Solutions by Industry

Optimizing Customer Rewards Program : A Case Study

A major U.S. grocery chain, with more than 800 stores in North America and annual revenues of over $5 billion, wanted to enhance its customer loyalty program.

  • The client wanted to understand if its recently upgraded membership program, with rewards for frequent high-spending customers, was profitable.
  • The client was not able to discern the effect of the program as customers participating were self-selected (non-random), membership cards were inconsistently used over time, and there was an untimely opening of a major competitor in the middle of the test market.
  • APT's Test and Learn™ Management System showed that while the program was on average unprofitable, specific types of customers responded profitably to the program.
  • The analysis saved millions of dollars in potential loss from rolling out an unprofitable program. Furthermore, a targeted rollout of the updated rewards program generated significant profit for the client. 
The Company:

A major U.S. grocery chain, with more than 800 stores in North America and annual revenues of over $5 billion

The Challenge:

The client wanted to enhance its customer loyalty program. It had recently upgraded its membership program in selected markets to include rewards for frequent, high-spending customers. The new program had the potential to drive sales up but would also be very costly. Their initial analysis indicated that sales were up amongst participating customers, but was the program actually profitable? With over 10 million customers in their loyalty program, and many different factors influencing the data, it was a challenge for the grocer to evaluate.

In the past, when the grocer's analysis team wanted to evaluate customer data, they called up the data warehouse where they had outsourced their customer database and requested specific sales reports. They did not have direct access to the data, and that made it hard to go beyond simple reporting and basic analysis. Reporting is inherently different from testing in that no control groups are selected; it is also difficult to objectively compare sets of customers. Reports on the new rewards program indicated that it was a huge success. However, the effect was not discernible at the store level. The grocer sought an integrated view of the new program in order to explain this discrepancy and find out how the new initiative really affected customers and stores.

There were several other challenges in accurately gathering and analyzing customer data that could not be dealt with using the grocer's typical reporting methods. First, the participants in the new program were self-selected, meaning that they did not constitute a randomly-selected group and were not representative of the general customer population. A second difficultly was the faulty assumption that customers use their membership cards consistently over time. In reality, members of the new program had an increased incentive to use their cards (compared to members of the old program), since they received rewards on their purchases. To top it all off, the test was affected by the untimely opening of a major competitor in the middle of the test market at the start of the test period. All stores in the test market saw an instant drop in sales.

The Solution:

Using Test & Learn for Customers™, APT identified a representative subset of the members of the new loyalty program. This test group was compared to a closely-matched set of control customers drawn from stores where the program had not been offered. In initial analyses, the test group showed a significant sales lift compared to expected performance. However, deeper analysis revealed that this lift was driven by an even greater lift in the number of transactions per active customer. The average transaction value had actually dropped. At the same time, there was a marked decrease in the number of anonymous transactions (transactions without a membership card) within the test market. This confirmed that the increase in transactions per customer was likely a reflection of increased card usage, rather than an actual increase in the number of transactions. APT adjusted for changes in card usage rate and estimated the actual spending increase to be much lower. The effect of the program on retention was also examined, and it was found that customers participating in the new program had a retention rate much higher than the control group.

APT quantified the effect of the new program on eight different demographic segments of the customer population. The population was also broken down by demographic attributes, such as income, age, household size, as well as behavioral patterns (e.g. number of shopping trips per week). After accounting for the cost of the program (on an individual customer basis), it became clear that the profitability of the program varied significantly by customer according to some of these customer characteristics.

The Results:

It had been clear from the beginning that rolling out this new loyalty program across the entire chain would require a very significant investment - likely tens of millions of dollars. What was less clear was whether or not it would change customer behavior enough to generate an attractive return on that investment. APT's analysis revealed a fundamental insight, critical to making a sound economic decision: the program would indeed change customer behavior, but that change would not be dramatic enough to pay back the investment. Armed with this finding, the client avoided what would have been a costly investment with an unattractive payback.

APT's analysis yielded another important insight. While the program was on average unprofitable, the analysis did identify specific types of customers that responded profitably to the program. In fact, if targeted on just these customers, the program could be profitable, potentially generating up to $ 5 million of annual profit improvement. The client has since observed similar patterns across other direct marketing initiatives. These insights are helping them to create further value as they test different variants of the original loyalty program as well as enhance the targeting of other direct marketing campaigns.