Retailers and product manufacturers alike heavily invest in promotions such as coupons, special displays with price discounts, “buy one, get one free” deals, and other promotional techniques. A high percentage of the overall marketing budget is linked to promotions, which have a significant effect on the top line of consumer product companies and retailers.

Imagine a promotion with a price reduction during the holiday season, the busiest shopping time of the year. A retailer who runs out of stock during the promotion would suffer a major loss of sales. With sufficient supply, the promotion would have run successfully.

We will describe how SAP Supply Network Collaboration (SAP SNC) responsive promotion management can react to the actual sales data in a timely and automated fashion, adjust promotion replenishment dates or quantities, prevent out-of-stock situations, and increase product sales at the store. For more information on SAP SNC, see our article Improve Forecast Quality with Sales Forecast Collaboration Using SNC and APO. 

Promotion Management

Before we explain the responsive promotion types, we want to show the different patterns relevant for promotion replenishment planning, the sales forecast, the promotion forecast, and the retailer’s distribution center promotion demand.

The sales forecast is mostly determined by historical data or market analysis. The sales forecast describes the expected sales activity over the runtime of the promotion. Based on this, the promotion forecast is determined, describing the movements from the retailer’s distribution center to the retailer’s stores. The distribution center promotion demand defines the promotion product demand at the retailer’s distribution center. The promotion demand definition typically allows for an initial inventory buildup before the actual start of the promotion sales and additional shipments during the runtime of the promotion.

Shipments for promotional product are pre-planned and independent from the actual sales occurring at the store during the promotion runtime. A promotion that runs better than expected can lead to an out-of-stock situation. In the opposite case, the retailer can become overstocked with promotional product. Figure 1 shows a schematic view of the promotion demand, promotion forecast, and sales forecast — each has a different time and quantity distribution.

Figure 1

Promotion demand, promotion forecast, and sales forecast

As part of the promotion management process in SAP SNC, three types of promotions are available: static, dynamic, and reactive. Static promotions do not allow for an automated adjustment of replenishment quantities during the runtime of a promotion. Most promotions executed today are static. This is appropriate when the promotion forecast is highly accurate and no variations from the forecast are expected. However, if variations exist, a more responsive promotion management is required. SAP SNC provides two types of responsive promotions: dynamic and reactive.

With dynamic promotions, the delivery dates of the promotion demand are adjusted based on the actual sales data. If sales are higher than expected, quantities are automatically delivered earlier to the retailer’s distribution center, avoiding stock-out situations. If promotion sales are slower than expected, additional deliveries are automatically postponed or even cancelled.

In contrast, reactive promotions have a fixed delivery schedule of promotion demands based on a predefined distribution pattern. The delivery schedule is not changed during the runtime of reactive promotion. Instead, based on the actual sales data, the total promotion quantity forecast is recalculated each time the system receives new data (which can happen daily or even sub-daily). Based on the actual data, a new promotion forecast quantity is calculated and the promotion demand is updated correspondingly using the distribution pattern. For example, if the promotion is running better then expected, the promotion total quantity is increased and more product units are delivered to the retailer’s distribution center by the next scheduled delivery date to meet the future demand and to prevent stock-outs.

Examples of Promotion Management

In the following sections, we simulate a promotion using static, dynamic, and reactive sales promotion management. We show how reactive and dynamic sales promotion management can help your company avoid stock-out and over-stock situations.

Static Promotion

The first example is a static promotion for an article called MKB_RR_MAT_A@Q43002. Here, the customer location CUMKB_CU_CA@Q43002 is the retailer location, which is typically the retailer’s distribution center. Figure 2 shows the SAP SNC Promotion Planning screen.

Figure 2

Example of a static promotion

Final Promotion Demand is the demand at the retailer’s distribution center. Final Promotion Forecast is the planned quantity to be shipped from the retailer’s distribution center to the stores. Promotion History shows the actual shipments from the distribution center to the stores during the promotion runtime.

In this example, the total promotion quantity is 1,000 cases. The promotion shown in Figure 2 starts with a build-up period on October 7 with 500 cases, followed by two further deliveries to the distribution center on October 12 and October 16. The original plan for shipments from the retailer’s distribution center to the retailer’s stores is 140 cases the first day (October 8), followed by 130 cases on the second day, 130 cases on the third day, and so on.

The quantity of actual shipments (Promotion History) was higher than originally planned, possibly because a higher-than-planned demand at the stores led the stores to request more deliveries. Instead of 140 cases, the first shipment was 155 cases. Instead of 130 cases on the second day, 140 cases were shipped, and so on. In this way, the initial build-up quantity of 500 cases was exhausted on October 11, leading to a stock-out of 30 cases. These 30 cases could have been sold if they were available at the store at the right time. The next day’s delivery of 400 units to the distribution center and the last shipment of 100 units on October 16 were on time and allowed for sufficient inventory at the retailer’s distribution center for the remainder of the promotion. Overall, the promotion sales could have been higher if sufficient inventory had been available.

Dynamic Promotion

If you take the same example and define the promotion as a dynamic promotion, the picture changes. For a dynamic promotion, the promotion demand quantities at the retailer’s distribution center depend on the actual quantities shipped to the stores and are defined by the dynamic promotion pattern. Figure 3 shows the pattern for this example.

Figure 3

Dynamic promotion pattern

The promotion demand starts on October 7. The first entry of the dynamic pattern defines that without any promotion sales (consumption), 50% of the promotion total quantity should be delivered to the retailer’s distribution center. That means the promotion demand on October 7 is 500 cases. Because the promotion is planned several weeks or even months before the start date, no real promotion sales history is available at the time. Therefore, the Planned Promotion Forecast data is used to define the second and third delivery of the promotion demand based on the second and third entries in the dynamic promotion pattern, respectively. For example, the second entry of the pattern defines that 40% of the promotion total quantity is needed if the Planned Promotion Forecast consumes 85% of the delivered quantity (i.e., 85% of 500 cases, which would be 425 cases in this example).

As shown in Figure 4, as of October 11 the sum of the promotion forecast is 500 cases (140 + 130 + 130 + 100 cases), which is greater than 425 cases. The second delivery of 400 cases (40% of actual promotion quantity) is planned for October 11.

Figure 4

Dynamic promotion initial setup

Depending on the number of entries in the distribution pattern, the series of deliveries (or promotion demands) is planned. With a carefully defined percentage of the consumptions, there will be no out-of-stock situation for the promotion if the Promotion History matches the Planned Promotion Forecast. Figure 4 shows the dynamic promotion in the planning stage.

Once the promotion start date has passed, the promotion starts running and promotion sales quantities are sent back into the system, populating the Promotion History key figure. Each time a new promotion sales data point is available, the system triggers a dynamic promotion adjustment by using the Promotion History data (instead of the promotion forecast data) to recalculate the promotion demand points in the future buckets.

Figure 5 shows that the first day’s sales are 155 cases instead of the forecasted 140 cases. This does not change the calculation of the second demand point on October 11, but it does affect the third demand point and causes the demand to be required one day earlier than originally planned — on October 14 rather than October 15. Without this adjustment, there is a risk of an out-of-stock situation on October 15 because the 90% threshold defined in the dynamic promotion pattern (Figure 3) will be violated.

Figure 5

Dynamic promotion after first day of sales

On October 9, the sales data comes in as 140 cases (Figure 6). Based on this data, the 85% threshold of the second distribution pattern entry will be reached on October 10. Because of this, the delivery of 400 cases will be moved forward to October 10.

Figure 6

Dynamic promotion after second day of sales

The same process runs each day after the sales data is registered in the system. With dynamic promotion, the promotion total quantity is not changed. The dynamic pattern controls the promotion demand. By carefully defining the distribution pattern, dynamic promotions can effectively react to the deviation of the real promotion sales and planned promotion forecast and then move forward or push out the promotion demands. In this sense, dynamic promotions can prevent out-of-stock situations as well as overstock situations. Consider again the example above: If the third distribution pattern threshold (90%) will not be reached due to low promotion sales, the third promotion demand will not be placed during the promotion run.

Reactive Promotion

The next example illustrates a reactive promotion. The idea behind the reactive promotion is to prevent out-of-stock and overstock situations by adjusting the promotion total quantity. For example, if the promotion is running well (i.e., promotion sales are better than the planned sales forecast) the promotion total quantity increases, as does the future promotion demand. In the opposite case — if promotion sales are slow — the total quantity and the future demand decrease correspondingly. Figure 7 shows a planned reactive promotion before its start.

Figure 7

Reactive promotion initial setup

On October 8, the promotion starts running and the sales of the day are sent into the system as 155 cases (Figure 8).

Figure 8

Reactive promotion after first day of sales

Because the real sales are higher than the planned sales (155 vs. 140), the system reacts to this data by increasing the promotion total quantity proportionally from 1000 to 1107 cases. The promotion demand for the upcoming buckets is increased accordingly, as is the Planned Promotion Forecast.

On the next day, October 9, the promotion sales data comes in as 140 cases and populates the promotion history key figure (Figure 9). The system compares the up-to-date total Promotion History volume (295 = 155+140) with the total promotion forecast in the same period (286 =140+146) and readjusts the promotion total accordingly from the original 1000 cases to 1031 cases.

Figure 9

Reactive promotion after second day of sales

In this example, the reactive promotion adjusts and redistributes the promotion total quantity on a daily basis. With more sales data available during promotion execution, you can form a more realistic picture of how the promotion is running. As of October 11, the total promotion sales and the forecast show no significant difference (Figure 10), so the promotion total quantity is updated almost back to the original level of 1000.

Figure 10

Reactive promotions after fourth day of sales

After the initial hype, if sales turn out to be lacking in the middle of the promotion, the promotion total will be recalculated to 978, less than the original planned 1000 cases (Figure 11). The system updates the remaining promotion demands correspondingly.

Figure 11

Reactive promotions with lower actual sales

There are often irregular spikes in promotion sales, especially at the beginning of the promotion execution. The reactive promotion allows you to choose the start-of-reaction point, such as three days after the promotion start date. In this way, the system does not react to the first three sales data points and avoids jumpy adjustments of promotion total quantity. 

You need to define the type of promotion (static, dynamic, or reactive) up front and you cannot change it once the promotion becomes active. A careful analysis of past promotion data is required to define the most suitable type of promotion. Lead times also play an important role. The shorter the transportation lead times, the better the supply chain can react to demand changes. Long lead times, in contrast, could lead to situations where no adjustments can be made within the runtime of a promotion.

Christian Butzlaff

Dr. Christian Butzlaff is responsible for SAP Value Prototyping services at SAP America. Before that he held various positions in the SAP Development Organization. In particular, he was Development Manager for SAP Supply Network Collaboration (SAP SNC). In his current role he works with customers on prototype engagements, for example, in the area of SAP Integrated Business Planning.

Christian holds a PhD from the University of Hamburg, Germany. He is PMI certified and is currently pursuing a specialization in machine learning.

 

You may contact the author at c.butzlaff@sap.com.

If you have comments about this article or publication, or would like to submit an article idea, please contact the editor.

V. Krishna Anaparthi

V. Krishna Anaparthi has vast experience in the field of enterprise consulting. Playing the roles of project manager and solution architect, he is instrumental in the successful implementation of multiple SAP projects across the consumer goods, pharmaceutical, and high-tech industries. He is currently serving as the process architect at SAP Labs India Pvt. Ltd. and is a part of the SAP Business By Design development team.

Krishna holds an MBA from NITIE, Mumbai, India. He has PMP, CPIM, and CSCP certifications and is proficient in supply chain and project management processes.

You may contact the author at v.krishna.anaparthi@sap.com.

If you have comments about this article or publication, or would like to submit an article idea, please contact the editor.