Case Article—The RealPro Customer Benefits Program: Rekindling Shopper Loyalty Through a Subscription Service

1. Introduction

With a market share of 4%, Real is the smallest of the five largest retail groups in Germany. The retailer faces stiff competition from its larger competitors such as Aldi and Lidl. To thrive in the highly competitive German grocery retail sector, Real had developed an aggressive high–low pricing strategy with deep price discounts to attract price-sensitive customers. In recent years, Real has been facing unfavorable consumer trends, resulting in dwindling numbers of store visitors and reduced revenues. To counter this downward trend by enticing more customers to shop at Real, the firm implemented several strategic business initiatives that included new store formats, store refurbishments, and a strong online presence. Despite the promising results of each initiative, the monetary and time investment required to implement any of them fully would be considerable. So instead of relying on these initiatives for a turnaround, Real’s management team developed a new one: the customer benefits program RealPro. Customers pay a fixed annual fee to become a member of the RealPro subscription program, which entitles them to substantial and permanent discounts on nonpromoted food items available in Real stores. The management team expects major benefits from the implementation of RealPro—in particular, an increase in the share of wallet of Real’s regular and top customers. To evaluate the potential of RealPro under realistic conditions, a market test was implemented in seven Real stores in Germany. The test ran from February 28 through November 30, 2019.

This case study uses an actual management challenge to teach students about the increased relevance of data in business decision making. In the end, the students decide whether RealPro should be launched and if so, under which program conditions. Making the best possible decision requires both strategic considerations and data-based insights.

2. Case Questions and Objectives

The case follows Real’s management team in their task of turning Real around. The case and the associated case questions emulate the development of the decision process associated with the RealPro subscription program.

First, the students assess the current situation of Real, the need for a new initiative to put Real back on the right track, and the opportunities and threats associated with RealPro. The students first learn about the unfavorable situation of Real, including decreasing revenues and a shrinking market share. Before envisioning RealPro, Real developed and tested multiple initiatives aimed at improving the situation. As stated in the case, these initiatives have shared limitations, namely the time and money required for their implementation. As a result, the Real team envisioned, developed, and finally tested the novel RealPro program.

Second, the students use the test market data set to analyze the influence of RealPro on consumer behavior and the financial impact of the program. This requires the students to employ different analytical methods. The instructor must decide up front which program or programming language should be used by the students. We supply detailed solutions in both Excel and R, but any other data science oriented application (e.g., Python or Stata) can certainly be employed as well. Furthermore, we provide a case extension that involves the use of more advanced statistical tools, including difference-in-differences analyses and, optionally, two covariate balancing methods. Analyzing the impact of RealPro on customer behavior requires students to clearly distinguish between customer groups (RealPro vs. Control) and years (2018 vs. 2019). Depending on the data analytics proficiency of the students, the questions on consumer behavior can be extended or reduced. We provide two possible extensions in the form of store- and weekday-specific analyses, but the data set also contains additional insights that can be explored. Next, the students evaluate the financial impact of RealPro. The consumer behavior analysis paints a positive picture of RealPro, but the program’s impact on profitability must also be assessed. Students calculate the gross profit of RealPro on the basis of the market test conditions and the associated customer response. Subsequently, students use a sensitivity analysis to evaluate the profitability of RealPro under varying program conditions and customer responses. This encompasses analyzing the impact of changes in the membership fee, both financially and strategically. As students must use parts of the results of their customer behavior analysis for this section, instructors can consider providing students with reference values before starting the financial analysis section.

Third, the students must decide whether to launch RealPro for all Real markets, and if so, under which program conditions. This step requires students to critically reflect on their data-based insights and think about strategic implications to make the best possible decision concerning RealPro. Furthermore, students are asked to assess the design of the RealPro market test and how it could be optimized. Finally, the class can discuss how the RealPro program could be improved further.

3. Subscription Programs

Subscription programs are a novel but increasingly popular concept in the (food) retailing sector and research on them is still limited. Existing research on warehouse clubs can provide additional guidance and insights, as there exist strong similarities between these club concepts and subscription programs. Kim and Choi (2007, p. 178) investigate the role of the membership fee of warehouse clubs in competition with high–low pricing supermarkets. They state that club membership fees are “an optimal competitive reaction to the supermarket’s promotional activity.” According to Ailawadi et al. (2018), shopping at a warehouse club format substantially increases the quantity of packaged foods that customers purchase. The authors list four drivers for this increase in purchase quantity, including low prices at the club stores, larger package sizes, membership fees, and low store density. Particularly the low prices and the membership fee are two drivers that also apply in the context of many retail subscription programs and in the case of the RealPro program.

Existing studies on subscription programs underline the significantly positive effect on purchase or consumption quantity and variety these programs tend to have. Datta et al. (2018) empirically highlight a large increase in quantity and diversity of music consumption for adopters of an online music streaming service. A large increase in online purchases of beauty products based on the membership in a subscription program is reported by Iyengar et al. (2020). Wagner et al. (2021) investigate the effect of subscription services in the context of online grocery retailing. They find that subscription customers spend more per month and place more purchase orders. Despite the increase in revenues, the authors also find that the subscription service actually has a negative impact on the retailer’s profit.

Research on loyalty programs can also provide additional insights into subscription programs. However, we emphasize that loyalty and subscription programs are distinctively different with respect to some of their key characteristics. A loyalty program can be defined as “a program that allows consumers to accumulate free rewards when they make repeated purchases with a firm” (Liu 2007, p. 20). The RealPro program does not fit this definition based on the required membership fee and the absence of an accumulation mechanism of, for example, points based on repeated purchases. Instead, we define RealPro as a subscription program. The difference between loyalty and subscription programs and their expected benefits is also highlighted by Real introducing the RealPro subscription program alongside the Payback loyalty program and running both programs in parallel. Real expected RealPro would increase the share of wallet and not only the loyalty of consumers. Keiningham et al. (2011) highlight the relevance of share of wallet for retailers, which only weakly correlates with customer satisfaction as a proxy for consumer loyalty. Discussing the differences between a loyalty and a subscription program before assigning the case can help students better understand the motivation behind introducing RealPro.

Loyalty programs are seen as a fundamental tool for retaining customers by many retailers. Despite their popularity, the literature provides contradictory evidence on the effects of loyalty schemes on consumer buying behavior. Some studies show that loyalty programs increase the consumers share of wallet and positively influence consumers’ repatronage behavior (e.g., Verhoef 2003, Lewis 2004). Conversely, other studies suggest that loyalty programs do not or only sometimes foster consumer loyalty, making them cost-ineffective investments (e.g., Dowling 2002). A report by Doppelt and Nadeau (2013) even states that among publicly traded companies in North America and Europe, those companies without a loyalty program grow faster, in terms of revenue, than companies with a loyalty program. Furthermore, the authors find that companies in traditional loyalty sectors such as airline, car rental, or food retailing, had lower EBITDA (earnings before interest, taxes, depreciation, and amortization) margins if they spend more on consumer loyalty. Boudet et al. (2020) highlight that subscription programs can deliver value by locking in customers in highly fragmented or undifferentiated lines of business, such as the food retailing sector. The question hence arises: Are loyalty programs still worth their money in the grocery retailing sector or are there better investments to positively influence customer repatronage behavior?

This case provides the opportunity to thoroughly analyze a novel customer-oriented program, both empirically and strategically.

4. The Data

The accompanying data set contains recorded transactions from the RealPro market test. Overall, the data set contains approximately 75 thousand transactions from RealPro customers/households and—for the purpose of creating a control group—from customers/households who have not joined the RealPro program. Key variables, including the ID of the purchasing customer, the date of the purchase, and the total purchase amount, are recorded for each transaction. To assess changes in the purchasing behavior of RealPro customers after joining the program, the data set also contains transactions from the same time span of the previous year for all customers. This allows the students to examine changes in the purchasing behavior of RealPro customers after joining the program and to assess whether these changes can be attributed to a RealPro membership or can also be observed in the control group. To ensure the validity of comparisons made between the two groups and limit potential self-selection biases, the covariate balance of the provided data set has been improved by applying propensity score matching to it. As a consequence, several hundred control group customers were excluded from the original data set of 83 thousand transactions.

5. Classroom Experience

This case study is geared toward graduate and advanced undergraduate (bachelor’s, master’s, and MBA) students in marketing, retail operations, strategic management, or business analytics courses that involve data analytics methods. Students should be familiar with basic statistics and the chosen program or programming language for the data analysis.

We taught the case without the extension in two first semester courses of a master in management program, where the large majority of students had a background in business administration. All students were familiar with Excel but not with R or alternative data analytics related programming languages. Hence, we chose to conduct the case analysis in Excel. We provided the students with the case and the large data set up front. Alongside the case and the case data, a short technical note with links to potential refreshers on PivotTables and Data Tables in Excel was provided. Groups of no more than five students were formed. The groups were given one week to complete the case and hand in their written solutions alongside their Excel files.

Following the case solution submissions, we used one 90-minute class session to discuss and analyze the case together with the students. First, we discussed the competitive situation of Real and the opportunities and threats associated with the novel RealPro program. The students combined the information in the case and their own customer experience from grocery shopping to assess the viability of RealPro. This resulted in a large list of possible opportunities and threats associated with RealPro, with some student groups having an optimistic and others having a more pessimistic outlook for the program.

Second, we conducted the analysis on customer purchase behavior and RealPro’s gross profit using Excel. Volunteers from the student groups were asked to briefly present their analytic approach and their findings. As expected, the groups focused on different customer behavior aspects for their analyses. Letting the groups share their analytic approach and the focus of their analyses provided each group with a larger, more holistic picture of the effects of RealPro. We moderated the presentations and provided additional interpretation of the results. During the discussions, it became evident that some groups had calculated different results and had come to different conclusions than those proposed by us. For the customer group analysis (Question III), several groups had reassigned customers to transaction segments based on their transactions in 2019 rather than keeping them in the same segment based on the transactions in 2018. This deviation from our proposed solution was not critical to the successful completion of the case, but it restricted the ability of the groups to see the pronounced differences in the changes of shopping behavior of different RealPro and control group segments. In response, we have adapted the customer group analysis templates to make them more intuitive. The sensitivity matrix in Question V (Question VI in the case extension) was another source of frequent deviation from our proposed solution. Several groups interpreted the matrix columns to represent the increase in revenue per customer from 2018 to 2019 in percent, rather than the revenue per customer in 2019, which was to be increased or decreased by the displayed percentages. This resulted in very different sensitivity matrices, which also influenced the answers concerning the potential rollout of and improvements for the RealPro program. Hence, we have adapted the column and row labels of the sensitivity matrix to make them intuitively understandable. Furthermore, we have also included some visual cues in the form of dashed lines in combination with a small note, to highlight that the center of the sensitivity matrix represents the gross profit under the original market test conditions. We occasionally observed other sources of deviations from the proposed solutions in the gross profit calculation (Question IV). Some groups failed to integrate the RealPro membership fee into the calculation, whereas others used the full, annual membership fees and reduction in marketing and personnel costs rather than only 7/12 of these annual values.

Third, based on the results from the analyses and the previous assessment of opportunities and threats of RealPro, we discussed whether RealPro should be fully implemented. Finally, the students presented their suggestions for further improving the RealPro program.

The students responded with enthusiasm to the opportunity of working with a large data set containing real data. They debated vividly on their analyses and subsequently on the decision of launching or stopping RealPro. After the class, we asked students for their feedback. All students stated that they thoroughly enjoyed analyzing big data and using the analysis in the decision making process—something only a few of them had done before. Several students also announced their motivation and intention to select courses on data analysis for the next semester. Other students’ feedback included an interest in analyzing a novel business concept and the accessibility of grocery retailing and loyalty/subscription programs as a topic.

Working analytically with large data sets and making business decisions based on the results of one’s own analysis is a challenging task for students. The case questions and content can be flexibly adjusted to reflect students’ technical skills and their familiarity with the chosen data analytics program. The higher the skills level of the students, the less structure must be provided and vice versa. The level of structure in the case can be easily adjusted using the level of detail in the case questions and the number of analysis templates provided to the students along with the data set. Additionally, the number and depth of the requested analysis can be flexibly increased or decreased. At the minimum, the analysis should include an analysis of changes in purchase behavior per customer group between 2018 and 2019, as well as gross profit calculations of the program.

We provide a case extension that implements difference-in-differences analyses and, optionally, two covariate balancing methods on the original test market data set. Integrating these tools into the case gives students the opportunity to run in-depth statistical analyses and make statistically grounded decisions concerning the future of the RealPro program. As many students will be involved in designing experiments, formally or informally, academically or in their later careers, this case extension provides the opportunity to discuss and implement the more technical aspects of designing and assessing experiments.

6. Case Extensions

Outside of the technically oriented case extension, there are opportunities to discuss different retail related issues on the basis of the case story. In the case, the main challenge for Real is to grow revenue. This is often done by one of the following strategies: high–low pricing, everyday low pricing (EDLP), and the subscription-based model. Each of these pricing concepts comes with its own challenges concerning planning, operations, and logistics. A large factor in variation between the three pricing concepts is the demand side.

6.1. High–Low Pricing

High–low pricing is a pricing strategy that involves lowering the price on a varying selection of brands. These price promotions vary in timing and the level of discount offered and serve as a mean of attracting customers to the store. High–low pricing retailers experience substantial peaks in demand during their promotions. Breiter and Huchzermeier (2015) propose a two consumer segment demand forecasting model combined with supply contracts to match demand and supply in a high–low pricing environment. The authors’ forecasting model works well for nonseasonal products, but faces issues when forecasting products with seasonal variation in demand. Wolters and Huchzermeier (2021) develop a novel forecasting model for such frequently promoted, seasonal products.

6.2. EDLP Pricing

EDLP is a pricing strategy that, in its purest form, involves setting a constant, and often low, price for each brand in each category. Hence, a pure EDLP strategy does not involve the usage of price promotions. EDLP strategies are popular among discount stores such as Walmart or Aldi. Demand at EDLP retailers is generally assumed to be less volatile and hence easier to predict than that of high–low pricing retailers (Ortmeyer et al. 1991). However, EDLP retailers are still subject to variations in demand due to multiple factors, such as promotions and marketing activity by competitors. Recently, the traditional EDLP retailer Aldi has started promoting some portions of its assortment to attract more customers into its stores (Schulz 2019).

Demand forecasting for retailers with a subscription-based model has received little attention in the literature. Nonetheless, the potential implications of a subscription-based model are substantial for demand forecasting, with potentially very high forecasting accuracy based on customer lock-in effects and the absence of promotions. The most prominent example of a retailer following a subscription-based model is Costco (Bell and Leamon 1998).

7. Conclusion

The RealPro case presents an engaging way of analyzing the effectiveness of an innovative customer subscription scheme that turns apparent weaknesses of a one-stop-shopping retailer into major strengths. Students can explore a very large data set and develop strategic initiatives for revenue growth that are potentially actionable. On the positive side, students are eager to learn more about data analytics methods and the required programs/programming languages; on the slightly negative side, students would like to extend their analysis to customers’ shopping baskets, which are not provided in this data set. As this case study documents a real business situation and subscription programs are on the rise, further research should be encouraged.

8. Supplemental Material

The teaching note for the case and the case extension, including the data sets, the case solutions in Excel and R, and a PowerPoint presentation for the complete class session, are available upon request.