Skip to main content

Can we make behavioural changes stick?

Excerpts from Nudgeapalooza

Katy Milkman, codirector of the Behavior Change for Good Initiative, shares some surprising findings on what kinds of nudges can help people make long-term behavioural changes.

This study took four years of my life and basically what I learned in the end is nope, I don't have the magic solution. That's a bummer.
—Katy Milkman, Codirector, Behavior Change for Good Initiative, The Wharton School, The University of Pennsylvania

TANYA OTT: Thanks for joining us today for the Press Room. I’m Tanya Ott and that

KATY MILKMAN: “… nope, I don’t have the magic solution.”

TANYA: … is Katherine Milkman. She goes by Katy, and she’s professor of operations, information and decisions at the Wharton School of the University of Pennsylvania. Katy uses big data—as well as insights from economics and psychology—to study how people can make long-term behaviour changes. Changes like not giving up if things get hard or don’t seem to be going your way.

Katy codirects an initiative called Behavior Change for Good with Angela Duckworth, a Penn psychology professor who studies grit and self-control.

KATY: She studies outcomes in education. I've studied outcomes in health and financial wellness and both of us had become obsessed with, “How do we figure out how to go beyond these nudges to change a single decision—as potent as they can be—to nudges or other interventions (it doesn't have to be a nudge, I'll take anything) that can help transform people's lives in a lasting way?”

TANYA: She talked about their work from the stage of Deloitte’s latest Nudgeapalooza event at Georgetown University.

KATY: So first, a couple [of] statistics that really motivate our work. The first one is the most astounding and that is that, at least, 40% of premature deaths in the US appear to be the result of behaviours that could be changed.1 So not genetic factors, not environmental factors, not bad luck or accidents, but things like diet, exercise, drinking, smoking, failure to get recommended screenings. And this is where the kind of work that we all think about, behaviour change research, nudges, have the opportunity to have a huge impact.

TANYA: There’s another statistic that shocked her. One in four students in the US never earns a high school diploma.2

KATY: On this one, this has not been parceled out into how much of that is a behaviour problem and how much of that are other major barriers that face students, right? So I do not want to suggest that there are not other huge contributors to this problem, like inequality. But there does seem to be some evidence that part of it is behaviour and that we can make an impact on outcomes for kids when we help them figure out how to get even more motivated and focussed.

TANYA: And finally, their third major focus—finances.

KATY: Another one that has many, many contributors, but an astounding statistic, which is that one in three families in the US has no savings at all, zero savings.3 In, certainly, at least some of these families, it's not because they have no paychecks. And so there could be some shift we could make, again, to behaviour that might help with this sort of extraordinary problem.

TANYA: Those are the kinds of problems that motivate Katy Milkman and Angela Duckworth’s work. But let’s break down exactly how they do what they do.

KATY: So imagine that both Liz and Jenna want to start a workout routine. We can even say it's New Year's, because that's when so many of us create resolutions like this. And they decide that the best way to do this is to sign up for a month of sessions with a personal trainer. And let's say, by random luck, they end up with personal trainers who have different philosophies about the best way to kick-start them on a successful and lasting workout habit.

Jenna's trainer is all about routine. So Jenna's trainer first says, “Hey, Jenna, when do you prefer to work out?” And she says, “I'm a 7 a.m. exercise person, that's my ideal time.” And the trainer says, “Great, here's our plan for the next month. We're going to meet three times a week at 7 a.m. to exercise and at the end of this month, you are going to have a lasting habit. You're going to have a stable routine. You're going to exercise consistently. You won't need my help anymore. You're going to be off to the races.” Jenna says, “Great!”

Liz's trainer, on the other hand, has a different philosophy, let's imagine. Liz's trainer also asked Liz when's her ideal time to exercise and let's say Liz is also a 7 a.m. gym rat, but her trainer says, “All right, that's great, sometimes we'll meet at 7 a.m. Good to know that's your ideal time, but the key is we're going to try to be flexible. Sometimes we'll meet at other times. We're going to move things around. We'll exercise three times a week. And by the end of this month of sessions, you're going to build a lasting habit. You're going to know how to roll with the punches, deal with it when life gets in the way of your ideal time and you're going to have built that sustainable exercise habit.”

TANYA: So those are two different kinds of philosophies—one is about building routines; the other stresses flexibility. They both make sense in their own way … but when Katy and Angela surveyed professors at top psychology programmes around the country, 77% of them thought “routine Jenna” would win at changing her behaviour. The others said it either wouldn’t matter which approach was taken or that “flexible Liz” would win.

To test it, they launched a four-week exercise programme at a well-known Fortune 500 company.

KATY: Everyone who signed up for the programme did a couple things. First, as they signed up, they told us the two-hour workout window when they prefer to exercise. Just like Liz and Jenna said 7 a.m., everybody says, this is my ideal workout time. So maybe it's 7 to 9 a.m., maybe it's 5 to 7 p.m.; everybody's different. And then everybody's going to get daily workout reminders 15 minutes before the start of their 2-hour workout window.

TANYA: So that was the setup. They did a randomised controlled trial—that’s the gold standard of evidence-based research design—and assigned a third of the participants to the control group. They made a plan and got the reminders to work out, but nothing else. The second group was meant to work like “flexible Liz”—they were paid for going to the gym during scheduled times, but also other days and times. The final group was “routine Jenna’s”—they got paid for gym visits only during the regularly scheduled time.

What did they see?

KATY: The more we paid people, the more they went to the gym; so that's good. Basic economic theory held up in our dataset, which always sort of makes me breathe a sigh of relief. Whenever you see something in data that you're expecting, it's like, phew, I didn't do everything wrong.

The other thing that we also expected is that when it was easier to collect your cash, because you could go any time to get it, you went more. So the flexible folks are just going to the gym more than the routine folks. We knew that was going to be a feature of our design. And that's actually why we varied the incentive amount, because what we really want to do is zoom in on two types of people who are going to the gym basically the same frequency, but one of whom is going consistently at routine times and the other who's going more flexibly all over the place.

TANYA: Katy says she and her team were really surprised to see that—despite what psychologists predicted—routine was actually not the key to successful habit formation. Flexibility seemed to matter more.

KATY: And let me tell you what we found in the data that helped us explain this. So, let's go back to Liz and Jenna and imagine that both of them were told that 7 a.m. was their ideal time. The first thing that we see when we analyse our data is that “routine Jenna”—the person who is in that routine condition—does actually form a sticky routine around going to the gym after our intervention period, at the same time they were going before, so they were going more at their scheduled time.

So now you are probably thinking, “Okay, that's confusing, because you just said that routine lost.” But this is why routine loses: If they can’t go at their scheduled time, they don't go at all. Whereas the flexible folks, they learn to go to the gym, first best at their scheduled time when they are getting reminders, but second best, some other time and that stuck with them after our intervention period.

TANYA: That got Katy and her research team talking about what really does create lasting behaviour change.

KATY: All the research that we have seen before sort of flew in the face of this and it's just that we need to know a lot more than we do. Our current state of knowledge is really insufficient for figuring out how do we create a lasting change. How can we learn much more, much faster? This study took four years of my life and basically what I learned in the end is nope, I don't have the magic solution. That's a bummer.

TANYA: So they launched the Behavior Change for Good Initiative.

KATY: The premise is that there are lots of people out there—probably most of you in this room, I am certainly one of them—maybe just about everyone out there that would love to change their behaviour on some dimension in these areas that we're interested in: Financial well-being, health, sort of study habits. There are always people looking for solutions, right, that the self-help industry caters to these folks. But how do we reach them and how do we do research with them? One thought was there a lot of large organisations that already serve these people and are collecting data on their decisions. They observe their daily decisions, whether they're trying to intervene and improve outcomes—some are, some aren't—but they're at least able to passively observe what's going on.

TANYA: The team partnered with a national pharmacy chain, a major bank, an organisation that does lots of student testing and others. They assembled an interdisciplinary dream team of scientists with expertise in education, health care, consumer and financial services, psychology and neuroscience. They built a digital platform that would be customised to each scientist’s specifications and would collect all of the data.

And they’ve got some early results. The biggest experiment they’ve run is one focussed on exercise, since 9 percent of premature deaths worldwide are due to physical inactivity.4

KATY: One of the reasons we started by doing something with gyms is not that it is the most pressing health problem, but I think of it [this way]: As the fruit fly is to genetics research, the gym is to habits research. It's this beautiful setting where there's something someone's struggling or trying to do on a pretty regular, repeated basis and it's very measurable and observable on an intent-to-treat basis. You don't have to have people suffer the way we would with eating or often with medication adherence. So it's a really nice, clean outcome variable.

TANYA: TWhen people enrolled in the programme, they were randomly assigned to one of 50 different experiment conditions built by those scientists on the team. The scientists fiddled with the rewards schemes for people who worked out during the initial 28 days. They tested a range of hypothesis from whether it’s best to hit the gym hard with the activity that will burn the most calories or to focus on doing something fun—maybe going to a class with your best friend. Another group of researchers tried changing up the incentives—offer more money if people came back to working out after a period of not showing up at the gym.

Early results showed that all of the specially designed programmes did better than the placebo treatment group—the ones who got US$1.25 in points regardless of whether they showed up the gym.

But Katy says there was some bad news … and it’s what happens after the initial 28 days.

KATY: We were trying to figure out how do you create lasting change. And when we look at the post-intervention period, things look a lot less glorious. So, in fact, the placebo with immediate cash transfer is no longer being outperformed by our best practice treatment.5

TANYA: That’s the glass half empty. The glass half full is that several approaches seems to work. One of them is that have fun laughing with your bestie at the Zumba class rather than pounding out a grueling workout.

Another approach that shows promise is one she built for herself … but she likes to frame from “flexible Liz’s” perspective. Imagine Liz wishes she exercised more, but she lacks the will power to actually do it. Also imagine Liz loves trashy novels, but feels guilty wasting her time reading junk.

KATY: So what I'll propose as a solution to both of Liz's problems and I call it temptation bundling. And the idea is very simple. What if Liz only allowed herself to read trashy novels while exercising at the gym? She'd stop wasting time at home on literary garbage and start craving trips to the gym to find out what happens next in her latest thriller, right? And not only that, she'll actually start to enjoy her novel and her workout more combined, because she won't feel any guilt about reading the novel and time will fly at the gym.

So we tried to teach people about this idea of temptation bundling the way I just taught you. We have a video about it. We have a cartoon about it. We also [gave] out free audiobooks to all of our participants that they could download.

We did that in the control condition, too, importantly. So everybody's getting an audiobook, but in the control condition it's just this weird gift. In the treatment group, it comes with an explanation of a way you might use it as a tool to help yourself crave your workouts.

TANYA: Those who were taught about temptation bundling did significantly better sticking to their exercise routines.

KATY: It's a 10 percent boost in exercise that we're seeing in the treatment group relative to the control during the intervention period.6 And although there's a weird dip in the week right after the intervention ends, maybe everybody is just feeling a little sad, who knows, it actually is sustained. So statistically speaking, there's no difference between the benefits during the four-week intervention and after which is great.

Because remember, we were looking for lasting change. And this, again, is the kind of thing that, if it's an idea we're giving to you and planting, [you would expect] it should wear off. So that's exciting.

TANYA: Exciting not just in the gym, but maybe the classroom too.

KATY: So [we’re] at one large public school district in Florida, for about an hour in January, at the beginning of the third marking period for students, which is right after they've gotten their report cards for the first half of the year. They're all feeling little bit dazed by the reality that school's a little hard. They weren't straight A students probably for the most part and they were looking to improve. It's a good moment to try to intervene.

TANYA: The team of scientists came up with various computer activities and wanted to see which ones would improve the students’ grades the most. In all, 14,000 students participated in five different experiments. Katy explains one intervention was clearly the most successful.

KATY: Most of the time we give kids advice on how to do better in school. We pair them up with mentors. But some really creative work by Lauren Eskreis-Winkler, who's a postdoc with me, has suggested that that may not always be the right thing to do.

There's a psych science paper with lab data showing that when we ask people for advice, that often boosts their confidence in themselves and leads them to introspect about what will work.7 And they actually know the right strategies; they just may not be using them. And this can actually be more effective and more potent than giving advice which goes in one ear and out the other.

TANYA: So they asked kids to spend the session writing down advice to other kids. It wasn’t advice on life. It was advice on how to perform better in school.

KATY: How do you stay motivated? What are the right spots to study in to be effective? How do you avoid distraction from things like cell phones? And they were really smart, really good advice; not surprisingly, these kids know sophisticated strategies. They may just not be using them, or maybe they are using them, [but this] just motivated them to use them even more.

TANYA: And then, at the end, they wrote a short letter to another student just offering general advice on how to improve performance in school. The result of this ten minute intervention—just asking kids for advice on doing well in school?

KATY: We boosted their third marking period grade significantly about 1 percentage point in both math and in their target GPA class.8 By estimates, we're trying to benchmark this.

This is something like getting kids to go to school for between an extra day and an extra eight days in terms of absences. So that's pretty amazing for an eight-minute intervention. But it's also pretty small and therefore kind of believable to me, right, when we analyse the data.

If we'd seen that this moved their grade by 10 percentage points, I would have said, “No way we did something wrong,” like that's a fluke. So this is exciting but I also just want to highlight all of the studies I've talked about are relatively small effects for your free nudge, so we're not getting everybody to go to the gym enough that they're going to prevent a heart attack. We're not getting kids to improve their performance in school enough that they're all going to go to college now or they're all going to graduate from high school. But we're making progress and we're generating new insight and hopefully can be useful to make a lot of progress.

TANYA: That was Katy Milkman of Wharton School at the University of Pennsylvania. She codirects the Behavior Change for Good initiative with Penn psychologist Angela Duckworth. Katy also has a podcast on behaviour changes called Choiceology. You can find it where you found this podcast!

This is the Press Room and I’m Tanya Ott. We’re on Twitter at @deloitteinsight and I’m on Twitter at @tanyaott1. You can learn more about Deloitte’s annual Nudgeapolooza event and behaviour economics at our website, We’ll have another podcast in about two weeks, so make sure you check back or, better yet, subscribe to the podcast. It’s free and you can have the latest episode delivered to your device of choice for listening wherever it is you listening.

This podcast is provided by Deloitte and is intended to provide general information only. This podcast is not intended to constitute advice or services of any kind. For additional information about Deloitte, go to

Deloitte Consulting


No matter how complex your business questions, we have the capabilities and experience to deliver the answers you need to move forward. As the world’s largest consulting firm, we can help you take decisive action and achieve sustainable results.

Learn more

  1. Centers for Disease Control and Prevention (CDC) Newsroom, “Up to 40 percent of annual deaths from each of five leading US causes are preventable,” news release, accessed November 6, 2019. 

    View in Article
  2. Martha Ross and Nicole Bateman, “Millions of young adults have entered the workforce with no more than a high school diploma,” Brookings, January 31, 2018.

    View in Article
  3. Amanda Dixon, “A growing percentage of Americans have no emergency savings whatsoever,” Bankrate, July 1, 2019 . 

    View in Article
  4. I-Min Lee et al., “Effect of physical inactivity on major non-communicable diseases worldwide: An analysis of burden of disease and life expectancy,” Lancet 380, no. 9838 (2012): pp.219–29. 

    View in Article
  5. Stephen J. Dubner, “How goes the behavior-change revolution?,” Freakanomics Radio, June 19, 2019.

    View in Article
  6. Katherine L. Milkman, Julia A. Minson, Kevin G. M. Volpp, “Holding the hunger games hostage at the gym: An evaluation of temptation bundling,” Management Science 60, no. 2 (2013).

    View in Article
  7. Lauren Eskreis-Winkler et al., “A large-scale field experiment shows giving advice improves academic outcomes for the advisor,” PNAS 116, no. 30 (2019).

    View in Article
  8. Ibid. 

    View in Article

Did you find this useful?

Thanks for your feedback

If you would like to help improve further, please complete a 3-minute survey