Marketing and Consumers

Understanding the ‘Want’ vs. ’Should’ Decision

Pizza or salad? Consumers use different approaches to buying things they want (pizza) versus items they should buy (salad). In their research on online grocery-buying habits and DVD rentals, Harvard Business School's Katy Milkman and Todd Rogers, along with Professor Max Bazerman, provide insights on the want-should conflict and the implications for managers in areas such as demand forecasting, consumer spending habits, and effective store layout.

Like having both an angel and a devil whispering advice in our ears, consumers often wrestle with the "want" versus "should" decision. Yes, I want to purchase that chocolate bar, but I should snack on granola instead. Yes, I want the gleaming red Ferrari, but given my budget, I should purchase the used Toyota Camry.

How consumers weigh those decisions is crucial information for retailers, and is the subject of recent research by Harvard Business School doctoral candidates Todd Rogers and Katy Milkman, in collaboration with Professor Max Bazerman.

In this e-mail Q&A with Milkman and Rogers, they discuss their working paper "I'll Have the Ice Cream Soon and the Vegetables Later: Decreasing Impatience over Time in Online Grocery Orders," which looks at our tendency as consumers to choose "want" items in the short run but weigh toward "should" items when the impact will be felt in the future.

They also talk about their related research on DVD rentals—should I rent the good-for-me documentary An Inconvenient Truth or the entertaining Pirates of the Caribbean—and offer implications for managers in areas such as demand forecasting, consumer spending habits, and effective store layout.

Sarah Jane Gilbert: What is the difference between the "want-self" and the "should-self"? How does psychology play a role in the internal conflict between the 2?

Todd Rogers and Katy Milkman: People often behave as if they possess multiple selves with different, competing interests. We call 2 of these metaphorical selves the "want-self" and the "should-self."

The want-self is myopic and desires instant gratification. If left to its own devices, the want-self would always act on immediate, visceral desires (e.g., spending instead of saving money, eating junk food instead of health food). The should-self, on the other hand, prefers to behave in a way that will maximize long-run benefits. If left to its own devices, the should-self would always act on behalf of an individual's long-term best interests (e.g., saving money or donating it to a good cause instead of spending frivolously, eating health food instead of junk food).

People focused a higher percentage of their spending on ice cream (and other 'want' groceries) when placing orders for delivery in the near future.

The "multiple selves" metaphor resonates with many people because most of us regularly struggle with choices between 2 options, one of which we know we should choose because it would be virtuous to do so and one of which we want to choose to satisfy our visceral desires. An obvious example of this type of dilemma is the choice of what to eat for lunch: a slice of pizza or a salad? What do you think you should eat? Now, what do you feel, viscerally, that you want to eat? For most people, this is an easy one: They want to eat the delicious, greasy pizza but know they should eat the healthy salad. This is a classic example of want-should conflict.

Past decision-making research has shown that when a decision will take effect in the future, people weigh the interests of the want-self less relative to the interests of the should-self. For instance, you are more likely to agree to put additional money in your savings account if that money will be withdrawn from a future paycheck rather than the paycheck in your pocket (Benartzi and Thaler, 2003). Similarly, you are more likely to donate money to a good cause if the donation will come out of a future paycheck (Breman, 2007).

Q: How do you apply this to online grocery shopping?

A: We predicted that online grocery shoppers would order healthier groceries when ordering for delivery in the distant future (i.e., 5 days from now) than when ordering for delivery in the more immediate future (i.e., tomorrow). This is exactly what we found.

We analyzed a year of individual-level data from a North American online grocer to determine how the delay between when a person's order was completed and when it was delivered affected the content of the order. In general, as the delay between order completion and the date a customer selected for delivery increased, customers spent less money per order (or behaved less impulsively), ordered a higher percentage of "should" items (e.g., vegetables), and ordered a lower percentage of "want" items (e.g., ice cream).

A simple way of describing our findings is that people focused a higher percentage of their spending on ice cream and other want groceries when placing orders for delivery in the near future than when ordering for delivery in the more distant future. On the flip side, they focused a lower percentage of their spending on fresh vegetables and other should groceries when ordering for delivery in the near future than when ordering for delivery in the more distant future.

Q: Did you find similarities in other areas where consumers are making choices?

A: Our research team has completed 2 other projects that examine the dynamics of the want-should conflict. The first project, conducted by the same team of Katy, Todd, and Max, analyzed the implications of want-should conflict in the context of online DVD rentals. The second project, conducted by Todd and Max, tackled the prescriptive policy question: How can we increase the odds that people will select should options, which is in their long-term best interest to do?

For the first project we obtained access (with the help of HBS professor Anita Elberse) to a data set containing information about the film rental and return behaviors of a sample of the customers of an Australian online DVD rental company. With this data we examined whether the choices people anticipate making in the future (represented by the order in which they rent movies) systematically differ from the choices people make in the "heat of the moment" (represented by the order in which they watch and return the movies they rent).

In "Film Rentals and Procrastination: A Study of Intertemporal Reversals in Preferences and Intrapersonal Conflict," we predicted and found that people are more likely to rent DVDs in one order and return them in the reverse order when should DVDs (e.g., documentaries) are rented before want DVDs (e.g., action films). Similarly, we predicted and found that should DVDs are held by online renters longer than want DVDs, meaning people procrastinate considerably more about watching documentaries, for example, than action films.

For the second related project Todd and Max looked at how time-to-implementation from the moment of decision affects preferences for should options. We looked at scenarios that involved want-should conflict like support for a policy that would benefit one's employer but would be costly to oneself; support for a tax on gas to reduce consumption and pollution; support for a personal savings plan; and willingness to donate money. We found that people are much more likely to make binding selections of should options when the options will be implemented in the distant future than when they will be implemented in the near future.

This research offers prescriptive guidance for how we can facilitate the selection of should choices: structure these choices so that they are binding and implemented in the more distant future. This research is in our working paper "Future Lock-in: Or, I'll Agree to Do the Right Thing…Next Week."

Q: What implications do your research findings have for online retailers or other business channels?

A: Our study has implications for online and catalogue retailers that offer a range of goods for sale and also offer different delivery options. Such companies might be able to improve their demand forecasting by taking into account the fact that their customers appear likely to order a higher percentage of want goods and a lower percentage of should goods for delivery in the near future relative to the more distant future.

In addition, online and catalogue retailers should anticipate that the further in advance of delivery an order is placed, the less a customer is likely to spend.

Q: Could your findings be applied to in-person grocery shopping?

A: If we were to extrapolate from our results, ignoring the major differences between shopping for the present and shopping for the future, it seems that to maximize the sale of healthy items or should groceries, brick-and-mortar grocery stores might display healthy foods near the entrance of the store so they will be encountered early on in the shopping experience and as far as possible from the moment when actual consumption will take place. Brick-and-mortar grocers may have already realized this, as fresh vegetables tend to be displayed in the first area customers encounter when they enter most major grocery stores.

Q: What are you working on next?

A: With HBS doctoral candidate John Beshears, we are examining how people's online grocery orders are affected by the receipt of a coupon for $10 off their bill.

A rational actor model suggests that people without extreme wealth constraints (meaning anyone who can afford to shop for groceries online) should buy the same items they would have bought without a coupon when redeeming a $10 off discount. According to the rational actor model, the $10 saved should be considered a small addition to an individual's lifetime wealth, which will eventually be spent on a broad array of future purchases.

This is not the way we expected people to think about $10 discount coupons, however. We anticipated that the receipt of a $10 discount would cause people to "splurge" on some extra goodies they might not otherwise have purchased. This is exactly what we observed, and this finding contradicts standard economic theory. On average, people buy an extra $2 worth of groceries per order when they receive a $10 off coupon. This finding has important implications for economic models of life-cycle consumption, as well as the models marketers use to estimate the true costs of coupon offers.

Another area of research that Todd is particularly excited about involves using behavioral principles to more effectively mobilize citizens to vote. Voting is an important domain where people know what they should do (they should vote), but do not always follow through on it. Todd is working on a handful of other projects looking at how to facilitate people's decisions to follow through on the should choice to vote.

Katy is also pursuing a study of the earnings forecasts made by sell-side stock analysts. Along with John, she is examining biases in the way analysts update their earnings forecasts in response to new information.

Latest from HBS faculty experts

Expertly curated insights, precisely tailored to address the challenges you are tackling today.

Strategy and Innovation

Social Responsibility

Diversity and Inclusion