An aggregate inventory-based model for predicting redemption and liability in loyalty reward programs industry

Loyalty Reward programs (LRPs) are marketing programs aimed at rewarding customers for repeat purchasing of a product or a service. Although different types of LRPs exist today across a spectrum of industries (travel, hotel, retail, telecommunication, banking, gasoline, etc.), most of the modern systems have their roots from AAdvantage®, the loyalty program introduced by American Airlines in 1981 (also called frequent flyer program in the airline industry). Consumers in these systems are given incentives or rewards for repeat business, which in turn serve as motivation for them to continue buying a product or service. In general, these systems involve at minimum a promotional currency (e.g. points or miles); single or multiple reward tier(s); a comprehensive database of individual consumers’ demographics and detailed transaction information; and an advanced technology to manage the program (e.g. redeem rewards directly or through internet), operate the contact center, and to analyze the members’ database. Hence broadly, in a typical LRP, customers become members of the program, earn points (based on some specified “accumulation scheme”) on their purchases of products or services throughout the network of LRP’s commercial partners. These points can be redeemed based on a “reward chart” pre-established by the firm that owns the LRP (i.e., host firm). Points that are not redeemed are saved in the customer’s account (under some conditions, e.g., being active) and constitute the LRP outstanding balance (“liability”). Points earned by customers during a given period (e.g., a year) constitute the LRP issued points (“accumulation”), whereas points redeemed by customers for rewards during a given period (e.g., a year) constitute the LRP redeemed points (“redemption”). Loyalty reward programs have increasingly become prevalent in recent years. For instance, in the airline industry alone, more than 130 companies currently have a LRP, and 163 million people throughout the world are enrolled in their programs. Geographically, LRPs have been quite popular in the United States, United Kingdom, Canada, and a host of other countries. Some studies show that 90% of Americans and 92% of UK consumers are members of at least one LRP. In the Canadian market settings, a special article featuring loyalty programs reports that according to Visa Canada Association, more than 25 million VISA cards are in circulation in Canada and about 78% of all card holders belong to one or more LRPs. A study from ACNielsen reports that 95% of Canadians belonged to LRPs of department stores, mass retailers, general merchandisers, or warehouse clubs. Despite the prevalence of LRPs and the increased complexities in their management and control, there are few academic models that specifically deal with LRPs to support planning and operational decision making. Most of existing work focuses on the leverage of the individual consumer information accumulated in the LRP databases to improve marketing and sales decisions. One of the challenges faced by LRPs is that of developing aggregate and disaggregate predictions of redemption, liability, and accumulation to support short, medium, and long term planning and operational decision-making. LRP managers rely on good predictions of redemption and liability to plan for rewards supply, set program budget, maintain a balance between customer service level and overall costs of rewarding customers, and to assess the growth of the program and the risk level associated with this growth. The lack of availability of rewards at the time of redemption results in a poor service level and/or an increase of the reward supply costs to meet customers’ demands, since the LRP’s host firm will have to acquire the additional rewards at a higher cost. On the other hand, too much availability will result in a higher cost as well (although the level of customer service would be high in that case). The unused availability will result in a penalty whenever the LRP’s host firm decides to reduce or cancel their reservation of rewards or return unused rewards to reward suppliers (i.e., LRP partners). In effect, in setting up long term contracts with partners, LRP managers must decide the volume of rewards to purchase in advance. This results in the needs for good predictions of redemption. Good forecasts of redemption are also required in establishing proper budgeting plans or forward financial statements. Moreover, good predictions of redemption provide LRPs managers with the ability to develop promotion plans that seek for better management of redemption demand between peak and off-peak periods. Liability of LRPs is widely recognized in the industry as a risk indicator for firms’ future LRP operations. It represents the value of future redemption obligation of points earned by LRP members. LRP organizations will face a higher risk and corresponding challenges in LRP operations when the liability level is too high. Therefore, a good prediction of LRP liability (i.e. with a greater degree of accuracy) provides managers with the ability to anticipate the growth of the program as well as the liability associated risks (e.g. hyperinflation and devaluation of points). Furthermore, risk mitigation plans can be developed which may include strategies such as revision of reward scheme, reward pricing, changes in management policies, etc. We propose a predictive model of redemption and liability to support short, medium, and long term planning and operational decision-making in Loyalty Reward Programs (LRPs). The proposed approach is an aggregate inventory model in which the liability of points is modeled as a stochastic process. An illustrative example is discussed as well as a real-life implementation of the approach to facilitate use and deployment considerations in the context of a frequent flyer program, an airline industry based LRP.