Default settings - the most powerful tool in the behavioural scientist’s toolbox?
“Of all of the tools in the choice architect’s repertoire, default rules may be the most promising; they are almost certainly the most discussed. Whether the area involves savings behavior, poverty reduction, or the environment, default rules have had significant effects on outcomes.”
Cass Sunstein, 2017
The concept of creating or changing default settings is possibly one of the most powerful ‘nudges’ in any behavioural scientist’s ‘toolbox’ - one of the surest ways of changing people’s behaviour. By defaults, we mean that when people are presented with default options already set, or are automatically enrolled into something, they tend to accept them. We accept the status quo and go with the flow, without considering other options or investigating further. A very simple example is the default (factory) password setting for our mobile phones or our voicemail, which an astoundingly high proportion of people never get around to changing.
In this article, the second in our series on the new frontiers of behavioural science, we look into the concept of default effects more closely and delve into the latest research and findings to better understand the concept - how we can change or create new default settings. Firstly, we summarise just how powerful default settings can be and look at four simple rules or conditions potentially explaining why they are often so effective in influencing behaviour. Secondly, we look at some cases where they have not been successful, to see what we can learn. Thirdly, we explore new findings about potential drawbacks or what nuances might make them more or less effective- the contexts, choice architecture or characteristics of people - that might shape whether a default is accepted or not, particularly in the longer term.
Yesterday - where we were
The power of defaults in shaping behaviour
Richard Thaler and Cass Sunstein drew considerable attention to the concept of defaults in their 2008 bestselling book ‘Nudge’, classifying defaults as one of six important types of nudge or choice architecture. Thaler commented, “The combination of loss aversion with mindless choosing implies that if an option is designated as the “default”, it will attract a large market share. Default options thus act as powerful nudges.”
Just how powerful is illustrated in these next examples. Research by tech expert, Nir Eyal, has found that two thirds of smartphone owners never change their notification settings, usually defaulted to ‘on’; they stick to the default settings. Similarly, in a recent UK survey, 82% of people said they had never changed their Wifi router's default administrative password. Findings from research by Jared Spool also found a high degree of status quo bias in people’s computer settings. He analysed the settings of people’s software and applications and found that less than 5% of the users he surveyed had changed any settings at all. More than 95% had kept the settings in the exact configuration as when the programme was installed.
When he interviewed a sample of the users, they all told him the same thing: they assumed Microsoft had delivered it with certain settings for a reason, therefore who were they to change it? “Microsoft must know what they are doing” several of the participants told him. In fact, on speaking to Microsoft, some of the settings were not recommendations at all, but developers waiting for further guidance before making a setting the default. Companies often realise the immense value of being the default choice. For example, in 2016 Google paid Apple $1 billion to remain the default search engine on the iPhone, also paying to remain the default on popular web browsers.
A 2017 study also highlights just how influential defaults - in the form of automatic subscription renewals - can be. When US telemarketing firm Suntasia faced a federal lawsuit after charging hundreds of thousands of customers an average of $239 each for worthless subscriptions, a team of researchers ran an experiment to see which method of cancelling subscriptions worked best. When subscriptions were automatically cancelled by default, 99.8% of customers accepted and cancelled their subscription. However, when they had to take action to cancel, just 36.4% did so.
Of course, one of the most well-known applications of defaults is in the area of enrolment into retirement savings schemes and pensions. Here, employees are automatically opted into a scheme by their employer, although they have complete freedom to opt out if they so choose. Research conducted in the US by Behavioural Economists Brigitte Madrian and Dennis Shea found that auto-enrolling employees into a retirement savings scheme raised enrolment from 49% to 86% and further research has found auto-enrolment boosts participation rates to over 90%.
In the US, auto-enrolment into retirement savings schemes (known as 401(k) private pensions or defined contribution pensions) has boosted enrolment rates dramatically. Automatic enrolment has increased 401(k) participation rates from around 65-75% to 90% or more. The UK also adopted auto-enrolment for pensions in 2012, an action which has led to participation rates of around 86-92%.
The mechanisms behind defaults
If we were to make a wholly rational decision about a choice, we would consider the different options open to us in a considered and balanced way, uninfluenced by the option which has been selected for us. Yet, defaults have a number of powerful psychological effects on us. Researchers have identified three main mechanisms at work:
- Inertia: We tend to stick with the status quo and go along with what has been pre-selected for us. Making a more active, engaged choice can take effort and focused attention. James Choi, Professor of Finance at Yale School of Management says, “It’s about inertia… the status quo is the choice that takes the least amount of effort.” People might also procrastinate over a decision - perhaps if it is complex or unfamiliar, if it is boring, or if they are busy - to the extent that in the end they go with the default because they haven’t reached any other decision.
- A default acts as an implicit recommendation: When we are not sure what to do and lack expertise in the area in question, we may assume the default has been pre-selected for us for a reason. We consider the default as a form of advice, believing it to have originated from those more knowledgeable than ourselves; perhaps peers or experts.
- Loss aversion: Because people dislike losses more than they enjoy gains of an equal size, they might be averse to opting out of a default. What they could lose by switching or opting out will be salient and front of mind, making them reluctant to change.
These mechanisms mean that, broadly, default settings are likely to be most effective under the conditions below:
- The default is seen as an implicit recommendation
- The decision is complex and unfamiliar
- The individual does not have a well-defined preference for any particular option
- The decision is dull or not seen as worth the effort
For example, there is considerable evidence that setting a default which goes against people’s preferences too much results in low take up. One study of UK pensions contribution rates found that only 25% of employees accepted the default savings rate set at 12% (the recommended level but relatively high compared to most rates). 60% of employees chose to shift down to a lower contribution rate. Similar findings are apparent when part of workers’ tax refunds are defaulted into US savings bonds.
Beyond finance, changing the central heating default setting in the OECD building backfired. Reducing the default setting by 1 degree in winter helped to reduce the average setting. But when the default was reduced by 2 degrees, there was a boomerang effect and the reduction in the average setting was smaller. A significant number of people found the temperature cold enough that they overrode the default.
Further understanding of what makes default effects work: The four conditions outlined above are useful at a general level but new evidence and understanding suggests that these rules are not so black and white and that there might be further nuances and variables at play. Whilst they make sense at a topline level, the more we learn about the nuances of applying default settings, the more we understand about how the specific context or topic might affect outcomes.
For example, what might explain the backlash by Dutch citizens in 2016 against the proposed bill to automatically enrol people as organ donors unless they opted out? Before the bill was passed into law, the number of citizens opting out spiked to 40 times the numbers seen in previous months, including people who had previously consented. Are Dutch people more sensitive to what they see as coercion or are there other causes?
Today, researchers are identifying further factors and nuances at play as well, which could impact on the effectiveness of a default. Although default effects are undoubtedly powerful, it’s important to consider the specific context in which they will be applied to identify any potential barriers to their acceptance.
For example, recent research reviews the effectiveness and potential mechanisms behind the default effect. A 2018 study by Jon Jachimowicz, Shannon Duncan, Elke Weber and Eric Johnson reviewed a plethora of studies applying defaults - 58 in total. The majority of studies (46) found a positive and reasonably large impact in participation from changing the default setting; on average increasing participation by around 27%. Notably, this impact was much higher than the impact of other types of behavioural science interventions recently reviewed, such as risky choice framing, using social norms to promote energy savings, or choice architecture to encourage healthy eating. However, there was considerable variation of impact, a handful of studies found no effect (10), and a couple even identified a negative impact - like the boomerang effect highlighted above in the OECD default temperature change initiative.
When the research team analysed what type of mechanism produced the strongest impact, they found that when a default was seen by people as an implicit recommendation and/or was perceived as the status quo - an effect termed ‘endowment’ - interventions produced higher increases in participation. For example, to illustrate the endowment mechanism, one study found that people considered an option more attractive when it was labelled as the ‘status quo’. An interesting example of the recommendation mechanism might be found in typical credit card statements, particularly statements viewed online, which usually present a number of repayment options. Research by the consumer body Which? found that 48% of consumers they surveyed thought that the minimum repayment figure was the amount their credit card provider recommended they repay. And in many online statements, it’s this option that is ‘pre-ticked’ or set as the default.
Surprisingly, the research team could find no evidence that ease and laziness were driving the default effect, although it may be that it’s more difficult to measure and research this mechanism in the studies they analysed. The team themselves note that real-world applications certainly show evidence of variations in ease of opt-out impacting participation rates; if opting out is a hassle or complicated, participation rates tend to be higher. Their findings both support earlier thinking and go against it. They support previous evidence that implicit recommendations are a key driving factor behind the default effect, yet place a question mark over inertia and laziness and introduce a new potential mechanism of endowment.
New thinking on defaults
i) Our past experience with defaults can impact our future acceptance: Research published in 2018 by Thomas de Haan and Jona Linde found that if people have accepted a default option that has resulted in a negative effect of some kind, they are less likely to trust defaults in the future, therefore rendering any default less effective. The opposite is also true; if our experience of defaults has been positive, we are more likely to trust that a default option is a good choice. 
This has implications in the frequency of use of defaults. Rather than using default options at every possible opportunity, it may be wiser to utilise them only when they are most necessary. It also puts pressure on practitioners and policy makers to get defaults right first time - it seems we might only have one chance. Blow it and you lose people’s trust.
ii) People like defaults - or most of them: The finding above is significant, since recent research suggests that people are sensitive to exactly what they are being nudged to do. In 2016, Cass Sunstein polled Americans to see if they approved of particular defaults, such as automatic enrolment in retirement saving plans or automatically registering eligible citizens as voters. Most Americans had no opinion or objection about defaults in general.
However, the purpose and effect of a default mattered: whilst defaults that are seen as legitimate and important like the ones above were widely accepted, they rejected defaults they viewed as manipulative that could conflict with people’s values. For example, 73% disapproved of a default which automatically donated $50 of someone’s tax return to the Red Cross unless they opted out. People expect a default to be well-intentioned and to help them make better decisions, yet not push them into doing something that they are unhappy with.
iii) People still like defaults when they know they are being nudged: One concern and frequent criticism of nudging using defaults regards the often-covert nature of changing the default. In response, in 2015 George Loewenstein and colleagues examined the impact of warning people about being opted into certain medical treatments. Participants in the study were asked to complete a hypothetical end-of-life care choice form.
- Firstly, the form asked them to indicate whether they would want medics to make every effort to prolong their life or, alternatively, focus on palliative care to maximise their comfort and reduce any pain.
- Secondly, participants were asked to indicate their preference for particular medical treatments which can help prolong life, such as feeding tube insertion, CPR or dialysis.
However, some participants saw versions of the form where they had been defaulted into choosing to prolong their life, and others saw forms where they had been defaulted into palliative care. Unlike typical nudge interventions however, participants were informed about the default and were free to accept or reject these pre-selected choices and to opt out.
Surprisingly, knowing about the nudge still did not eliminate the default effect. Just as people who set their watch 5 minutes fast to avoid being late find this works, even though they know the real time, knowing our behaviour is being steered by a nudge still seems to push us in the same direction. Although further research is needed to explore other contexts, it seems that telling someone they are being nudged may not decrease its effectiveness.
iv) Active engagement is often better than a default: A second insight is that defaults can sometimes be a double-edged sword; the main strength of a default is also its weakness because it asks nothing from individuals in terms of engagement. Behaviour change in this case is achieved by relying on our tendency to accept an implicit recommendation, settle for the status quo, and not explore other options, meaning that the choice ‘made’ often requires next to no effort in both thought and action on our part and may leave us feeling uncommitted and unengaged. Not consciously choosing to do something can result in a lower level of commitment and responsibility or ownership. Therefore, in most cases, generating more active engagement might be more beneficial than applying new defaults, as the lack of engagement might work against optimal behaviour in the long term.
Take auto-enrolment in pensions, as discussed above. An individual could be auto-enrolled into a pension in their current job, accept the default fund(s) and the (often low) default contribution rate, but have very little understanding or engagement with their pension. Individuals might not even know or remember they are enrolled in a pension plan, let alone think about how to optimise it for their needs. In the US, this lack of engagement also means that people are cashing out their retirement savings, not understanding how vital it is to keep these savings untouched. One in four households with a defined contribution fund cashes out its savings, meaning that as much as $70 billion is withdrawn from 401(k)s on an annual basis.
Jonathan Rowson of the Royal Society of Arts (RSA) summarises the issue succinctly. A default “changes the environment in such a way that people change their behaviour, but it doesn’t change people at any deeper level in terms of attitudes, values, motivations etc.” He and other social scientists still see a need for a more conscious, thoughtful engagement in some cases, which aims to “foster the transformative learning we need, to make significant and enduring changes to our behaviour.”
v) People will still override defaults when it matters: However, new findings from a study conducted among healthcare clinicians provide evidence that individuals can quickly make a conscious choice if the default is not appropriate. If an individual does have some expertise in an area and the decision is important, they will override a default if they do not judge it to be the best option. David Olshan, Charles Rareshide and Mitesh Patel - based at what is the first ever ‘Nudge Unit’ for healthcare - at the Penn Medicine Center, ran a long-term trial to test whether they could encourage a higher rate of prescription of generic drugs. Generics are notably cheaper, yet almost always equally effective, so they can save considerable amounts of money.
They made a tiny tweak to the prescription order system on the University of Pennsylvania Electronic Health Record system. When doctors select the drug they want to prescribe, they click on a drop-down menu. Previously, branded drugs were listed at the top of that menu and generics at the bottom. The researchers flipped the order so that generic drugs were listed first - effectively making them the default choice. It had an astounding effect. Before the trial began, the generics prescribing rate at Penn Medicine was around 75.0%. Immediately after the change to the drop-down order, the generic prescribing rate increased rapidly to 98.4% and remained there for the entire 2.5-year evaluation period. This is an impressive finding in its own right and will save healthcare services, insurers, and patients millions of dollars. Yet that was only one of the findings.
Figure 1: Generic prescription rates for each medication by month. Each line represents one of the 76 medications evaluated in this study. Pre-intervention period was from January to October 2014. Post-intervention period was from November 2014 to May 2017. Black line ranging from 76.1 to 80.8% in the post-intervention period represents the generic prescription rate for levothyroxine. Source: Olshan et al, 2018
There was one exception. Clinicians opted out of generic drug levothyroxine for the brand name prescription 20– 25% of the time. (See figure above.) Clinicians recognised that generic and brand versions of this medication can differ in formulation and that patients already on the brand formulation, perhaps arriving from another healthcare system, should remain on the branded drug. This is a fascinating finding, reassuring that professionals are still capable of making informed decisions and opting out of a default when it matters.
Summary and conclusion
Default settings are almost certainly one of the most effective tools available to any behavioural science practitioner. However, we could argue that earlier implementations of the concept didn’t fully understand the mechanisms and nuances at play. We were using a powerful tool, despite not having a complete grasp of all of its settings.
Today, thanks to new analysis and continually evolving research, we are gradually moving to a stronger position, one in which we can better tailor default settings to suit the context; perhaps by being more open about their use, or even using them more sparingly – in certain contexts, substituting defaults for other more suitable and equally effective tools, such as more active or better-structured choice, providing feedback or simply increasing cognitive ease. The instruction label for this tool should perhaps now read ‘Handle with care. Follow guidelines such as: 1) Use a default if people will see it as an implicit recommendation or it is perceived as the status quo (research this first to find out). 2) For ethical reasons, make sure it's easy and simply to opt out. 3) Use sparingly to ensure people don't lose trust in the default - active choice may be a better option.’
-  Sunstein, C. (2017) “Nudges that fail” Behavioral Public Policy, Volume 1, Issue 1, pp. 4-25
-  Joel Rosenblatt and Adam Satariano. ‘Google Paid Apple $1 Billion to Keep Search Bar on iPhone’, January 21, 2016 Bloomberg News
-  Madrian and Shea 1999
-  J. J. Choi, D. Laibson, B.C. Madrian, and A. Metrick, "For Better or for Worse: Default Effects and 401(k) Savings Behavior", NBER Working Paper No. 8651, December 2001
-  Sunstein, C. (2017) “Nudges that fail” Behavioral Public Policy, Volume 1, Issue 1, pp. 4-25
-  Sunstein, 2017
-  John Beshears et al., The Limitations of Defaults (Sept. 15, 2010) (unpublished manuscript)
-  Erin Todd Bronchetti et al., When a Default Isn’t Enough: Defaults and Saving Among Low-Income Tax Filers 28-29 (Nat’l Bureau of Econ. Research, Working Paper No. 16887, 2011)
-  Zachary Brown et al., Testing the Effects of Defaults on the Thermostat Settings of OECD
- Employees, 39 Energy Econ. 128 (2013).
-  Krijnen, J. M., Tannenbaum, D., & Fox, C. R. (2017). Choice architecture 2.0: Behavioral policy as an implicit social interaction. Behavioral Science & Policy, 3(2), 1-18.
-  Moshinsky, A., & Bar-Hillel, M. (2010). Loss Aversion and Status Quo Label Bias. Social Cognition, 28(2), 191–204.
-  Sunstein, C. (2016). Do People Like Nudges? Administrative Law Review, Forthcoming
-  Loewenstein, G., Bryce, C., Hagmann, D., & Rajpal, S. (2015). Warning: You Are About to Be Nudged. Behavioral Science & Policy, 1(1), 35-42.
-  Fellowes, M., & Willemin, K. (2013). The Retirement Breach in Defined Contribution Plans. Available at info.hellowallet.com/rs/hellowallet/images/HelloWallet_The%20RetirementBreachInDefinedContributionPlans.pdf
-  Rowson, J. (2011) ‘Transforming behaviour change: Beyond nudge and neuromania’ RSA
About the authors
Crawford Hollingworth is co-Founder of The Behavioural Architects, which he launched in 2011 with co-Founders Sian Davies and Sarah Davies. He was also founder of HeadlightVision in London and New York, a behavioural trends research consultancy. HeadlightVision was acquired by WPP in 2003. He has written and spoken widely on the subject of behavioural economics for various institutions and publications, including the Market Research Society, Marketing Society, Market Leader, Aura, AQR, London Business School and Impact magazine. Crawford is a Fellow of The Marketing Society and Royal Society of Arts.
Liz Barker is Global Head of BE Intelligence & Networks at The Behavioural Architects, advancing the application of behavioural science by bridging the worlds of academia and business. Her background is in Economics, particularly the application of behavioural economics across a wide range of fields, from global business and finance to international development. Liz has a BA and MSc in Economics from Cambridge and Oxford.