At a government office in the UK, a small tweak in the phrasing of a letter changed everything. Officials were struggling to get people to pay their taxes on time until someone added a simple sentence: “Most people in your town have already paid.” Suddenly, compliance shot up, without needing to introduce any new laws or hefty fines. Just a simple shift in wording was enough to have a huge impact. Behind that quiet revolution was one of the world’s first “nudge units.”

These so-called nudge teams—often known as behavioral insights teams, particularly in the case of the original nudge unit, now referred to as the Behavioral Insights Team (BIT)—are small groups of experts who apply findings from psychology, economics, sociology, and behavioral science to public policy. Instead of relying on top-down mandates, they focus on subtle changes in how choices are presented (nudges) to help people make decisions that are in their own best interest. Want to increase vaccination rates, reduce energy use, or improve job training enrollment? A nudge team might suggest a default setting, a timely reminder, or even a well-placed message at just the right moment.

The term “nudge” comes from the 2008 book Nudge by Richard Thaler and Cass Sunstein, which popularized the idea that behavioral science could be used to design better systems without restricting freedom of choice. Over the last decade, governments around the world have embraced this approach. Nudge units now operate across sectors, testing low-cost interventions with big potential payoffs. But their influence goes beyond clever tweaks: they’ve challenged how policymakers think about human behavior and how governments can help people lead better lives. 

Nudges: Subtle changes in the way choices are presented that steer people toward better decisions without restricting their options or significantly altering economic incentives. They preserve freedom of choice while influencing behavior in predictable ways.

Choice Architecture: The deliberate crafting of decision-making environments. By subtly shaping how options are presented, choice architecture influences individual decision-making, often without explicit awareness.

Homo Economicus: Latin for “economic man,” this term describes a hypothetical figure who represents the concept of unconditional rationality. This caricature is what traditional economic models were based on—although today, we recognize that no actual person acts in a vacuum without cultural, emotional, physical, or mental influences.

Libertarian Paternalism: A concept in behavioral economics and public policy that aims to influence individuals' choices in a way that will make them better off, as judged by themselves, while still preserving their freedom to choose.

System 1 and System 2 Thinking:  The two distinct modes of cognitive processing introduced by Daniel Kahneman in his book Thinking, Fast and Slow. System 1 is fast, automatic, and intuitive, operating with little to no effort, allowing us to make quick decisions and judgments based on patterns and experiences. In contrast, System 2 is slow, deliberate, and conscious, requiring intentional effort, used for complex problem-solving and analytical tasks where more thought and consideration are necessary.

Default Option: An option that is selected automatically unless an individual chooses otherwise.

Framing Effect: The cognitive bias that suggests our decisions are influenced by the way information is presented. Equivalent information can be more or less attractive depending on what features are highlighted. 

Behavioral Economics: A method of economic analysis that applies psychological insights to human behavior to explain economic decision-making.

The origins of nudge theory are rooted in behavioral economics and psychology, particularly the work of Daniel Kahneman, Amos Tversky, Richard Thaler, and Cass Sunstein. Thaler and Sunstein’s 2008 book Nudge: Improving Decisions About Health, Wealth, and Happiness popularized the term “nudge,” which describes how we can subtly alter the environment or context in which people make decisions with the aim of influencing their behavior. As research into nudges expanded, their use became widespread, designed to guide decisions in predictable ways by leveraging cognitive biases without restricting freedom of choice or changing incentives.

Before the publication of Thaler and Sunstein’s book on nudges, they introduced the concept of libertarian paternalism, gently guiding behavior without restricting freedom. Although their paper titled “Libertarian Paternalism is Not an Oxymoron” laid a solid foundation for justifying the use of nudges in public policy, the paper was met with much criticism.¹ Some academics argued that one-size-fits-all nudges ignore the individual, social, and cultural differences between people, and that designing systems that attempt to emphasize the “right” choices forces one social and cultural perspective onto everyone.² Critics also fear that policymakers, who are just as prone to cognitive biases as anyone else, may inadvertently (or purposely) design nudges for the public with the intention of shaping things in the best interest of the designer.³ Once we allow for some level of libertarian paternalism through corporate or governmental nudge teams, it’s a slippery slope to allowing even more paternalistic or authoritarian forms of intervention.⁴ 

Despite these criticisms, the first nudge team was established within the UK Cabinet Office in 2010 as the Behavioural Insights Team (BIT).⁵ Their mission was to apply behavioral science to public policy problems, including, of course, through the use of nudges. The early team included only seven members, who tackled practical issues like increasing tax compliance and supporting people into employment. As a team of data-driven and research-minded individuals, their approach usually involved the use of large-scale randomized controlled trials (RCTs) to test interventions before scaling.⁵ Their results were often published in peer-reviewed journals, helping to legitimize the field.

Focused on an empirical approach, the team developed a number of frameworks for designing and testing their interventions. For example, EAST (Easy, Attractive, Social, Timely) is a simple, actionable framework for designing behavioral interventions, and TESTS (Target, Explore, Solution, Trial, Scale) is a structured approach to building and evaluating behavioral insights. As the latter name suggests, the framework is helpful for testing out different behavioral intervention approaches. BIT has also designed online experimental platforms for testing, like Predictiv, which allows teams to test interventions before real-world implementation.⁵

BIT was the world’s first government institution dedicated to behavioral science in policy, and its success sparked rapid global interest. In 2014, BIT became independent, separating itself from the UK government and expanding internationally. Many other countries saw this as a good thing: in their first non-UK project, they tripled tax compliance in Guatemala. BIT has also participated in major philanthropic partnerships, such as a $42M initiative to improve city governance in the US. As they’ve increased focus on sectors like health, education, labor, and sustainability, the rest of the world has continued to take note. In fact, over 200 behavioral insights teams now exist worldwide, in countries like the US, Australia, Singapore, Canada, and beyond.⁵ What’s more, the concept has spread to the private sector, NGOs, and international development agencies, who have also begun adopting behavioral approaches. As BIT continues to gain success and notoriety, the team attracts more leading academics and experts, cementing its reputation as a hub of applied behavioral science.

Today, the original nudge team at BIT has evolved into a global research and innovation consultancy with more than 220 staff members, over 1,800 projects, and offices in seven countries. Their work includes systems change, digital design, social innovation, and complex policy evaluation—well beyond basic nudges. The teams have been known to collaborate with academic experts in the field, like Richard Thaler, as well as renowned institutions like Harvard, LSE, Princeton, and more. In 2021, BIT was acquired by Nesta, a UK-based innovation charity focused on social impact.⁵ Ideally, their shared goals and vision for testable and scalable solutions, along with Nesta’s resources and infrastructure, will continue to be mutually beneficial. 

Herbert Simon

An American economist and cognitive psychologist known for his concept of bounded rationality, which explains how cognitive limitations affect decision-making. His work highlights why people might need guidance to make better choices.

Daniel Kahneman

An Israeli-American psychologist and Nobel laureate known for his work on the psychology of decision-making, particularly his development of prospect theory. His research on cognitive biases and heuristics explains how people often make irrational choices and how small interventions can improve decision-making.

Amos Tversky

An Israeli cognitive psychologist, best known for his work on decision-making and judgment under uncertainty, who co-developed prospect theory with Daniel Kahneman. His research on cognitive biases and heuristics laid the groundwork for understanding how people make irrational choices.

Richard Thaler

An American economist and Nobel laureate known for his contributions to behavioral economics, particularly through his work on nudge theory. He co-authored the book Nudge, which explores how small interventions can help people make better decisions by addressing cognitive biases, and his work has shaped behavioral science in policy and economics.

Cass Sunstein

An American legal scholar and co-author of the book Nudge alongside Richard Thaler. He’s known for his work on behavioral economics and public policy, particularly how governments and organizations can use nudges to improve decision-making and promote better outcomes without restricting individual freedom.

Even if we focus only on the original nudge unit, the Behavioral Insights Team (BIT), the impacts of their work are wide-ranging. Their main areas of focus include AI and technology, the economy, education, the environment, government and society, health, and transportation.⁵ 

Education 

Education is one of the strongest and most reliable predictors of various life outcomes, for everything from earning potential and future professional success to things like life expectancy and political affiliation.⁶ A stable and supportive education is incredibly influential in supporting long-term success. Unfortunately, education, like many other publicly funded services, is often underfunded. That’s why it’s important that education interventions be both high-impact and low-cost. Traditional top-down and authoritarian mandates usually don’t work in the classroom—if they did, we could just require every student to pay attention in class or get an A on every assignment.⁷ Instead, nudge teams looking to improve education have had to find creative ways to weave behavioral insights into the experience of students, teachers, school administrators, and parents in an attempt to improve education outcomes. 

What did this look like in practice? In one project, the team tackled absenteeism in Uruguay. To address this kind of complex issue, the team first had to identify the main behavioral barriers, which included parents’ lack of knowledge and optimism bias, which led parents to underestimate their student’s absences, as well as a saliency and survivorship bias, where schools mostly focused on students with many consecutive absences but sometimes overlooked other cases where students could be helped. BIT had previously tested absentee interventions in the US and the UK, so they replicated some of the behavioral interventions they’d tested before, including a personalized letter sent home to parents, an informational brochure, a school calendar, and a chatbot designed to report absences. They compared different combinations of these treatment groups to a control condition (where no intervention occurred), and ultimately found that those in the full treatment condition experienced a 6% increase in attendance, or an additional 16,920 days of school attended in total.⁷ 

In older students, they’ve tested a social action program that develops employability skills, behavioral interventions to encourage university applications, and awareness campaigns to increase the use of free early childcare opportunities in the UK. Overall, BIT has tackled over 60 projects in education, testing strategies in more than 500 schools with 180,000 students; that’s a big impact!⁷

Government and society

The original nudge teams often worked for and as a part of the government. Although they don’t exclusively function this way anymore, anyone who’s worked in the space likely feels there are ways to make some bureaucratic processes more efficient. Nudge teams have worked to increase economic mobility with innovative strategies like connecting cities across a nation, leveraging the power of social pressure and social norms to improve their chances of sticking to their goals. The benefit of this type of intervention is that it supports cities in a way that traditional advertising or information campaigns can’t fully address. Not only does the established “social network” between cities give a sense of accountability to leaders, but it also connects local governments with insights into what may or may not work for their own city, as they can better learn from the triumphs and trials of others.⁸ 

This type of work involves BIT’s collaboration with Results for America (among other philanthropies) to support the design and evaluation of behavioral science-informed programs aimed at improving citizen uptake of governmental programs.⁹ Using the EAST framework, the teams improved recruitment and retention efforts, expanded the availability of affordable housing units in New Jersey, and even increased preschool attendance in Dayton, Ohio—a personal favorite of mine, as I attended a wonderful preschool in Dayton. By encouraging governments to make these publicly-minded programs easy to apply for, attractive to residents, leveraging social connections, and timely for beneficiaries, the EAST framework has helped numerous local and national governments improve the quality, quantity, and use of their resources.⁹

Health

Nudge teams can take a variety of approaches to improving health outcomes. Often, they either focus on improving public health outcomes or specific healthcare systems. On the public health side of things, the team at BIT has tackled issues like the global spread of disease, rising obesity rates, and vaccine misinformation. In one study, they analyzed how to keep public drinking water safe in the United States. 

While many people may assume that tap water is high quality everywhere in the country, the number of lead pipes and the outdated infrastructure in many cities often compromise the safety of the water.¹⁰ This is another issue close to my heart; for most of my time in middle and high school, the water fountains were connected to lead pipes or unsafe service lines. Years later, the city discovered that there were high levels of lead, PFAs, and other dangerous chemicals in the water that the children in my district were growing up with. Across the United States, contaminated water disproportionately impacts communities of color, leading to long-term negative health effects.¹⁰ 

Meanwhile, in the city of Chicago, citizens have access to free lead testing kits. Unfortunately, most of these kits were going unused. In 2022, BIT partnered with Bloomberg Philanthropies’ What Works Cities Certification program to support the Chicago Department of Water Management (DWM), not only to understand why the current testing program wasn’t taking off, but also how to make it better.¹⁰ 

To begin, the team developed a behavioral map outlining all the steps needed to successfully complete the lead testing kit, along with the barriers that might impede progress. Through qualitative research like user testing and interviews, the team discovered that residents found the materials in the kit intimidating: they were complicated and confusing. What’s more, they had a hard time remembering (or even finding the time) to let the water stagnate for six full hours before the testing. Not only did they have to remind themselves not to run the water for those six hours, but they also had to remind everyone else in the house not to use the water.¹⁰ 

Using this understanding of the barriers to testing, the team came up with a number of behavioral interventions to make the process easier. First, they simplified the kit’s instructions, cutting out any unnecessary details and using color coding to make the key information accessible. They also added “DON’T USE” stickers to the kit, which could be placed on their faucets and toilets. These visible and physical barriers made it easier to remember. Ultimately, the city’s rate of return increased by over 20 percentage points, and for those who received SMS reminders, the rate of return increased an additional 3.8 percentage points.¹⁰ Sometimes, a little simplification is all it takes to get things flowing. 

Although BIT and other nudge teams have seen real results, not everyone is convinced that nudging lives up to the hype. As the field has matured, critics have raised sharp questions about its evidence base, ethical foundations, and long-term effectiveness. From overstated impact claims and shaky replication rates to concerns about manipulation and oversimplified policy design, nudge teams now face a reckoning: can their subtle interventions really hold up in a complex, messy world?

Overstating effect sizes and evidence base

Despite their power, some behavioral economics interventions, including nudges, have consistently overpromised their impact and underdelivered in real-world settings. While studies on these interventions might promise an average impact of around 8.7%, the actual impact observed in the real world is significantly lower, averaging about 1.4%.¹¹ Seems like false advertising, right? 

This discrepancy is partly attributed to the nature of academic research, where the stakes are much lower for researchers compared to, say, a surgeon, if they get things wrong. Critics of nudges (or at least those who question the validity of nudge research) have a number of concerns and theories about why this discrepancy exists. First, it’s possible that the studies’ interventions aren’t always done with sufficient care in experimental design, recruiting, or analysis.¹² There's a concern that researchers aren’t always showing the full data sets collected, potentially hiding unfavorable results.

There are also low replication rates: the behavioral sciences, including areas related to nudging, have a notoriously low replication rate. Fewer than half, possibly closer to 36%, of published studies are able to be replicated, and this weakens the overall evidence base.¹² Academic incentives favor publishing "surprising findings" and statistically significant results over replication or reporting null results, further contributing to a skewed literature.

In fact, a major meta-analysis of nudges found evidence of publication bias, which is when studies with positive or statistically significant results are more likely to be published than those with null or negative findings.¹³ This, in turn, leads to an overrepresentation of positive results, distorting the scientific knowledge base and potentially misleading professionals, policymakers, and the public. A response to this meta-analysis specifically claimed that after appropriately correcting for the detected publication bias using advanced statistical methods, "no evidence for nudging after adjusting for publication bias" remains for an overall effect, suggesting the actual efficacy of nudges may be negligible or null.¹² For those who are statistically inclined, the meta-analysis found an average effect size of Cohen's d = 0.45, which is typically considered a moderate effect but is surprisingly large in the context of such minor interventions, causing the claim that small nudges could have such huge results to come under scrutiny.¹³ 

Critics point out that the meta-analysis likely included studies subject to selection bias, where studies reporting statistically significant results would overestimate effect sizes. It also included studies later retracted for fraud, further contaminating the results. There’s also a question about whether it makes sense to aggregate vastly different types of interventions (like organ donation defaults vs. persuasive messages) under the single label of "nudge," as the diversity of these nudges means that averaging everything may not tell us anything meaningful.^12,13 

Don’t lose hope yet, though. If we can distinguish between the academic literature, which may suffer from publication bias and other issues, and real-world applications by practitioner groups, then there may be some good news. Studies examining the full universe of trials conducted by specific behavioral science organizations (including unpublished results) have shown clear, positive effects, averaging an 8.1% improvement on outcomes across millions of people.¹⁴ This suggests that while the published academic record might be skewed, real-world interventions designed by practitioners can still be effective.

Ethics of intervention

Critics of nudge teams often concurrently dispute the ethics of libertarian paternalism. One of the concept’s central tenets is that it preserves freedom of choice because individuals can easily opt out of the suggested arrangements. However, these critics argue that if nudges are effective precisely because they exploit cognitive biases, inertia, and bounded rationality, then the right to opt out may not translate into true freedom of choice for the very people the policies target.^2,3,4 Instead, default rules and framing effects mean people don't (or can’t) easily reverse the suggested options in practice. For example, the dramatic difference in organ donation rates between countries with opt-in vs. opt-out systems is cited as evidence of nudging’s power for good, but it also demonstrates that defaults significantly constrain actual choices, reducing effective freedom and decisional autonomy.³

Some critics also worry that the means used by nudge units to achieve their goals aren’t transparent. This lack of transparency can lead to an accountability deficit, making it harder to monitor and evaluate the actions of planners. Nudge units must often assume that the "planner" (whether in government or a private institution) will always act in the best interests of the individual being nudged. Critics argue this ignores the premise that individuals within government (politicians and bureaucrats) have their own interests, agendas, biases, and bounded rationality, just like the general population. Thus, this raises concerns about the potential for nudge policies to be influenced by others’ selfish interests or malicious intent. If voter choices are easily manipulated by factors like framing or heuristics, then even democratic mechanisms like voting can’t ensure that planners' interests align with citizens' true concerns.^2,3,4

There’s an additional concern that adopting even mild forms of "libertarian paternalism" could lead down a slippery slope to more intrusive interventions. This slope isn’t just about moving from soft to hard paternalism (e.g., from opt-out defaults to outright bans), but also within the realm of libertarian paternalism itself, moving towards progressively less visible and more manipulative forms of nudging (e.g., unflattering mirrors or subliminal advertising). The way proponents tend to frame the debate is as "how much" paternalism rather than "whether" contributes to this risk.⁴

Critics also question how the planner knows what the individual "really" wants or what choices would genuinely promote their welfare. Behavioral economics shows that people have inconsistent preferences; think about when you intentionally set your alarm extra early the night before, and your morning self decides to sleep in anyway. Deciding which preference or state reflects the individual's "true" welfare involves subjective judgments and potentially imposes the planner's or some other socially approved preferences under the guise of objective science. This requires difficult welfare calculations that may not be easily agreed upon.^2,3

By constantly emphasizing people's cognitive deficiencies and using techniques that work around these limitations rather than helping people overcome them (through training or better information processing tools), nudge units might implicitly treat individuals as perpetually irrational. This could become a self-fulfilling prophecy, discouraging individuals from developing better decision-making skills and reinforcing reliance on System 1 thinking.^2,3,4

Lastly, critics argue that some nudge policies can inadvertently lead to the redistribution of resources from “more rational” individuals to “less rational” ones. Policies that encourage more people to take advantage of a limited pool of funds or resources (like employer contributions to retirement accounts or lower food prices in a cafeteria) can mean less is available per person, or costs must be externalized. These costs are often borne by rational individuals or the public through taxes, but this redistribution, even if unintended, raises ethical concerns for those who oppose state-mandated transfers.²

Complex systems perspective

It’s still not clear if an approach based primarily on individual decision-making, as with traditional nudging, increases equity or inequalities. Who are the leaders on the nudge teams or behavioral insights teams, and how do they decide what’s best for individuals? 

The increasingly heavy focus on nudging in the early 2010s has also been criticized because many other behavior change intervention types and approaches exist that might be more appropriate for addressing certain policy problems. While nudging is a well-known and often cost-effective way to use behavior intervention, some critics have suggested that when we use behavioral interventions in policy, it shouldn’t be equated with just nudges or nudge teams. Instead, we should move toward more extensive approaches.

Part of this perspective stems from a concern that nudge-inspired public policies are insufficient for solving complex problems. Although nudges can be successful, many problems are so complex that these types of interventions can’t address the underlying causes of the issues at hand. As the global arena becomes more interconnected and complex, any type of nudge team work needs to adapt beyond simple nudges if they want to match the challenges posed by non-linear changes and unintended consequences.¹⁵ In other words, simple interventions (like nudges) often can’t solve complex problems where the outcomes are hard to predict or have unforeseen positive and negative effects based on the interconnectedness of the systems in which they exist.

There are also concerns about simplistic views of behavioral insights, which make it harder to assess or use more complex approaches. At times, popularized views can lead to focusing more on individual behaviors than the system, or seeing behavioral science teams as a synonym for nudges or consisting only of basic concepts. This oversimplification can hinder the integration of complex systems perspectives, which is an important part of shifting responsibility from the individual to the broader systems that are often responsible for the conditions people are stuck in in the first place.¹⁵ For example, much effort has been put into nudging people to eat more fruits and vegetables and reduce their consumption of highly processed or fast food. However, this overlooks the much more relevant role of food prices, limited access to fresh produce, and major food companies and government subsidies favoring highly processed ingredients. 

The context of public policy, including pressure for straightforward and quick results, can even hinder the adoption of systems approaches, leading to a continued focus on simpler interventions like communication-based nudges. When nudge teams meet, the advisory sessions can be brief, which could force experts to start with simple advice. While simple interventions have historically been seen as a way to justify the value of using behavioral insights, especially in the beginning, this focus can prevent tackling more complex issues effectively.¹⁵

Penn Medicine Nudge Unit

Although we’ve mainly focused on the original nudge team, BIT, there are many other nudge teams around the globe, specializing in a variety of fields. In 2016, the Penn Medicine Nudge Unit opened as the world's first behavioral design team embedded within a health system. Their self-described mission is to design, implement, evaluate, and scale evidence-based, behaviorally informed nudges that steer decisions toward higher-value care, better patient outcomes, enhanced public health, and greater health equity.¹⁶

In their first five years, the unit worked on more than 100 projects, which included more than 25 randomized trials, and these efforts resulted in over 75 publications in leading medical journals. For the Penn Medicine Nudge Unit, projects typically move through five phases, very similar to other nudge teams: contextual inquiry and data analysis, design, implementation, evaluation, and scale and dissemination. Because they’re focused on health outcomes, their work spans from targeting workflow improvements and decision making in healthcare providers to implementing nudges for sustained change in patients, to implementing interventions aimed at reducing the public health burden caused by epidemics like distracted driving, opioid addiction, and gun violence.¹⁶

In one study, the team sought a way to support patients with opioid use disorder. Often, these patients are seen in emergency departments but are then medically cleared, discharged, and left to navigate a complex treatment system after discharge. The Penn Medicine group gathered insights from a diverse group of clinicians and designed a screening protocol to identify patients early in their emergency room stay. Using a participatory design approach, they created a nurse-driven protocol for opioid use disorder screening as part of triage and coupled this with automated prompts to both nurses and physicians so that they could assess and treat the disorder head-on. 

Because the team was focused on understanding the true behavioral barriers involved (and worked to collect data on the processes involved), they learned that physicians wanted nurses to drive more aspects of this type of care, and nurses were eager to do this. This insight led to a nurse-driven triage protocol that initiated OUD care well before a patient encountered their treating clinician. They designed the prompts with this in mind, as well as recognized the importance of peer recovery support early in the visit to maximize opportunities for discharge planning. In this study, as with others at Penn Medicine, relatively simple interventions have had a big impact on real-world health outcomes.¹⁷ 

Summer nudging: Can you believe it? 

When summer hits, most high school students are swept away with dreams of going to the beach, ice cream on a hot day, maybe even thinking about all the extra cash they’ll earn in tips at their summer job. Few students have the same level of excitement about their FAFSA applications for financial aid. Unfortunately, just like their ice cream, many students fall victim to "summer melt," a phenomenon where students, particularly from low-income backgrounds, who intended to enroll in college after high school graduation, fail to matriculate.¹⁸ This attrition occurs because students often face complexities related to financial aid, managing a high volume of college correspondence, and overcoming the psychological burden of new, uncertain situations. Many low-income students also lose access to high school guidance counselors during the summer, and their families may lack college experience, exacerbating these challenges.¹⁸

To mitigate summer melt, two large-scale randomized trials were designed and implemented by nudge teams in the summer of 2012, testing two promising approaches: personalized text messages and peer mentor outreach. The studies were conducted across various sites, including the Dallas Independent School District (Dallas ISD), uAspire in Massachusetts, and Mastery Charter Schools in Philadelphia, Pennsylvania.¹⁸ The target population was college-intending high school graduates, typically identified by FAFSA completion, prior engagement with advisors, or self-reported college plans on exit surveys. The samples predominantly included students of color and those eligible for free or reduced-price lunch.¹⁸ The nudge team’s goal was to see if there was a way to nudge these at-risk students into the college attendance they’d planned for themselves. 

In the first intervention, organizers sent 8-10 personalized text messages to students and their parents, at approximately five-day intervals mid-summer. These messages reminded recipients of essential college tasks like logging into their college web portals, registering for orientation or placement tests, completing housing forms, and navigating financial aid processes. Many of the messages included direct links to make it easier for students to immediately click and complete the task, and the messages even offered the option to request follow-up assistance from a counselor by simply replying to the text.¹⁸ 

The second intervention involved college students, who were typically alumni of the high schools in the study, proactively reaching out to high school graduates. These peer mentors provided encouragement, shared first-hand college experiences, assessed students' readiness for matriculation, and connected them to professional counseling for more complex issues like financial aid. Students across the sites were randomly assigned to either a text message group, a peer mentor group, or a control group.¹⁸

The results of these summer nudging campaigns demonstrated notable successes, particularly given their low cost: the text message intervention had a positive impact on college enrollment in several intervention sites, with enrollment rates 4 to 7 percentage points higher among students who received messages compared to those who didn’t. These effects were primarily concentrated among students in communities with limited educational attainment, those who qualified for free or reduced-price lunch, and those with less defined college plans. In Dallas, students in the text message group were 4.9 percentage points more likely to enroll at a two-year college. In two of the Massachusetts sites, the text intervention led to a 7.1 percentage point increase in overall college enrollment. Again, the text message intervention was remarkably cost-effective, at approximately $7 per participant.¹⁸ 

The peer mentor intervention also proved successful, increasing four-year college enrollment by 4.5 percentage points overall, with the largest effects seen for males and students with less defined college plans. For students with undefined college plans in Massachusetts, the intervention increased enrollment by 16.0 percentage points, and for those with fewer than four advising meetings, it increased enrollment by 10.8 percentage points. While this intervention cost approximately $80 per participant, both strategies were relatively cost-effective and supported an increase in college entry among traditionally underrepresented populations.¹⁸

However, the efficacy of nudges, particularly when scaled, is not always guaranteed, and this raises important questions about their universal trustworthiness. Many of the promising results for nudges come from relatively small-scale studies, often involving hundreds or thousands of students in local settings.¹⁹ A crucial question in behavioral science and public policy is whether these interventions maintain their effectiveness when expanded to a state or nationwide ("global") scale. Recent research has made conclusions a bit more complicated and provided evidence suggesting that nudges that work locally may be hard to scale effectively.¹⁹ 

For instance, a study on FAFSA completion campaigns, representing the largest FAFSA nudge campaigns to date and collectively reaching over 800,000 students nationwide, found no impacts on financial aid receipt or college enrollment overall, nor for any student subgroups.¹⁹ This null finding persisted across various experimental variations, including different behavioral framings, delivery channels (mail, email, text), offers of one-on-one advising, and timing of messages.¹⁹ The large sample sizes in this study allowed researchers to rule out even very small effects.

This implies an obvious discrepancy between small-scale successes and large-scale failures and highlights several reasons why nudging may not always be trustworthy when scaled up. One primary hypothesis is the importance of the relationship between the student and the source of the nudge.¹⁹ Most prior successful interventions involved a local partner with closer connections to and knowledge of the individual students. Students may respond more favorably to messages from people whom they think are specifically invested in them and their communities, like in the peer network. 

In contrast, global scale-ups often involve more generic and less personalized messaging. Even when a significant portion of the target audience engages with the messages (e.g., over 40% in the large-scale FAFSA study), this engagement doesn't necessarily translate into the desired behavioral change or positive outcomes. Other factors contributing to null results at scale include the possibility that current cohorts of students may already possess better information, diminishing the marginal impact of nudge campaigns.¹⁹ 

Lastly, challenges in replicating pilot programs at scale, like selection bias (who is being chosen as the recipient of these nudges?) and context dependence, can also undermine effectiveness.¹⁹ Even though one-on-one advising has shown potential, staffing sufficient advisors for hundreds of thousands of students would obviously be challenging for any organization. These findings from this second study underscore that simply replicating successful local nudge designs on a larger scale may not yield similar positive results, and nudge teams must instead be thoughtful about applying nudges before attempting to scale up. 

Insuring Behavior Change

Nudge teams are an ideal tool to help firms provide a better customer experience and tackle fraud, which means they have huge potential for influence in the insurance industry in particular. TDL leveraged behavioral science tools to pilot a nudge unit in this sector. Read here to learn more about how and why the team tried this out and the results of the study. 

Nudge Theory 

If you’re looking to understand more about how and why nudges work (and where they fall short), this piece can help unpack the core concepts that nudge teams are built upon. Although a focus on nudging individuals can be effective, it’s important to acknowledge some of the limitations of an individual-focused approach to behavior change.

1. Sunstein, C. R., & Thaler, R. H. (2003). Libertarian Paternalism Is Not an Oxymoron. The University of Chicago Law Review, 70(4), 1159–1202. https://doi.org/10.2307/1600573 
2. Mitchell, G. (2005). Libertarian paternalism is an oxymoron (FSU College of Law, Public Law Research Paper No. 136; Law and Economics Paper No. 05-02). Florida State University College of Law. https://ssrn.com/abstract=615562 
3. Rebonato, R. (2013). A critical assessment of libertarian paternalism. SSRN. https://doi.org/10.2139/ssrn.2346212 
4. Whitman, G. (2010, April 5). The rise of the new paternalism. Cato Unbound. https://www.cato-unbound.org/2010/04/05/glen-whitman/rise-new-paternalism 
5. Behavioural Insights Team. (n.d.). Our history. https://www.bi.team/about-us/our-history/
6. Wolf, Z. B. (2024, October 14). Why education level has become the best predictor for how someone will vote. CNN. https://www.cnn.com/2024/10/14/politics/the-biggest-predictor-of-how-someone-will-vote
7. Ramos, R., Cabrera, M. A., Sánchez, I., González, I., Suarez, G., & Rosich, L. (2024, May 8). Let’s get to school! Behavioural Insights Team. https://www.bi.team/blogs/lets-get-to-school/
8. Hallsworth, M., & Tublin, L. (2022, June 13). Increasing economic mobility by bringing together nine US cities. Behavioural Insights Team. https://www.bi.team/blogs/increasing-economic-mobility-by-bringing-together-nine-us-cities/ 
9. Bennett, E., Cardon, E., & Toth, C. (2022, May 10). How can governments increase program uptake? Behavioural Insights Team. https://www.bi.team/blogs/how-can-governments-increase-program-uptake/
10. Cardon, E., McNease, M., Mirpuri, S., & Tublin, L. (2023, September 7). Keeping household water quality safe. Behavioural Insights Team. https://www.bi.team/blogs/keeping-household-water-quality-safe/ 
11. DellaVigna, S., & Linos, E. (2022). RCTs to scale: Comprehensive evidence from two nudge units. Econometrica, 90(1), 81–116. https://doi.org/10.3982/ECTA18709
12. Maier, M., Bartoš, F., Stanley, T. D., Shanks, D. R., Harris, A. J. L., & Wagenmakers, E.-J. (2022). No evidence for nudging after adjusting for publication bias. Proceedings of the National Academy of Sciences, 119(31), e2200300119. https://doi.org/10.1073/pnas.2200300119 
13. Mertens, S., Herberz, M., Hahnel, U. J. J., & Brosch, T. (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119(1), e2107346118. https://doi.org/10.1073/pnas.2107346118 
14. Hallsworth, M. (2022, August 2). Making sense of the “Do Nudges Work?” debate. Behavioral Scientist.
15. Puukko, S., Heino, M. T. J., Kostamo, K., Saurio, K., Sniehotta, F. F., & Hankonen, N. (2024). How do behavioral public policy experts see the role of complex systems perspectives? An expert interview study. Translational Behavioral Medicine, 14(7), 417–425. https://doi.org/10.1093/tbm/ibae024 
16. Center for Health Care Transformation and Innovation. (n.d.). Nudge Unit. University of Pennsylvania. https://chti.upenn.edu/nudge-unit 
17. Lowenstein, M., McFadden, R., Abdel-Rahman, D., Perrone, J., Meisel, Z. F., O’Donnell, N., Wood, C., Solomon, G., Beidas, R., & Delgado, M. K. (2022). Redesign of opioid use disorder screening and treatment in the ED. NEJM Catalyst Innovations in Care Delivery, 3(1). https://doi.org/10.1056/CAT.21.0297 
18. Castleman, B. L., & Page, L. C. (2013). Summer nudging: Can personalized text messages and peer mentor outreach increase college going among low-income high school graduates? Center for Education Policy Research, Harvard University. https://cepr.harvard.edu/files/cepr/files/summer_nudging.pdf
19. Bird, K. A., Castleman, B. L., Denning, J. T., Goodman, J., Lamberton, C., & Rosinger, K. O. (2019). Nudging at scale: Experimental evidence from FAFSA completion campaigns (NBER Working Paper No. 26158). National Bureau of Economic Research. https://doi.org/10.3386/w26158