SYNOPSIS: The fastest way to doom your loyalty program to being only average in the minds of your customers is to use your Average Customer Profile to design it. Taking a more nuanced approach will help you avoid major miscalculations in your financial modeling and forecasts.
The Average Customer Profile is the favorite tool of financial analysts and management consultants because using averages cuts through the clutter and quickly provides reasonable estimates. However, averages work less well for designing and evaluating the financial success of a loyalty program, where it is imperative to look at real scenarios for real people. Working with only the Average Customer can in fact lead to gross miscalculations for program lift and liability.
The Average Customer doesn’t buy your brand. Real people do. People of different ages and experiences. People living alone or in households of vastly different sizes and configurations.
A car company could look at their customer base and see an average household size of 2.5, then assume that they need to make only one type of car – a sedan. However, if they segment their customer base they might find that 50% of households are childless singles who want a fun small car and the other 50% are families with two or three children who want an SUV.
Or a theme park may work from an assumption that the average food and beverage purchase is 0.5 hot dog and 0.75 of a fountain soda. Their park visitors may in fact eat only half their hot dog, but they still had to purchase the whole hot dog. What is happening in that case is that only half of the park visitors are buying a hotdog with a soda and another 25% buy only a fountain soda. This marketing task – to get someone to buy any form of lunch at a theme park rather than eating elsewhere – is a very different marketing task from getting someone to buy more hotdogs while they’re at the park.
Many loyalty marketers do the functional equivalent of designing their loyalty program around hypothetical household sizes and parents who magically buy less than a full hot dog at the park.
Here are the key points that the marketer needs to consider for more detailed segmentation during program design and financial modeling:
- Spend per member: Resist the temptation to use your current average annual customer spend for your loyalty program. Customers in the upper percentiles are far more likely to enroll in a program than customers who are less engaged with your brand. You’ll often see retailers brag that their average loyalty customer spends more than their average non-loyalty customer; this often represents self-selection bias more than it signals true incremental spend. But that’s a different, although related, concern.
From a modeling standpoint, spend usually determines the number of points earned, which in turn determines when a customer earns benefits. Failure to look at the range of enrollment rates and spend across your customer base could lead to severely under-estimating the number of members who will reach rewards and therefore your actual costs and timing.
- Reasonability of forecasted growth: Another danger point is using one flat average growth rate across all customers. Many clients will look for a 10% rate across the membership base. When a once-a-year customer becomes a twice-a-year customer, that’s a 100% growth rate. You need to look at the one-timers and make reasonable estimates about what percentage of them will increase from one to two trips rather than expecting all of them to add a fractional visit. Similarly, a 10% increase for some top tier customers could result in an unrealistic number of visits or spend per visit – for example, there may be set limits for how often a prescription can be re-filled for a drugstore customer. Determining reasonable growth within customer segments drives greater accuracy for many calculations ranging from promotion design to staffing levels at point of sale and also to forecasting revenue for quarterly reports to Wall Street.
It will also help you identify and personalize your marketing strategy to increase the relevance of your program. Our 2017 research on desired program attributes showed that personalization is consistently in the top four chosen by program members.
- Estimating liability: Breakage takes multiple forms – there are members who never reach a reward and there are members who reach a reward, but don’t redeem it. Breakage for those who never reach a reward is 100%. Breakage for those who fail to redeem a reward they’ve earned can range from 5% for travel reward programs to over 80% for some retail loyalty programs. Because of this variation, your liability calculations will not move in a straight line depending on how many points you require for your first reward or whether you have a first purchase bonus. You need to estimate the “never reach a reward” customer base separately from the “reach a reward but still don’t redeem” customer. Otherwise, you could be setting unrealistic financial expectations in the marketplace.
- Setting your top tier and program rules: Many loyalty marketers have heard that elite tiers should target the top 10% of a base and build their program forecasts using one average for the top 10% of customers and another average for the remaining 90%. While this is an improvement over using one companywide average, imposing an external structure based on received wisdom about loyalty programs rather than looking for splits in your own data can be just as misleading. The spend curve for some customer bases can literally look like a hockey stick where only the top 5% determine the profit for the company – extending elite benefits beyond the top 5% will detract from overall margin. Other times, the top customers receive so many layered incentives that they, in fact, become less profitable than customers towards the middle of the database. In that case, program rules need to be adjusted to channel top tier participation appropriately while rewarding their loyalty with increased experiential recognition rather than hard value rewards and discounts. But, you won’t even know whether you have an issue unless you look beyond the averages.
Conclusion: The Average Customer may not be like your real customers at all. Programs that design their programs by looking only at this mythical average end up delivering experiences to their real customers that can be just as odd and dissatisfying as a quick-serve restaurant where they serve a cup that is only three-quarters full and the beverage in it is a mixture of 50% hot coffee, 15% iced tea and 10% fruit smoothie (with a quarter of a quarter pounder on the side). We recommend taking a more nuanced approach and start by looking at the behavior of your customers at least by decile if not by individual percentile (or other known behavioral segmentations) to see where the true leverage points in your base lie.