year’s worth of merchandise can result in a significant amount of overstock.

One key to knowing how well an item will do on a sale is to have a database tracking past sales on the item. How well did it do and what was the overhang of merchandise after the sale? This should also be on the promotional item record. Here again it is imperative that all regular stock demand is shown only on the regular item usage file. All promotional demand for promotions is kept on the promotional item file. This is the only way to make the distinction of what sells best in each category.

The promotional usage file tracks by item, sale, and warehouse. This tracks the best timing and analysis by sale and by month. An item may be on sale for six months or so, placing the item on six circulars with each circular running for a month. Each sale is coded by the month in which it is run. This outlines the monthly analysis of how the sale worked. If the PRU is high, the promotion may result in overstocks. This is an opportunity to negotiate with the vendor to have the item shipped at different times so that the performance of the promotion can be evaluated.

Demographic information can also be used to selectively change some of the items on the sale circulars. A list of which items sell best in each region can enhance sales demographically. It also helps in actually increasing the sales of merchandise and effectively lessening the chance for poor sales and overstocks. Increasing sales results in lower after-promotion merchandise in the inventory, which decreases overhead.

After the regression analysis is completed, the supplier can be provided with the forecast in an electronic document called an EDI 830 transaction. With the added predictability of the regression analysis, it is possible to predict the overall monthly usage of promotional demand with 80% accuracy. Offering this information to suppliers in advance allows them to run their Material Requirement Programs. This is definitely a way to minimize the volatility of the downstream supply chain.

• It can minimize the expediting of merchandise throughout the promotion.

• Suppliers can ship on the correct date because of the added information. They have a clear plan for the Master Production Schedule.

• The supplier was able to run a more effective MRP program because unpredictable demand was eliminated. The demand is more predictable because of the regression forecasts.

• Running weekly simulations on the program can alert the supplier to any changes.

• The savings is in the out-of-stocks on promotional merchandise with a reduction from 20%-plus to around 10%-plus. The inventory has been reduced by 25% throughout the sales. The key reason for this is that the long lead times on promotional items are longer than those on the normal stocking items. The lead times are longer because the amount of product being purchased is greater for a one-time event over a four- to six-week period.

• The schedule is dependable and the supplier can plan ahead. The orders are shipped on schedule, and if changes need to be made, they are usually small and the supplier can react to them. With this process, the order is not covering the entire sale period but is placed by the week throughout the sale period, lowering inventory amounts.

• The average inventory is considered as 1/2 the Order Quantity (AI = 1/2 OQ) of a forecast system. The order quantity is dictated as OQ = D × (LT + RT) + k × MAD × (LT + RT).

• LT is the lead time in weeks.

• RT is the review time in weeks.

• If inventory is reviewed every other week, the average LT = 18.5 working days for stock orders and LT = 25 days for promotional orders or 5 weeks. Lead times are now reduced for promotional orders from 25 days to 7 days, or 1.4 weeks of lead time. The reduction of Order Quantity should be days or 48.57% decrease. If the order quantity is reduced by 48.57%, the average inventory is reduced by 48.57%.

Figure 9-1 shows graphically that the Average Inventory = 1/2 OQ Order Quantity. The following bullets describe the values on the graph.

• P1, P2, P3, and P4 are the order periods.

• P1 and P2 Order Quantity is 1,400 and it gives an average inventory of 700.

• P3 and P4 Order Quantity is 700 and it gives an average inventory of 350.

• The Y axis is the Order Quantity.

• The X axis is the time in periods 1 to 4.

Figure 9-1. The computation of average inventory per period

This program allows for a cut in corporate inventory by 4.6%, which is an inventory savings of $16,000,000.

• The OQ is reduced by 48.57%.

• The Sales Promotions account for approximately 1/4 the sales of the company.

• The average prices of the promotional items average 40% of the average regular stocking items as a sample. By buying the merchandise on promotions, you can save 20% to 50% on the price, compared to nonpromotional periods.

• The reduction of overall inventory is calculated as .4857 × .25 × .40 = 4.6%.

• The Sales are $936,824,568.

• Turns before the Promotional Forecast program = 5.17 turns.

• Inventory reduction from the Promotional Forecast program is 4.6%.

• The actual Inventory Reduction is 4.6% × $181,093,991 = $8,330,323.

• The new inventory is $181,093,991 − $8,330,323.59 = $172,763,667.

• Carrying cost reduction is 26.6% × $7,742,287 = $2,215,866.

• Freed-up cost of capital is 2% × $7,742,287 = $166,606.

• The new turns are $936,824,568 / $172,763,667 = 5.42 turns.

Total Lean Savings is $2,386,162.