Executive Summary

In this project, we analyzed the effectiveness of sales promotions for laundry detergent across four stores. We used scanner panel data to assess the profit implications of price cuts, consumer purchase behavior, brand choice, and consumers’ purchase quantity decisions. We created a binary logistic regression model to show the effectiveness of promotional tools in the detergent category purchasing process, a multinomial logit model to show the probability of choosing each brand, and semi-log models to estimate the purchase volume of each brand. We determined display ads were the best option for individual brands, and featured ads were the best to increase category attractiveness. We also found that consumers are sensitive to price cuts, so brands should keep a higher price and implement frequent price cuts to attract buyers. Finally, we found that Tide will bring the largest profit increase for the retailer during a price cut with a pass-through rate of 100%, and Cheer will bring the largest profit increase for the retailer during a price cut with a pass-through rate of 70%.

Introduction & Background

Liquid laundry detergent is one of the best-selling products of our customers. We were hired to find ways to maximize our customer’s profit by prioritizing certain brands and to analyze the effectiveness of different promotional tools (price cuts, display promotions, and feature advertising). We also had to find an optimal pass- through rate to maximize the profit of our customers and maintain manufacturers’ incentives to cooperate with our promotion plan. To facilitate our research, we were given scanner data for four laundry detergent brands (Wisk, All, Tide, and Cheer) across four stores and 178 households over 135 weeks.

Data and Methodology

The data used in this analysis is scanner panel data collected by IRI and consists of two datasets. One helps to determine the overall performance of the laundry detergent category by showing liquid laundry detergent purchases made by 178 households from four stores during 135 weeks. The other selects information from the previous dataset, focusing on the observations about the real occurrence of category purchases, with a total of 781 choice occasions and 3,124 observations. It was used to analyze the specific performance of different brands. We analyzed the above two data files in SAS and presented the critical values of the output results in Microsoft Excel to assist us in calculations and recommendations. First, we created a binary logistic regression model to show the effect of three promotions - Price-cut, Display Advertising, and Feature Advertising - on the purchase rate of laundry detergent. This helps us to determine which promotion is more likely to help increase purchase rates and use this as a basis to suggest the most desirable and effective promotions for retailers. Next, we ran a Multinomial Logit Model to analyze the probability of families choosing each of the four brands. From this part, we understood the decisions of families’ brand selection and how these decisions were affected by different promotion methods. Based on this, we can provide suggestions to improve brand attraction and capture consumers. Finally, four semi- log models for the four brands were created to estimate the purchase volume of these brands respectively by families. This model reflected the influence of the three promotion tools on purchase volume. Since price-cutting will cause changes in sales volume, thus affecting the gross profit of both manufacturers and retailers, we focused on this model when deciding on price reduction promotion.

Key Findings

Effects of Promotional Tools toward Purchase Incidence: Binary logit model is applied to this topic. Based on the parameter estimation, the estimated promotional tool “price-cut”, “display promotion” and “feature ad” does have an impact on increasing the attractiveness of liquid laundry detergent purchase, with one unit of increase, attractiveness increases by 1.07,0.49 and 0.56 respectively (Table 1) Thus, customers' brand choice decisions can be augmented mostly by price cut. The current regular price hurts the attractiveness of purchase incidence. The dependent variables of display promotion and feature advertising are used to demonstrate the state of the entire category which means no matter which brand conducts the display promotion and feature advertising, the attractiveness will increase by 0.49 and 0.56, respectively. (Table 1)

Effects of Promotional Tools on Brand Choice: We use a multinomial model to evaluate the promotional tools on the four brand choices. We are using Cheer as our baseline brand. After computing price cuts, display promotion and feature advertising will increase the utility of the brand choice by 0.71, 1.31, and 0.39. (Table 2) Therefore, customers’ choices can be most affected by display promotion. Promotional tool comparisons within 4 brands are conducted. Since the sum of the utility of 4 brands should be 1, it means whenever a brand increases its utility will surely decrease the other 3 brands’ utility. The tables reveal the same. Tide, with the largest intercept, has the largest competitive advantage when promotional tools are absent. When promotional tools are applied, it also sees the most significant increase in its probability of being chosen. (Table 2 - Table 14)

Estimate of Purchase Quantity: We are using a semi-log model to investigate the purchased quantity of a certain brand. Average volume, regular price, price cut, and previous category purchase on promotion are included in the model. All is the only brand with a negative coefficient of regular price, and it also has the largest price cut and previous purchase coefficients. This reveals that All is the most effective brand when a price cut is conducted, at the same time, customers are tending to stockpile when shopping with All. (Table 15 - Table 18)

Promotional Tools’ Effects on Expected Quantity: Using the 3 models mentioned above, 4 brands’ expected quantities can be calculated. With every promotional tool being used alone, Tide has the largest increase among the other 3 brands, which is also reflected in the brand choice model. (Table 19 - Table 30)

Evaluation of Pass-through Policies: Assuming the retail margin is 20% and the manufacturer’s margin is 45%, knowing each brand’s regular price and price cut, the regular retail and manufacturer unit margin will be calculated before price promotion. We can also compare the profits from retailer and manufacturer margins after price cut promotion and compare the gross profit changes after price cut promotion (Table 31 - Table 34)

In a scenario of a 100% pass-through rate, the retail total gross profit of 4 brands increased by 20.8% (Wisk), 24.3% (All), 35.7% (Tide,) and 19.6% (Cheer) (Table 35). The manufacturer's total gross profit of brands 1- 4 increased 31.2% (Wisk), 17.9%(All), 18.5% (Tide), and 23.7% (Cheer) (Table 36). Based on this result, the retailers' gross profit increases the most with Tide being the only price-cut brand. In a scenario of a 70% pass- through rate, the retail total gross profit of 4 brands increased by 27.2% (Wisk), 37.3% (All), 59.8% (Tide), and 28.1% (Cheer) respectively (Table 37). Nevertheless, the manufacturer's total gross profit of Wisk increases by only 10.3%, All decreases by 31.9%, Tide decreases by 11.3%, and Cheer increases by 3.5%. The brand of maximum gross profit increase for retailers in this scenario is Tide, making a 59.8% profit increase. Manufacturers’ gross profits would increase less or even decrease (Table 38). The result implies manufacturers of All and Tide may not agree with the 70% pass-through policy with their profits shrinking.

Conclusions and Recommendations

1. If a retailer's goal is to increase category attractiveness, then having featured ads is the most effective. For individual brands, the most effective way to increase their probability of choosing is through display advertisement.

2. In the category purchase incidence model and brand choice model, coefficients for price cut are larger than regular price. This reveals that most customers for detergent are loss averse. Setting a higher regular price initially is optimal to leave some leeway for future price cuts.

3. In a 100% pass-through scenario, Tide is the brand that will bring the largest profit increase for the retailer when Tide is cutting the price. The manufacturer is also gaining 18.5% more profit from this brand than before. Cheer is also seeing a 23.7% increase in profit so P&G, the owner of both brands, would have incentives to execute this price-cut plan for Tide. We recommend Tide as our primary price-cutting proposal for the manufacturer.

4. In a 70% pass-through scenario, only Wish and Cheer are seeing profit increase. All and Tide will have a 31.9% and 11.3% profit decrease respectively thereby eliminating them from our price-cut proposal. To maximize the retailer’s total gross profit increase in this scenario, we recommend a price cut for Cheer which will bring us a 28.1% profit increase compared to 27.2% for Wisk.

Limitation: Pass-through rate of 100% or 70% might not be the optimal rate for retailers, further studies be conducted to find the balance between maximizing profit and keeping manufactures’ financial incentives.