The Behavioral Insights Team 5 Years On: A Conversation with Owain Service

This article was originally published on The Psych Report before it became part of the Behavioral Scientist in 2017.

“Nudge, nudge, wink wink” read one headline when the UK’s Behavioral Insights Team or ‘Nudge Unit’ announced the results of their first year’s work in 2011. The Monty Python reference, far from the only time it was used, revealed that many viewed the newly established team tasked with bringing behavioral science into public policy as something far less than a serious approach to government.

Today, however, nearly 200 randomized control trials later and with their findings permeating virtually all areas of public policy—including job assistance, organ donation, and tax collection—the creation of the BIT and the wedding of behavioral science and public policy might seem like forgone conclusions. But hindsight is twenty-twenty, and it’s hard to overestimate the changes to policy creation and implementation the BIT’s work has catalyzed. No better evidenced than by the fact that since its launch in 2010, the United States, Canada, Australia and many others have established their own behavioral science units using the BIT as a model.

We recently spoke with BIT Managing Director Owain Service, who’s been with the team since its start, to better understand the BIT’s path-breaking course over the past five years and find out where they’re headed next.

Evan Nesterak: How has the behavioral insights approach impacted the UK government on the whole?

Owain Service: We have now run close to 200 randomized controlled trials in pretty much every single area of government policy. Each of those has shown better ways of doing things or, in some areas, things that don’t work. [There are] two things I would point to that, personally, I am most proud of. The first is that I think we can say we have changed the way in which policy is made in Whitehall. People think about drawing on ideas from the behavioral sciences in a way that five years ago almost nobody did. Secondly, people now think about using randomized controlled trials as one of the policy tools that can be used to find out whether or not something works. Again, that was just not considered to be part of a policymaker’s toolbox five years ago. So rather than pointing to the successes of the interventions, I think I’m most proud of the fact that we’ve started to change the mindsets of policymakers in the UK government.

EN: Is there a particular example that you think illustrates how an initially skeptical government agency adopted this mindset?

OS: We’ve done a lot of work in job centers in the UK. Job centers are institutions that we have that help people back into work and administer their benefits. They’re run centrally by the UK government. I think that it would be fair to say that three or four years ago we had conversations around what you could do differently in relation to job centers, but it was always a bit tricky to get that program off the ground.

I think we can say we have changed the way in which policy is made in Whitehall. People think about drawing on ideas from the behavioral sciences in a way that five years ago almost nobody did.

Then, about three years ago we started a very very small piece of work in a single job center. We redesigned the process that an individual goes through when they go into the job center for the first time. We ran a very small pilot. Did some A/B testing. That showed a lot of promise. At that point I think we started to build belief around the ideas from within the job center community. We then turned that into a very large trial across a region of England and that replicated the findings of the original piece of work. Those interventions are now in every single job center across the country and they touched the lives of 9 million people across the country.

EN: The BIT is in the process of hiring a Ventures Start-up Manager who will help create and scale behaviorally informed products. Most of the BIT’s work has been in consulting, so this seems to be a new direction. Can you describe the vision for this new pursuit?

OS: We obviously used to be a team within government and then last February we signed ourselves out of government and set ourselves up as a social purpose company. One of the reasons we did that was because we found when we were inside government quite often the answer to a problem we were looking at was not ‘let’s change a policy.’ It was, there is some kind of product or service that will help people to make better decisions. But when you’re inside government it’s difficult to do that. What you normally have to do is you have to encourage somebody else to do so on behalf of the government. But now, one of the things we can do as a social purpose company, albeit one that is partly owned by the UK government, is to take the initiative ourselves. So we have a couple of what we’re calling ventures in the pipeline and we’re hiring somebody to support the development of those ideas.

EN: Can you let us in on some of the first ventures you’re planning?

OS: I can’t tell you the detail yet, but I can give you a bit of a heads up that one of our first ventures is around debiasing recruitment practices.

EN: Recruitment for?

OS: For jobs. [For] an organization that wants to do its recruitment in a smarter, more efficient way, but most importantly in a way that removes some of the inherent biases from the recruitment processes that most organizations put in place.

EN: One of your most well known interventions involved sending a letter to people who hadn’t paid their taxes on time, which explained that most people do pay their taxes on time. What was the impact of this social norms letter on tax collection?

OS: That was one of our original trials. In some ways the reason it became quite famous and we still talk about it a lot is because it was one of our first trials that enabled us to be able to say here is a very simple change that anybody could make to their administrative processes that has a real tangible effect.


The original trial, which was the social norms in tax letters trial, was on a group of late payers of self-assessment tax debt. The most powerful letter gets you about a five percentage point increase in tax repayment. We’re going from about 32-33 percent to about 38 percent in that particular trial. The most powerful message says the vast majority of people in your local area pay their tax on time and people with a tax debt like yours have paid by now.

The interesting thing is that some of these nuances are quite important. Telling people that they are part of a group of people that is more locally defined than just the country as a whole seems to be more powerful. And this message around people with a tax debt like yours have paid by now, seems to resonate even more strongly.

That was part of a group of trials we ran with HMRC (Her Majesty’s Revenue and Customs), which they estimate over the course of 12 months helped bring forward 200 million pounds of revenue. Now it’s important to know that we’re talking about bringing forward revenue rather than additional revenue. Most of the time the tax authority will get to most of the people eventually. What you’re really doing in this area, in a tax system where ultimately almost everybody pays eventually, is you’re making the processes much more efficient. You’re making them better for citizens. They don’t get chased up. They don’t get taken to court. And you’re saving costs to the administration as well. You’re not having to spend money on people constantly ringing people up, writing extra letters, and taking people to court.

EN: How are you thinking about this trial three and four years on from when it originally started? Have you continued to send a similar letter?

OS: HMRC has continued that program. They have found that the same approach to the same group of people has a similar effect the following year. What we haven’t done is a longitudinal study of the individuals then who receive the first letter and then sent them another one six months later which is similar in tone and message. People often ask with nudges in general do the effects wear off? The answer is it depends on what the actual intervention is. In some areas, for example automatically enrolling people into pension plans, the decision is a one off decision. You either do that or you don’t. In other areas, you have more habitualized decision making to deal with, and the sense would be that constantly bombarding somebody with the same approach will wear off.

EN: Where does the use of behavioral insights go next? Much of the discussion over the last four or five years has seemed to focus on same pros and cons—the low cost of the interventions, the use norms and defaults, and the need to nudge for good. Where’s the new territory where the work goes next? 


OS: I think a lot of it is about taking some of the ideas to scale. In the work that I mentioned in relation to job centers, that’s a good example of how you can take something relatively small scale and then scale it up so that the system as a whole is starting to use some of these practices. I can’t say that we’ve done that in every single area that we’ve been working in. Yet, we’re still a relatively small team and I think in lots of areas we’ve had amazing results that haven’t yet transformed whole services. That’s one area I’d like to us focus on.

The second is that alongside the policy side of things we hope that we will start to see the development of products and services as well policies that really help people to make better decisions for themselves.

Then the third is that I mentioned how the use of randomized controlled trials is starting to become more prevalent as part of the policy making process, but again we still feel that we are only scratching the surface in this particular area. I’d like to think that in 10, 15 years time the use of large-scale randomized evaluations will become the norm in the way that it is for clinical trials and it is becoming to an extent around A/B testing for internet firms. It was not so long ago that in the clinical world people didn’t say has this been tested and trialed before it’s been rolled out. Now, it would feel to anybody like that would be a crazy thing not to do before introducing a medicine or a new clinical process. I hope that in 15 years time we’re saying similar things about the policy making process.