We have updated our Privacy Policy and Privacy Options

Got It

Are You Rewarding Your Customers for One Action, But Hoping for Something Different?

Share

Rewarding positive behavior—or punishing negative behavior—is so common in our lives, we don’t always notice it happening. When we tell a child he or she must finish dinner before getting dessert, we create a motivational force driven by the promise of a reward. Likewise, when a coffee shop promises a free cup after ten punches, they inspire us to return again to work towards our reward. And when the completion of a successful project earns a sizable bonus, the award reinforces the idea that hard work leads to reward.

There are at least two distinct theories that apply to these situations. Expectancy theory argues that an individual will behave in a certain way—will choose one action or option among many—because he or she is motivated by an expected result. In other words: Buy 10 coffees, get one free. Operational conditioning, on the other hand, describes a type of learning in which behavior is modified by rewards or consequences. In other words: If hard work led to a bonus, more hard work will lead to another bonus.

It’s a simple idea but often these powerful motivational forces are misdirected. We hope for one outcome but reward something else entirely. Steven Kerr eloquently outlined this problem in his 1975 essay “On the folly of rewarding A, while hoping for B.” In too many examples, Kerr explains, reward systems are “fouled up,” often to the extent that unwanted behavior is actually encouraged.

Kerr’s essay is most often applied to management theory—and it applies to managing a testing team, too. Test velocity, for example, might be the key performance indicator for your testing group. At the same time, you hope that as the number of tests increases and time to launch decreases, broken tests stay the same or even decrease. Unfortunately, this is probably not realistic. Instead, the team in this example should have an explicit goal—one that is actually measured—to reduce the number of broken tests over a set time period.

Perhaps more interesting, however, is applying Kerr’s idea to actual customers. Imagine, for example, the coffee shop introduced above was interested in increasing sales of sandwiches. Introducing the “buy 10 get one free” coffee cards, they reason will encourage customers to return regularly, increasing the chances they may eventually purchase a sandwich.

But notice that the coffee shop is rewarding customers for purchasing coffee specifically—encouraging coffee purchases by building an expectation for a free cup and helping customers learn the shop is a place they want to purchase coffee. The shop is rewarding coffee purchases but hoping for sandwich purchases.

Similar situations can arise in e-commerce as well. Free shipping offers, for example, are commonly offered incentives intended to convert browsing into completed sales. But what’s the ideal threshold for the free shipping to be activated?

If the KPI is increased orders, setting a low threshold—say after $20—might be the best option. If increasing average order value, on the other hand, is the KPI, a higher threshold—say after $80—would seem better. But one variable is unlikely to improve both measures simultaneously. Lowering the free shipping threshold will reward a lower average order value and hopefully increase the total number of orders as a result—but it does nothing to reward higher order values. This doesn’t mean increasing both metrics is impossible—rather, it illustrates that one strategy cannot improve both on its own.

Ultimately, this is an ideal problem to solve through testing. If the goal is increased revenue, a test could easily determine whether free shipping threshold that encourages many low-value orders or fewer high-value orders is more effective. But focusing on one and hope for an improvement in the other is a recipe for failure.


Brooks Bell helps top brands deliver exceptional customer experiences through testing, personalization, and advanced analytics consulting.

Our goal is to help our clients to effectively leverage their data, learn about their customers and maximize their digital revenue. Find out more about our services.