When American Airlines offered the first loyalty program in 1981, it forever changed how companies engage with their customers. Today, loyalty programs are not only ubiquitous in the airline and hotel industry, but also a key feature of any competitive credit card offering.
Loyalty schemes attached to credit cards have evolved significantly over the years, adding millions of new members; developing partnerships in order to offer innovative earning and redemption options; creating co-branded credit cards; and introducing tiered membership and varying levels of elite perks.
While this growth has offered marketing executives new and exciting ways to engage with their customers, it has left finance executives with an increasingly complex problem to solve: How much do these programs cost?
Today, credit card issuers can hold up to several billions of dollars’ worth of loyalty program liabilities on their balance sheets. In fact, two of the leading companies held over $8 billion combined in loyalty program liabilities as of yearend 2013.
With such large amounts at stake, accurately understanding and forecasting costs is critical for any finance executive with responsibility for a loyalty program. If you get it wrong, eventually you'll have to true-up the liability, which can result in a significant hit to income.
Challenge of getting it right
Unfortunately, loyalty program liabilities are notoriously hard to estimate.
Under accounting rules, a liability must be held for each point issued and outstanding, even though the actual cost associated with that point will not be incurred until the point is redeemed.
For some programs, points will remain outstanding for years before being redeemed, while some points will never be redeemed at all, which is referred to as “breakage.”
Forecasting that far into the future is difficult under any circumstances, but it is especially challenging for loyalty programs because they tend to be constantly in flux. Program structures are often modified. For example, in 2013 American Express introduced a “Use Points for Charges” feature. This feature allows eligible American Express Card members to use points to pay for any eligible transactions, rather than redeeming points for travel or gifts.
Marketing is constantly rolling out new engagement strategies to change member behavior. Think of all the bonus offers and reminder emails that you receive.
In light of this, finance executives face a very demanding task.
First, they need to accurately estimate the loyalty program liability, despite all this noise.
Second, they need to create a management framework to monitor and control costs, ensuring they are in line with forecasts and eliminating unexpected volatility in income. Our research has made great strides in solving these problems.
A better way to estimate loyalty liabilities
Many companies use actuarial techniques or similar approaches to address the challenge of accurately estimating their loyalty program liability.
Actuaries are good at this kind of thing, because the basic mechanics of program costs are very similar to those in insurance—loyalty programs have costs if points are redeemed, just as insurance companies incur costs if accidents occur. The timing of these events will impact the costs associated with them.
However, unless trained experts very carefully— and frequently—calibrate these techniques, the resulting computations may be biased in dynamic environments common to loyalty programs. Without proper skill, these methods may not fully anticipate and predict changes in redemption patterns, resulting in model estimates lagging behind true costs. This is particularly troublesome in an environment where costs are trending upward, because estimates will constantly come in low.
The reason for this is quite simple: These methods were designed for the more stable insurance industry rather than the fluid loyalty reward environment. A new approach specifically designed for loyalty programs is needed.
The solution is to use more advanced predictive modeling techniques to understand loyalty program costs. Such models are capable of assessing changes in forward-looking predictors, such as point-earning behavior, to predict changes in redemption behavior. This allows the cost estimate to be much more responsive and accurate, quickly reflecting the impacts of all available information on the program liability.
These models have an additional benefit of identifying predictors that are drivers of costs. This gives finance executives an intuitive way to interpret changes in the loyalty program liability.
For example, if the models indicate that a member’s expected cost rises in proportion to his or her cumulative earned points, then a historical plot of how the mix of points has shifted toward members with higher average cumulative earned points can help the program operators to conceptualize why costs are changing.
In practice, there are several important predictors of costs, and observing the “mix shift” across these dimensions often provides a level of transparency and understanding of the liability never before available to finance executives.
Finding a common framework
We have seen loyalty program managers resort to drastic changes to the loyalty program structure in order to control costs, such as devaluing the currency or introducing more restrictive expiration rules. While these measures often prove effective at decreasing costs, they are also very disruptive to the customer’s experience. Fortunately, there is a less disruptive way to manage costs.
As mentioned above, one of the biggest drivers of cost increases is the effect of mix shift, where the distribution of points is shifting toward certain high-cost demographics. So the key to managing costs is managing how this mix shifts over time.
Mix shifts are partially driven by individual members’ natural behavior as they earn or redeem points over time. However, mix shift is also affected by marketing’s own engagement strategies. For example, if marketing is targeting a certain demographic with bonus points or other incentives, then it’s likely this particular segment will grow. If this segment tends to be a higher-cost demographic, then overall costs rise.
Unfortunately, more often than not, marketing efforts are undertaken without a detailed cost analysis. That is, the mix shift is not managed at all.
The more responsive models described above can be leveraged to give Marketing the insights required to manage mix shift. These models can provide liability estimates at the individual member level. This, in turn, can be used by Marketing to identify and target only members who make the most financial sense, fully knowing the potential impact on the liability of the company’s engagement campaign.
This provides a new lever for both marketing and finance executives to control costs. With these models, they can continually monitor costs, then deliberately modify their targeting efforts to steer the liability as needed. This keeps costs within the forecasted range and eliminates unexpected volatility in income.
Understanding and controlling loyalty program costs is a very challenging task, leaving finance executives with an important question: At a time when loyalty programs have never been more in flux and profit margins so stressed, are you confident you know the true cost of your program?
About the authors
Len Llaguno and Manolis Bardis are both senior consultants at Towers Watson.
- Why ‘Explainable AI’ is the Next Frontier in Financial Crime Fighting
- Regions Financial Corp. Catches the Eye of Institutional Investors
- FinTech Lending: Friend or Foe?
- What All Banks Can Learn from Credit Suisse Group AG’s Earnings Report
- Santander signs $700M deal with IBM to accelerate its transformation