Choice without Awareness: Ethical and Policy Implications of Defaults

Defaults Make a Difference

Suppose a customer has two options when completing a purchase: he or she can enroll in a “rewards” program (and receive promotional offers by e-mail) or not. In these situations, it is common to speak of “opt-in” and “opt-out” policies. In the opt-in system, the default is not to automatically enroll new customers and to enroll only those who make an active request. In the opt-out system, all new customers are enrolled by default and remain enrolled unless they take active steps to quit. The default setting firms use can have significant effects on behavior.

Defaults are surprisingly powerful in several consequential domains, including matters of life or death. For example, Johnson and Goldstein (2003) find that in European countries with opt-in organ donor pools, often less than a quarter of the population opted in. In contrast, in opt-out countries, typically more than 99% of the population did not opt out, leading to enormous differences in donor pool size between otherwise similar countries. Studies show that default enrollment in 401(k) retirement plans can lead to 95% participation within a few months of employment, compared with approximately 60% participation without the default (Beshears et al. 2009). Defaults can also sell millions of dollars of insurance. In the early 1990s, both New Jersey and Pennsylvania passed sweeping changes in legislation that required every driver to choose between two alternatives: a high-cost insurance policy that provided the right to sue or a low-cost policy that lacked this right. New Jersey law mandated the inexpensive policy as the default, and Pennsylvania law mandated the more expensive one. Defaults exerted tremendous influence in this choice: 21% of New Jerseyans purchased the right to sue, compared with 70% of people in Pennsylvania (Johnson et al. 1993). It is estimated that $140 million more auto insurance is purchased annually in Pennsylvania (more than $2 billion since the law was changed) because of the default.

Beyond two-alternative choices, defaults exhibit strong, and sometimes stronger, effects in the presence of several, hundreds, or thousands of alternatives. Cronqvist and Thaler (2004) document how, under privatization of social security, Swedish citizens were sent a catalog of mutual funds and given instructions on how to invest for their own future. Of the 456 possible funds, a full one-third of participants chose to put their entire investment in the default fund, despite an extensive educational campaign encouraging them to make active decisions. Park, Jun, and MacInnis (2000) find that consumers chose a car with a more expensive set of features if the default was a fully loaded car from which they could remove features versus a basic car to which they could add features; they find similar results for treadmills and personal computers. Consumer-defined products illustrate the scope and power of defaults beyond fixed and captive consumers: consumers buying a car can compare across offerings from competing car manufacturers (e.g., manufacturer A with the fully loaded default vs. B with the base model default) or even across categories (e.g., computers vs. cars, with different default offerings).

Computers come loaded with defaults (e.g., the operating system, web browser, and its search engine are most often set by default), and many people never change the default settings on the software they use (Mackay 1991). When less-technical consumers receive technical products that ship with defaults, many of them may not realize they even have choices over such settings. Although software defaults may be trivial, they can still have enormous economic impact. For example, it has been argued that AOL Inc.'s $4 billion purchase of Netscape was motivated less by its software and more by its enormously popular home page, which some 40% of Netscape users preserved as the default (Kesan and Shah 2006). Because search engines such as Google and Bing make billions of dollars by placing ads among search results, the dispute over default search engines has found its way to the U.S. Department of Justice, the U.S. Federal Trade Commission, and the European Commission (Johnson and Goldstein 2006).

Evidently, defaults make a difference. Within the defaults literature, earlier work has focused on the potential welfare-enhancing consequences of defaults, such as increased enrollment in retirement plans. However, as we discuss subsequently, a second generation of defaults research has also highlighted their potential downsides. As Beshears et al. (2009, p. 192) observe, “Defaults are not neutral—they can either facilitate or hinder better savings outcomes.”

Why Are Defaults Effective?

Three mechanisms with different ethical and practical implications are thought to drive default effects: implied endorsement, cognitive biases, and effort. We briefly describe these mechanisms in the following subsections.

Do Defaults Maximize Consumer Welfare?

A consequentialist approach suggests that one possible resolution of the quandary defaults present would be to pick the default that would produce the greatest overall goodness of the consequences, perhaps best conceived as maximizing consumer welfare. However appealing this may sound, there are at least two problems with this approach. First, the firm's and the consumer's interests are not necessarily aligned. Without an incentive, firms may find it difficult to engage in the kind of perspective taking needed to increase the welfare of another while decreasing their own.

Second, an outcome that maximizes consumer welfare overall may be suboptimal for some consumers in a context in which there is heterogeneity of preferences. Thus, a marketer may set a default consistent with consumer welfare maximization, but some minority of consumers will still be dissatisfied and possibly harmed. This potential for a “tyranny of the majority” is a classic weakness of consequentialist reasoning (and a catalyst for the development of nonconsequentialist theories). For example, the default for auto purchasers in the United States is to have an air bag installed in all new vehicles. Although this feature has produced a net savings in lives, it has endangered small-framed women as well as children. The welfare-producing benefits have mostly accrued to large-framed men; moreover, there is some speculation that those who are more likely to be in accidents—such as those who are inebriated—are particular beneficiaries. Thus, although welfare is improved on average, there are identifiable winners and losers and questions of responsibility. Similarly, recent research on defaults in the context of retirement savings has indicated significant negative welfare consequences: default settings were not well-suited to some of the people affected, who regretted their decision if they were defaulted into the program rather than if they chose it actively (Brown, Farrell, and Weisbenner 2011). Brown, Farrell, and Weisbenner (2011) suggest that the welfare implications of defaults depend critically on why people go with the default.

For situations in which default settings are accompanied by heterogeneity of consumer preferences, questions of marketer responsibility surely loom larger if the explanations for default effects are more attributable to nonconscious consumer influences, such as cognitive biases. The explanations for default effects are all the more salient in nonconsequentialist ethical analysis. As we show in the next subsection, even if defaults enhanced consumer welfare for all consumers, they remain ethically problematic because of their implications for consumer autonomy.

Do Defaults Violate Consumer Autonomy?

Defaults can both enhance and reduce consumer welfare. Setting defaults to maximize consumer welfare might appear to resolve ethical issues. However, a duty-based nonconsequentialist perspective suggests otherwise.

Various marketing ethicists have identified a duty of marketers not to mislead consumers (e.g., Laczniak and Murphy 1993; Martin and Smith 2008), and the American Marketing Association Statement of Ethics identifies honesty and openness as basic values required of marketers. ¹ The “implied endorsement” explanation of default effects suggests that defaults may mislead some consumers (e.g., to believe that the default is going to be better for them). However, some consumers (e.g., the “market savvy”) will be skeptical of marketers and view defaults as persuasion attempts (Brown and Krishna 2004).

Defaults also can be at odds with the consumer's right to choice, one of the four basic rights identified in a landmark speech by President John F. Kennedy in 1962 (Lampman and Douthitt 1997). Smith (1995), in reference to social contract theory, proposes a marketer's duty to ensure that consumers are capable of exercising informed choice. Whether due to implied endorsement, cognitive biases, or, to a lesser extent, effort, all three causes of default effects suggest that defaults can be “hidden persuaders” that are inconsistent with consumers exercising choice; thus, all three fail Smith's (1995) “consumer sovereignty test” (under which marketers ascertain whether consumers have sufficient capability, information, and choice). In summary, a nonconsequentialist, duty-based perspective on defaults suggests that they have major implications for what is best described as “consumer autonomy”: the right of consumers to make their own decisions.

The term “autonomy” comes from the Greek words autos (self) and nomos (rule or law) and, when applied to people, refers to their decisions and actions being their own. As Dworkin (1988) states, it is a moral, political, and social ideal. Autonomous people are more than simply self-determining, as Dworkin's (1988, p. 20) seminal analysis observes:

Autonomy is conceived of as a second-order capacity of persons to reflect critically upon their first-order preferences, desires, wishes, and so forth and the capacity to accept or attempt to change these in light of higher-order preferences and values. By exercising such a capacity, persons define their nature, give meaning and coherence to their lives, and take responsibility for the kind of person they are.

Dworkin uses the classic story of Odysseus—tied to his ship's mast so that he can resist the calls of the sirens—to explain the second-order reflection inherent in his conception of autonomy. Autonomy means that people can have a preference about their preferences in light of how they want to live their lives. For this reason, it is possible for autonomy to be maintained in the face of interference (or even coercion) that infringes on the voluntary character of one's actions. Dworkin (1988, p. 14) writes that “not every interference with the voluntary character of one's actions interferes with a person's ability to choose his mode of life.” Thus, some loss of liberty still may be consistent with Dworkin's conception of autonomy. Consider, for example, life-saving medical treatment rendered without patient consent in emergency situations (Dworkin 1988, p. 116).

Consumer autonomy involves people's self-determination as consumers. It reflects preferences about preferences (“metapreferences”) as well as immediate needs and wants. Thus, it can be conceived as accommodating both consumers who would want to always have as much choice as possible and those who might prefer to have their choices curbed (e.g., due to antimaterialistic values). ²

Under defaults, the consumer generally cedes some independence of choice to the marketer, and consumer autonomy is diminished. This may not be the case when the default effect is due to effort or to a correct assumption of an endorsement by the marketer; consumer autonomy may be maintained in these instances because the consumer is acting consistently with metapreferences if he or she desires the convenience and ease of decision making or chooses to defer to the marketer's judgment. However, autonomy would seem to be reduced when cognitive biases are the explanation for the default effect or when consumers do not view the default as a choice (e.g., the costs of opting out are viewed as prohibitively high). The same is true for cases of hidden defaults, in which consumers are not aware of the possibility of choice (e.g., not realizing that they can switch long-distance calling providers). Yet is autonomy always maintained when consumers recognize choice (i.e., when they understand that they can reject the default option) or, more simply, when there is a requirement for active choice?

What about Active Choice Instead?

Our examination of the theoretical basis for default effects suggests they raise ethical concerns from at least two quarters. First, although consequentialists might be somewhat reassured if defaults are set to maximize consumer welfare, this theory assumes a potential willingness of marketers to act inconsistently with their own interests when they are not fully aligned with those of consumers. Furthermore, because of the potential heterogeneity of consumer preferences, maximizing consumer welfare overall may entail diminished welfare for some minority of consumers. Second, because defaults can be hidden persuaders, particularly when they are attributable to cognitive biases, they undermine consumer autonomy, and this in itself is ethically problematic even if defaults were to maximize consumer welfare across all consumers in equal measure. This leads us to ask: What about active choice as an alternative to defaults?

Bovens (2008), a moral philosopher, has analyzed the ethics of Thaler and Sunstein's 2009 book, Nudge: Improving Decisions About Health, Wealth, and Happiness, drawing on their “Cafeteria” and “Save More Tomorrow” examples as paradigmatic cases. “Cafeteria” uses the placement of food items to nudge healthy eating. “Save More Tomorrow” nudges greater savings by asking employees to commit in advance to putting the next year's pay increase in their pension fund. Although these examples are not defaults as we have defined them, they are comparable cases in which the choice architecture has been selected to increase the likelihood of a particular outcome, and similar psychological mechanisms are at work (e.g., endowment effect).

As Bovens (2008) observes, a central consideration is the matter of who is doing the nudging. When people are in control of themselves, nudges are unproblematic. This is most apparent in the case of nudges people consciously establish to govern their own behavior (e.g., placing healthier food at the front of the refrigerator), consistent with our previous discussion of autonomy as a second-order capacity. People are far less in control if a government agency or company is doing the nudging. Nevertheless, Bovens acknowledges the potential appeal of nudging as a tool of social policy, particularly for addressing urgent problems such as obesity and for situations in which policy messages are competing with contradictory messages from marketers who may also be using nudge techniques (such as those that promote unhealthy eating).

Bovens (2008) differentiates between nudges and the use of social advertising, including subliminal advertising, by the state: Social advertising aims to achieve behavior change by providing information or through emotional appeals, and this attempt is obvious to the observer. Subliminal advertising, as we note at the beginning of the article, aims to influence behavior without the observer being aware. Bovens views nudges as falling between the two and argues that such nudges by the state are more morally permissible when each nudge is, in principle, transparent, “to ensure that everyone can unmask the manipulation if they wish…. This protects the rights of the minorities who do not wish to be so manipulated and it keeps a check on the government” (Bovens 2008, p. 15). Although this potential for unmasking provides greater control over the use of nudges, they remain morally problematic for Bovens in other ways—principally, when the resulting choices are not consistent with consumers' “general preference structure” (p. 217) and are not reflective of “genuine preference change” (p. 207).

At the root of Bovens's (2008) and others' (e.g., Singer et al. 1991) critiques is a belief in an overall preference structure. Bovens argues for greater individual control so that people are better able to ensure that choices are kept in line with their overall preference structure and reflect what they “genuinely” prefer. However, Bovens's (2008) argument ignores the larger body of research (of which Nudge is a part) that questions the assumption of rationality and active choice in many areas of human judgment and decision making. Thus, in “Cafeteria,” for example, the bias (to select items placed earlier and at eye level) does not go away if there is no choice architect deciding what goes where; some item must come first; there is always a “starting point” (Thaler and Sunstein 2003). The case is similar for defaults: in some respects, they too are unavoidable. ³

In contrast to the classical rational choice paradigm, decision researchers have shown that consumer choice is often constructed (Lichtenstein and Slovic 2006; Payne, Bettman, and Johnson 1992). Accordingly, one cannot conceive of an “overall preference structure” because it does not exist; consumers often do not have well-defined existing preferences. Bounded rationality, limited processing capacity, cognitive biases, and interactions with the task environment all lead to preferences being constructed rather than merely revealed, with various strategies used contingent on task demands (Bettman, Luce, and Payne 1998). This is not to say people do not have firm and stable preferences for some objects; it is simply more likely to be the case with objects and experiences that are already familiar to them. Furthermore, empirical research has demonstrated that people can generate preferences independent of conscious thought, through preconscious or nonconscious processing (Fitzsimons and Williams 2000; Janiszewski 1988; Lynch and Srull 1982).

Active choice might be advanced as an alternative to defaults on the assumption that it would be more likely to lead to welfare maximization and to more autonomous decisions that are consistent with some conception of a consumer's overall preference structure. However, this assumption ignores the likelihood that, under the active choice alternative, the consumer might not have established preferences, and decision making is by constructive choice and subject to nonconscious processes. Thus, active choice also might not result in an outcome that maximizes consumer welfare or is consistent with consumer autonomy. Indeed, in some contexts, active choice is less likely to produce a welfare-maximizing outcome than are defaults (Sunstein and Thaler 2003), suggesting that perhaps paternalism is the answer, at least relative to a consumer welfare criterion. Perhaps defaults should be set with the consumer's best interests in mind?

What if Defaults Are Set for the Consumer's Own Good?

Because defaults change choices, they can violate consumer autonomy and do so by serving the marketer's interest and not the consumer's. Yet this violation of autonomy can also occur paternalistically, for the consumer's own good. Paternalism is the “interference with a person's liberty of action justified by reasons referring exclusively to the welfare, good, happiness, needs, interests, or values of the person being coerced” (Dworkin 1972, p. 65). This trade-off between maximizing autonomy and consumer welfare is present in classic illustrations of paternalistic interventions by the state, such as laws requiring seat belt use in cars or helmets for motorcycle riders. Dworkin (1988, p. 123) explains that paternalism involves the “usurpation of decision making, either by preventing people from doing what they have decided or by interfering with the way in which they arrive at their decisions.”

Sunstein and Thaler (2003, p. 1161) argue strongly in favor of a form of paternalism, urging that default rules “should be chosen with the explicit goal of improving the welfare of the people affected by them.” Their rationale (2003, p. 1162) is that “in some cases individuals make inferior decisions in terms of their own welfare—decisions that they would change if they had complete information, unlimited cognitive abilities, and no lack of self-control.” Moreover, given their belief in constructed preferences, the authors suggest that in many situations, there is no alternative to a kind of paternalism. Somebody must set the default. This “weak paternalism” is still impossible to avoid even where planners avoid defaults and require active choices, because some people would choose not to choose.

Sunstein and Thaler (2003, p. 1162) advocate “libertarian paternalism,” under which, they suggest, paternalistic policies that are “self-consciously attempting to move people” would be acceptable from a libertarian perspective if choices are not blocked off and impose only “trivial costs on those who seek to depart from the planner's preferred option.” This is the central idea in Nudge (Thaler and Sunstein 2009). In setting defaults, marketers potentially could have “libertarian benevolence” in mind, whereby default rules are “enlisted in the interest of vulnerable parties” (Sunstein and Thaler 2003, p. 1162). It remains libertarian because the design makes it easy to reject the default option.

Nevertheless, even libertarian paternalism violates autonomy. Sunstein and Thaler (2003, p. 1167, footnote 22) acknowledge this concern up to a point, though they assert that it is “fanatical” in settings such as obesity “to treat autonomy … as a kind of trump not to be overridden on consequentialist grounds.” They continue by claiming respect for autonomy in stating that “autonomy is adequately accommodated by the libertarian aspect of libertarian paternalism.” If the effect of defaults results from laziness or implied endorsement, perhaps giving people the ability to change their choice does provide sufficiently for autonomy. However, this is not the case for our cognitive biases explanation of default effects, and yet Sunstein and Thaler (2003, p. 1168) also maintain that human judgment is profoundly biased: “People fail to make forecasts that are consistent with Bayes' rule, use heuristics that lead them to make systematic blunders, exhibit preference reversals (that is, that they prefer A to B and B to A), suffer from problems of self-control and make different choices depending on the framing of the problem.”

Consumers who embody this characterization would not be as “free to choose” as Sunstein and Thaler (2003, p. 1161) would have it, due to the very biases these authors view as fundamentally relevant. If loss aversion underlies the preference for default options (as Thaler, Kahneman, and Knetsch [1992] argue), the use of defaults as instruments of policy inevitably compromises autonomy. Thaler and Sunstein (2009) provide a chapter on objections to the use of nudges, including the slippery slope (modest paternalism gives way to the adoption of a more aggressive form), evil nudgers and bad nudges, and the right to be wrong. However, they still claim that their goal is “to allow people to go their own way at the lowest possible cost” (2009, p. 252). Although we might endorse this view, it would be a mistake to think that people who express preferences through defaults are truly “going their own way.” For those who are unable to detect or resist default effects—especially those due to cognitive biases—defaults may not offer the freedom of choice that libertarian paternalism suggests.

Nevertheless, we are again left asking what the alternative is. Consumers are often not so obviously going their own way through active choice, and indeed, they might be making worse decisions as a consequence. Viewed this way, we question how much more autonomous consumers are under active choice conditions compared with defaults given constructive consumer choice and the many sources of nonconscious influences. They might be no more autonomous (and arguably, less so) in some contexts. Active choice is likely to be preferable in familiar choice decisions in which there are established preferences but less so when consumers do not know their preferences, have difficulty with the decision, do not receive feedback, or cannot assume that the market corrects their mistakes (Thaler and Sunstein 2009).

More specifically, in the context of retirement saving, research has shown that requiring people to make active choices is optimal relative to a default of automatic enrollment or nonenrollment (uncoerced active choice) when consumers have a strong propensity to procrastinate and savings preferences are highly heterogeneous, whereas default enrollment is preferable to compulsory active decisions under conditions of financial illiteracy (Carroll et al. 2009). Building on this work, Keller et al. (2011) find support for enhanced active choice (forced choice supported by the endorsement of the preferred option by the choice architect) in the context of medication adherence.

Although concerns about consumer autonomy under defaults might be diminished by this argument about the limitations of active choice, we must also consider the role and intent of the person setting the default. Default settings have the potential to increase consumer welfare, but this assumes that the paternalistic choice architect is setting welfare-enhancing defaults. ⁴ Moreover, it is surely preferable from a public policy perspective that this intentional use of defaults to enhance consumer welfare should also be capable of being “unmasked,” as Bovens (2008) suggests and we further discuss subsequently.

Remedies to Enhance Consumer Welfare

Although an active choice requirement is a viable response to the dilemma of defaults, it could deprive consumers of their welfare-enhancing aspects—that is, efficient, assisted decision making. We propose that marketers can use defaults in many contexts in ways that exploit their potential benefits to enhance consumer welfare while addressing concerns about the misassignments (“incorrect” decisions) that may occur due to heterogeneity of consumer preferences. Such defaults may be paired with safeguards that minimize any diminution in consumer autonomy. We also consider the incentives that may be necessary for the choice architect to make a welfare-enhancing decision (rather than use defaults in a self-serving, manipulative way) and again note that active choice is still preferable in many decision-making contexts.

Default effects have certainly become more prevalent over the course of the twentieth century and into the twenty-first century as consumer choice has expanded; contrast Ford's Model T (“any color so long as it's black”) with the multiplicity of choices provided to today's automobile purchasers. As demonstrated by work on reason-based choice (Shafir, Simonson, and Tversky 1993), the addition of more options increases the tendency to remain with the status-quo default. Most important, perhaps, is that the expansion of choice significantly increases the effort involved in making a decision, even if decision makers use adaptive strategies (Payne, Bettman, and Johnson 1993). Is it possible to create new kinds of defaults that retain the benefits of default-guided decision making while also preserving some measure of consumer autonomy and leading to good outcomes?

An obvious conclusion, from a consumer welfare perspective, is that ignoring defaults can be a mistake for firms, consumers, and policy makers. Welfare may be reduced when defaults are set without regard to the consequences for choice, which Sunstein and Thaler (2003, p. 1202) refer to as “inept neglect.” We know of a large auto manufacturer that allowed its consumers to configure orders online. The manufacturer had, inadvertently, set the default alternative to the least expensive option for every choice. Not only did this strategy fail to maximize profits for the firm, it also hurt consumer welfare: when making choices in the absence of a default, customers systematically chose more expensive options. It is likely that the wrong default left both the marketers and the customers worse off. A better choice of default would generate a Pareto improvement in welfare to both parties. By assuming that active choice better reflected consumer preferences, in this case, the marketers' overweighting of consumer autonomy seemingly resulted in losses to both parties.

Because ignoring defaults is not an option—not least because there simply is not a world without defaults—we discuss the use of benign defaults as Sunstein and Thaler (2003) propose. We subsequently endorse a new kind of default: “smart defaults” (see also Goldstein et al. 2008).

Benign Defaults

Consistent with Sunstein and Thaler (2003; see also Thaler and Sunstein 2009), we believe that the more problematic cases are typically those in which defaults are set in a way that does not maximize consumer welfare. A certain number of people will be dissatisfied under almost any default. However, if the default is set to the choice most people would be presumed to make when making an active and unconstrained decision, the greatest number benefit.

Implementing such policies is not as simple as it seems, because providing the wrong defaults may have a great cost. Consider the case of organ donation (Johnson and Goldstein 2003). Governments deem organ donation to be welfare maximizing, and polls in the United States show that most people approve of organ donation. However, only a minority of Americans have joined organ donor pools, and only a minority have agreed to be donors in forced-choice situations, such as at motor vehicle registration agencies. Should stated preferences (polls) or revealed preferences (forced-choice questions about joining donor pools) be used to determine which choice is welfare maximizing? Policy makers and marketers also must look beyond the number of people affected by various defaults (as we have done here) to other stakeholders and the broader consequences. The families of willing organ donors may care little if their kin are defaulted into not being donors, whereas the families of unwilling donors may care a great amount if their loved ones are harvested for organs. Even if having more donors despite a few outrages is arguably better for societal welfare (including organ recipients), the negative press arising from the incidents could cause voters to put an end to the opt-out system, thus decreasing societal welfare.

Less controversial is the use of benign defaults by Disney theme park restaurants in providing healthy side dishes as the default offering with a main meal and leaving it to customers to specify a less healthy alternative if they so choose (Walt Disney Company 2009). Disney's policy helps address the obesity epidemic by providing healthier meals but makes it easy for customers to opt out, consistent with Thaler and Sunstein's libertarian paternalism (also see Halpern, Ubel, and Asch 2007).

In their approach to benign defaults, Sunstein and Thaler (2003) focus primarily on public policy maker use of defaults to identify welfare-enhancing interventions. They do, however, acknowledge the relevance of these interventions to the private sector. In their later work, Thaler and Sunstein (2009, p. 242) also recognize the possibility of private sector exploitation of defaults that reduce consumer welfare—in which private choice architects are “evil nudgers”—and include examples of welfare-enhancing nudges that firms might adopt.

Sunstein and Thaler (2003) propose four types of interventions: (1) minimal paternalism, in which a planner constructs a default rule with the goal of influencing behavior, but it is costless or nearly costless to depart from the default plan (this intervention is most consistent with their idea of libertarian paternalism); (2) required active choices, in which the planner is unsure which choice will promote welfare and so forces people to choose explicitly (though this can be problematic in many contexts, as we have discussed); (3) procedural constraints, which typically require more effort and are designed to ensure that departing from the default is voluntary and rational rather than a function of defective decision making (e.g., due to a lack of experience); and (4) substantive constraints, which allow people to reject the default but only on certain terms and potentially at considerable cost as well as effort.

Planners also have the option of denying choice altogether on the basis that people will reject a default in error. This option is more typical of public sector use of defaults; Menard (2010), for example, rejects libertarian paternalism in certain health care contexts in favor of more coercive measures (e.g., banning the sale of items with health-threatening ingredients). This is also, in some respects, what companies do in requiring consumers to read terms and conditions before committing to purchase: they do this arguably to enhance consumer welfare but more likely to reduce scope for subsequent complaints or litigation. It is also what a company does in determining a set of product attributes over which the consumer has no choice, although consumers certainly have choice when comparing different products of competing companies.

In determining the appropriate intervention, there are two approaches that seem to apply to marketers as well as to a public policy context (Sunstein and Thaler 2003). The first is a cost–benefit analysis that evaluates the gains and losses associated with the program options. If feasible, this approach would involve an objective assessment of which option maximizes consumer welfare and thus how planners should set defaults. The organ donation example illustrates the challenges this approach poses, and a standard critique of consequentialist ethics is the difficulty of forecasting all potential good and bad consequences for all affected parties. The second approach is to adopt one of the following rules of thumb: the approach that the majority would choose if (unconstrained) explicit choices were required and revealed (but what of the minority?); a forced-choice approach (but some would not choose and others would not make “good” choices, and this also abandons the efficiency of defaults); or an approach that minimizes the number of opt-outs (but this might result from cognitive biases). Ultimately, however, the approach of using these various benign default solutions is suboptimal in our view relative to what we call “smart defaults” that intentionally set out to exploit marketer expertise in mitigating default effects in marketing contexts. From a public policy perspective, this approach also has the virtue of not requiring policy maker intervention, if it can be assumed that there is sufficient incentive for marketers to follow it.

Smart Defaults

Marketers are in the business of understanding consumers' needs and predicting their behavior. Defaults can be set in a way that takes advantage of that knowledge. Consider the air bag example we used previously to illustrate heterogeneity of consumer needs. If the deployment of the air bag could react to the type of occupant in the seat, consumer welfare would be increased. Thus, many “Advanced Air Bag Systems,” required on all new vehicles in the United States since 2006, are designed to sense the weight of the seat occupant to determine whether to activate the air bag. This is a smart default: it uses customer information to generate individualized options that are more likely to be optimal for any specific customer than a generic set of alternatives. In helping customers make better decisions about the purchase of retirement investments, a smart default might be based on a simple linear model incorporating the purchaser's age, family status, and intended age at retirement. Other factors, such as the investor's risk preferences and loss aversion, also can be included. Such defaults would not suit all consumers perfectly but are superior to the traditional default contribution of not providing a safety net or the more recent “one-size-fits-all” default that many firms suggest.

The challenge for smart defaults is to gather enough information sufficiently quickly to produce a better-customized default than the one-size-fits-all approach. There will always be a trade-off between the amount of information gathered and the accuracy of the default calculation, but existing market research technology should enable firms to address this problem. From both a consequentialist ethics and a caveat venditor (seller beware) perspective, smart defaults are a dominant option. ⁵ We believe there may also be an advantage to the firms as well, assuming they (1) meet the challenge of creating the right smart default and (2) are able to communicate the effect of defaults on choice and to convince consumers that their choices are better using smart defaults (which is perhaps much more difficult). In this case, firms may profit from the long-term loyalty generated by increased consumer satisfaction.

Smart defaults require the presence of consumer-specific data, some of which may already be known to the marketer (e.g., age, gender, referring website, prior purchase history) and some of which might be collected explicitly to generate the default. ⁶ The benefit of smart defaults is that they return to the original idea of marketing as the understanding and meeting of consumer needs, including the differences across consumers evident in market segmentation. What is novel is that smart defaults assume that the firm must understand consumers better than consumers understand themselves at the beginning of the decision process. Much like a firm uses market research to produce products that meet consumer needs, smart defaults suggest that firms must produce decisions that meet the consumers' needs as well. This was not the case when our auto manufacturer, which had picked the least expensive default for every choice, failed to meet the needs of most consumers. The smart default design—selecting the right engine, body style, and accessories for both the high-performance connoisseur and the parent of a large family—would be a smart default that better meets those needs. In this way, market orientation becomes evident in consumer decision-making processes as well as outcomes. Firms thus use consumer information more effectively in meeting consumer needs and, in principle at least, would be rewarded in the market for so doing.

Similarly, in light of the Dodd–Frank bill introduced in the United States in 2010 to address “abusive financial products” (Morgenson 2010), retail banks aiming to be market-oriented could go beyond the provisions of the law and use smart defaults to better serve their customers. For example, in regard to overdraft charges, customers at automated teller machines making cash withdrawals could be given advance notice that the withdrawal would result in an overdraft, the amount of the fee that would be charged, and whether a smaller amount withdrawn would result in avoiding an overdraft fee. The default decision presented to consumers could be the smaller withdrawal amount, while providing minimal-effort alternative options of canceling the withdrawal or proceeding with knowledge of the fee. This is a smart default because it uses consumer information (account balance and terms) to set the default and guide the decision process.

Smart defaults aim to minimize misassignments (incorrect default settings) and to do so consistent with our three explanations for default effects. To the extent that default effects are due to cognitive biases or effort, the marketer aims to work with those biases and the consumer's desire for minimal effort to provide the correct assignment and a greater probability of a welfare-enhancing outcome. More fundamentally, to the extent that implied endorsement is at work, smart defaults provide consumers with exactly that: the endorsement of the marketer who is recommending the “right” decision. Moreover, the right decision for one consumer might not be the right decision for another, and thus, smart defaults can respond to consumer heterogeneity by aiming to identify the correct default using consumer characteristics.

Smart defaults can become even smarter by adapting to information provided by the consumer as part of the decision-making process. This approach uses each choice in a series to set other defaults and is particularly well suited to online contexts. Instead of the previously mentioned auto manufacturer's unfortunate choice of selecting the least expensive options as defaults, a more appropriate, benign, welfare-enhancing default might be the most likely option that a customer would select in the absence of a default.

Thus, marketers can supplement smart defaults that rely on extant and possibly rudimentary knowledge about the consumer by gathering additional data generated in real time. The marketer would present the options with defaults that represent the best guess of what might be chosen conditional on what the consumer has chosen to date. For example, someone who chooses a car with a powerful engine may be more likely to choose certain colors (red) and other consistent options, such as sporty wheels, a sports handling package, and performance tires. Note that unlike the widespread use of packages of options to limit choice, the idea of an adaptive default preserves considerable consumer autonomy (within marketer-determined boundaries) and strikes a balance between providing more choices and providing the right choices. It also addresses concerns of choice overload by limiting the options it presents to the consumer. Because the number of decisions that must be made is also reduced, ego depletion effects in choice might also be minimized (Baumeister, Muraven, and Tice 2000). In this use of real-time data, smart defaults can become akin to electronic recommendation agents or “smart agents” (Ariely, Lynch, and Aparicio 2004).

In proposing these smarter defaults, we assume that consumers will make good decisions as they go through the process. It is possible, however, that consumers will make mistakes that cause errors to compound, with the more limited consumer choice offerings restricting consumers' scope to correct earlier errors. The solution is to ensure the consumer's ability to switch out of the default to make an active choice at any time (e.g., “go back to start”), although we accept that this may not be a perfect remedy, especially for more vulnerable consumers.

As in other contexts in which marketers use consumer data, privacy issues also may arise, such as online tracking used to target advertising (Goldfarb and Tucker 2011; Goldman 2006). Although these privacy issues are beyond the scope of this article, at minimum, marketers will need to be attuned to how they gather and use consumer data as much in the context of smart defaults as in other contexts, especially those in which data are collected and used without consumer awareness or consent and those involving particularly sensitive behaviors and product categories.

Thus, although smart defaults have many merits, there may be some problems associated with their implementation. These clever defaults seem to be choices or judgments that the firm makes for the customer. Who is to blame when a smart default does not fit the consumer, such as when the air bag is inappropriate: the manufacturer, for having a poor algorithm, or the consumer, for passively accepting the default? Furthermore, can we assume in the first place that the incentives are sufficiently strong for marketers to create consumer welfare-enhancing smart defaults? Are the rewards so evident if the (smart) default effect occurs without awareness? Moreover, given the biases associated with motivated reasoning (Kunda 1990), marketers may be too ready to believe that default settings that serve their interests also enhance consumer welfare. Are penalties likely to be invoked when marketers use defaults to serve their own interests at the expense of consumer welfare? What of the privacy implications of the data collection and use associated with the “smarter” smart defaults? To some extent, at least, we might rely on others, such as consumer-interest nongovernmental organizations or the media, to discipline or regulate companies that manipulate consumers for their own ends or violate privacy rights. We address these points and return to the importance of the ability to unmask the default in the next section.

Other Remedies

In some circumstances, it might be appropriate to require procedural constraints (Sunstein and Thaler 2003) to reduce the prospect of consumers rejecting a welfare-enhancing default. These constraints typically raise the cost of departing from the default by requiring greater effort (e.g., software companies that provide recommended settings when installing software). In other circumstances, particularly those in which the default is not welfare enhancing for at least some consumers, warnings and disclosures may be warranted. Hidden defaults (i.e., defaults that do not inform customers that they have a choice of certain options) may be inappropriate or at least require disclosure if there are potentially significant negative consequences for consumer welfare.

In light of concerns about consumer autonomy and consumer welfare, defaults should, in general, be capable of being unmasked. It should be possible for observers in civil society (e.g., nongovernmental organizations, the media) as well as regulators to be able to unmask the use of defaults and hold the marketer accountable (e.g., through adverse publicity), even if the use and effects of these defaults are not immediately obvious to consumers. In some cases, it may be appropriate to warn consumers of default effects, not unlike how curved rearview mirrors come with warnings that they alter perceived distance. However, warnings would be of less practical value as a remedy under the assumption that cognitive biases are at work. If default selection reflects implied endorsement, it might be appropriate to require warnings to the effect that the default option is not endorsed by the company when it is not the consumer welfare–maximizing option (e.g., “default settings do not constitute a recommendation and may not be the preference of a majority of consumers”). In some instances, this type of warning might need to be mandated. However, in this context and others (e.g., in regard to paternalistic remedies), more research is required to increase understanding of the relative contribution of the different theories of default effects to default outcomes. One promising approach is the application of query theory, which Dinner et al. (2011) suggest can be used to isolate and mitigate causes of default effects.

More draconian, but arguably warranted in some contexts, are regulations that prevent the use of defaults or that restrict marketers from using a consumer welfare–reducing default or from unfairly loading the costs of not following the default. In view of defaults' demonstrated powerful effects, regulatory agencies should closely monitor their use. In some cases, defaults might be blocked, as in Microsoft's settlement with the European Commission, whereby users are required to make an active choice of browser software when loading its Windows operating system (BBC News 2010). In 2008, new European regulations prohibited airlines from adding insurance and other services as the default option on online ticketing websites, requiring instead that they be sold on an opt-in basis (Bachelor 2009). However, this move was primarily directed at increasing transparency in pricing rather than eliminating welfare-reducing default effects. A more general response would be to educate consumers such that they are better informed of how they might consciously respond to defaults (e.g., by encouraging discussions in various consumer media about examples of powerful default effects and the need to recognize both that a choice is being made and whether that choice is the best for the person making it). However, we note that research has often shown that educating consumers about the biases in decision making is ineffective (Thaler and Sunstein 2009).

Further Research

Although the powerful effects of defaults are well established, further research is needed to shed more light on their underlying causes as well as the possible remedies proposed, including consideration of their use relative to active choice and all the limitations thereof. The three theoretical explanations we identify for default option preference are well supported, but the relative contribution of each is less certain in regard to characteristics of the context or respondent (e.g., do individual difference variables predispose some people to be more inclined to believe that the default setting is endorsed by the marketer?). Accordingly, market segmentation would seem to be a key consideration in understanding differences in responses to defaults and in the requirement and form of remedies adopted. Thus, high levels of buyer expertise in a category might reduce concern about harmful default effects and eliminate a requirement to protect consumers through the use of smart defaults. Equally, these defaults could save time by identifying buyer expertise and fast-tracking those with high expertise, while research could explore the use of electronic recommendation agents as a possible form of implementation of smart defaults for buyers with less expertise.

Smart defaults are not appropriate solutions for all consumers. Research is needed to better understand not only when they are a useful remedy for reducing harmful default effects but also when consumers are likely to appreciate them and how firms can weigh their costs and benefits. As we show in the current study, choosing the right default can enhance customer satisfaction and reduce various risks for customers and the firm, but the trade-offs involved must be better understood (e.g., when are risks for the firm or the customer sufficiently large enough to require that consumers make active choices rather than be presented with defaults?). More narrowly, can marketers develop smart defaults that gather the information quickly enough to produce a superior customized default relative to, say, a benign default alternative?

Research could also explore the trade-offs consumer protection agencies face in regard to whether intervention is required over default effects. Certainly these agencies should be attuned to the possible manipulative use of defaults (e.g., the extent to which companies abuse default effects by setting defaults to their advantage and in ways that disadvantage consumers). In the Facebook example we described at the outset of this article, the firm was criticized for choosing a default setting that gave it an advantage with its advertisers at the expense of consumer privacy. In this instance, public backlash disciplined the firm. However, we cannot necessarily assume that markets will self-correct even when consumers object to default settings. Many commentators (e.g., Consumer Reports 2012) believe that Face-book has continued down the path of privacy invasion despite a public backlash and is able to do so because of its monopoly of power and the lack of competitive market forces. No doubt there are many lower-profile or undiscovered instances of default settings to the detriment of consumers. Research could also explore consumers' privacy concerns in the context of smart defaults that collect and use consumer information to generate default settings.

More generally, there is a need for further research on the policy implications of what are now more firmly established “hidden persuaders” in marketing. Default effects are not the only source of consumer choice without awareness, and greater research attention from a policy perspective could be given in particular to consumer choice under conditions of mere measurement, framing, anchoring, endowment, and placebo effects.