Applying Behavioural Science to Increase Engagement and Outcomes

Robbie Tilleard at UXDX EMEA. Video: https://youtu.be/kT9QXeVCOgU

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Introduction

[00:00:00] Hi everyone. I'm really glad to share with you some lessons on applying behavioral science to technology today. So what are we going to talk about? We're going to step through four core areas. First, introduce you to behavioral science: what is it? Second, give you the tools to apply behavioral science to technology. Third, suggest where behavioral science might fit into your product life cycle, the way that you develop your products today. And finally, share a practical example of behavioral science at work.

[00:00:30] Just really quickly, a little bit about me. Whilst my accent isn't as strong these days, I grew up in a regional town in Australia. When I was growing up, the largest employer left that town, taking a few hundred jobs with it. Good professional jobs. This always made me interested in how to create good jobs and how to help communities transition. But five years ago I realized a lot of this is psychology. It's about identity. It's about asking a steel worker to become a data scientist or a health worker. And this led me to a startup focused on good jobs and understanding users, where I think we could take a scaled approach in that area.

Why behavioral science, and the investment game

[00:01:05] So why do we even bother with behavioral science anyway? Haven't the economists worked all this out already, and we humans behave rationally at scale? Classical and modern economics has suggested that humans are broadly rational. We stack rank our preferences and then we're consistent in our decisions. It suggests we respond in expected and predictable ways to incentives, rules and information. Maybe that makes sense.

[00:01:35] But let's play a little game. I want you to look at this chart. You see there is an amount of time on the bottom and a percentage gain from an investment that you made from time zero on the Y axis. So you are now the investor trying to decide between which of these investments you would like to go into.

[00:01:56] We have five possible ones to choose. Power Trade, the dotted line. Fortitude, the kind of straight line in there. Alpha. Tobacco Trade. And the S&P 500, that thick bold line. Take a moment now to have a think: which one of these would you most like to invest in? Just your immediate gut feel. Write it down in front of you or say it out loud.

[00:02:26] If you chose Fortitude Investments, you are with the majority. 68% of people choose that fund on average in this exercise. However, the challenge is one of those funds is really well known to you. Perhaps many of you remember Bernie Madoff? He had a well-known Ponzi scheme that collapsed in the global financial crisis. It is also Fortitude Investments. Those of you who chose this are not alone, that's what the majority chose. But why do you think that is?

[00:02:55] It's really about certainty. Fortitude is the fund that provides a sense of a steady, increasing return over time without the scary peaks and troughs of some of the others.

[00:03:07] But interestingly, there is a simple nudge or technique that changes this dramatically. Any guesses? The answer is: which fund was the most suspicious? If you ask people that before they make the decision, about 17 percentage points of people change their approach, and so the amount of people choosing Fortitude declines to 51% of the time.

[00:03:33] Of course there are two lessons in this: providing certainty and the power of nudges. This is really why behavioral science matters. It's why we see Uber or ticketing or Deliveroo so successful with this approach. Arguably it's not because of their slick user interfaces, but it's actually the fact that they're sharing with you where exactly you are and what will happen next. You have certainty over when that taxi is going to arrive, where you are in the queue, or where you are in terms of your orders.

What behavioral science is

[00:04:04] So with that introduction, hopefully you can see why behavioral science matters for product. Let's make sure we're really clear on what it means, especially, to be fair, how it differs slightly from psychology in general.

[00:04:17] Behavioral science is the study of how and why humans make decisions and take action. Taking action is really important, as it's primarily interested in actual behavior and not stated behavior. For instance, we all say we want to go to the gym more, but a behavioral scientist is more interested in, did you go to the gym?

[00:04:37] Second, it's looking for systematic or predictive effects. This means for instance, if we say taking time to go for a walk in your day makes you more productive, this works on average for everyone. These are not one-offs, but they apply in general, on average, to everyone, all of your users, and so on.

Fast and slow thinking

[00:04:57] And broadly, these findings from behavioral science can be simplified into what might be called a pop academic model, but it's a useful mental model for how to think. And that is that we have two sides of our brain, a fast side and a slow one.

[00:05:13] Fast is intuitive, automatic, reacts to your environment. If I say 3 + 3, immediately six will likely come into your head. It's what happens when you catch a ball. You're not doing the complex geometric equations to know when the ball's going to land in your hand. Instead, it's an intuitive reaction.

[00:05:30] And then there's slow: rational, considerate, thoughtful. If you need to do large multiplication, or you're thinking about how you might win that game, not just catch the ball, this is more where that part of the brain will kick in. This is famously popularized by Danny Kahneman, one of the Nobel Prize winners in economics.

[00:05:51] And today what we're going to really be focusing on is that fast part of the brain. But both are really useful. And a key finding is, however, that we use our fast brain much more than we would like to admit.

[00:06:09] The slow brain responds to economic levers, rules, information and incentives that I mentioned at the start, but also of course the larger shifts that might be occurring in values, beliefs or attitudes in yourself or society. But the fast brain, well, that's looking for the easiest way out, and so it's often very responsive to the environment around it. And that's going to be much more the focus of today's presentation. But I want to make sure you don't leave thinking the only offering of behavioral science and wider psychology to technology are nudges. It does go much deeper than that.

Nudges and EAST

[00:06:47] So, thinking about nudges, just to be clear on what the definition is here. Broadly, they're an aspect of the choice architecture that alters people's behavior in a predictable way without forbidding any options. And this was popularized by Richard Thaler, who also won the Nobel Prize in economics, and an academic called Cass Sunstein. And so there are two areas that are really important here: choice architecture, so what's that environment that your user is in, and then how do we help people make the best decisions for themselves?

[00:07:20] And it turns out there are really simple ways to change that choice environment to impact that fast decision-making part of the brain. Richard Thaler has a very famous saying: if you want to change behaviors, make it easy. And then my old boss, David Halpern, CEO of the UK's Nudge Unit and the Behavioural Insights Team, summarized decades of psychological research into an easy mnemonic called EAST. If you want to change behavior, make it easy, attractive, social and timely.

[00:07:49] Let's break this down. Easy. People will do the easiest thing possible. Take desire lines in parks. People won't follow urban design necessarily, the concrete path. They'll take the shortest possible route. And so that's why you see these areas of heavy walking that don't necessarily apply to the concrete. And so think of choosing defaults for users as an example here.

[00:08:16] Then there's attractive. Make the desired action as salient as possible. Think of why Google won the search wars. Not only was it the best, it was also the simplest and most attractive to navigate. You literally wanted to do the search. And they famously spent months getting the right color blue on their buttons that made people more likely to search. I'm not sure if people remember, but like AltaVista or Ask Jeeves, it wasn't that same focus on making the page attractive.

[00:08:47] Then there's social. Have you ever wandered down the street, seen everyone looking up, and then you do as well? Well, we're social animals, so having a social comparison, understanding what other people are doing, makes you more likely to actually do that behavior.

[00:09:02] And finally, timely. Here we see searches for P60 or employment summaries essentially in the UK. It peaks every year in April with the tax year. So if you're selling tax products, guess which is the best time to nudge users. And so thinking about what is that best time, when people might be most likely to be changing the behavior, is what it means to be timely. So here we go: easy, attractive, social and timely. An easy way to apply behavioral science in your work.

Nudges all around us

[00:09:35] It's worth mentioning of course that none of this is particularly new. We have nudges all around us. Take driving, for instance. It's the ultimate guided choice environment. What nudges do people see here if you take a look for a second?

[00:09:52] So on the left, obviously you'll see the barriers, making sure that cars aren't going off the road. You have the line in the middle, which is making sure people are really clear about where they're going. And you even have an arrow there, making sure people are aware that there's a corner coming up. And then on the right you have something probably a bit more innovative, kind of a way of creating a really attractive, salient example that this is a pedestrian crossing. So people using these nudges in different ways.

[00:10:20] And whilst none of this is new, its translation into a science has, in the last 30 to 40 years, revolutionized the way we think about human behavior and developing new habits. So let's take an example where you might want to discourage littering. Here we see incentives being applied: there's a fine for littering. We see rules being applied, so littering is literally illegal. Then we might see information: what is the impact to the environment or to yourself of that littering? And then finally we see nudges at the end. These are two classic examples of making a behavior attractive. Who would want to litter when you could play hoops? And perhaps the footsteps are an attractive and salient guide to the rubbish bin that makes you more likely to take your litter there. Nudges aren't going to change everything, but they're a useful new tool that product managers and we in technology can use to actually see how we can influence user behavior.

What makes it a science

[00:11:23] And of course this is all great in theory, sounds awesome on podcasts like Freakonomics, or great at dinner parties. But what makes this science anyway? Well, the science component is driven by measurement and experimentation. Behavioral science runs randomized controlled experiments to understand if its findings truly impact people in that systematic and predictable way that I mentioned earlier.

[00:11:48] This is just like what we've seen with vaccine trials in COVID-19, for instance. Two groups are made at random. One group gets a placebo, the other gets the real thing, and then we can see if the real thing works by comparing the two groups. We then test this in different ways and with large populations to make sure it's a robust finding over time.

Sludge

[00:12:10] So that's nudge, but an important reminder for all technologists is not to fall into the trap of sludge. This is where, instead of helping people make better decisions, expanding their choice environment, these might often restrict the choice environment and make it harder for the user to achieve the outcome they want.

[00:12:32] So here are two famous examples. One is from The New York Times, where it's just not that easy to cancel your subscription. On the way in you can sign up online, put your credit card, it'll take you almost seconds. On the way out, the only real way to cancel your subscription is actually to talk to someone rather than just pressing a button.

[00:12:52] The other example is taken from The Economist, where they're using a decoy to encourage the purchase of a higher price. They have just the digital subscription, which is one price, then they have the print, and the print and web together, making it seem like it's just an obvious deal in order to basically double the cost that you're paying. This is probably not quite as clear an example of sludge as The New York Times, but it is certainly an example of perhaps not giving your user as much freedom in their choices as they might actually like. So I want to be clear that many tools in behavioral science can be used for nefarious ends, and it's our responsibility as technologists to ensure we nudge for good and we don't sludge.

[00:13:42] So, just thinking about that now. That was the summary of behavioral science and one of its tactical tools called nudging. But think back to your last project. Did you nudge? Did you help customers make the best decisions for themselves? How did you do that? Did you send some email notifications? Did you make it a really salient choice in order to do something? Then perhaps you're a nudger. Or did you sludge? Did you not take a user-centric approach, but instead you forbid options, or you impacted the user in ways they might not have understood? Have a think.

Where it fits in the product life cycle

[00:14:21] So now, having introduced you to behavioral science, let's get practical. When would you actually use behavioral science in the product life cycle? At Multiverse, where I work, we break our life cycle into five core phases. This is inspired, or maybe even copied, from my team from my time at the Behavioural Insights Team. So we call that target, explore, solution, take off and scale.

[00:14:45] In target, you want to choose your target objective. What are you actually aiming for? What's the metric you want to shift? In explore, understand the problem space and the barriers to that objective. Why aren't people doing the thing that you'd like them to do? And then that will throw up a bunch of solution areas.

[00:15:04] In the solution phase, therefore, you confirm your proposed approach. You do the trade-offs and prioritization to work out what you're going to build. Then you launch your MVP to users in take off, before learning those lessons and scaling to users. You likely have a similar approach. Your company is driven by the double diamond or some kind of scientific method area, but mapping this is useful if we think about where things are going to actually apply to behavioral science.

[00:15:32] So in the explore phase, you want to really understand the capability, opportunity and motivation of your users. Why are people able to do the thing? Can they? Are they motivated to do it? These are the kinds of questions you want to ask.

[00:15:49] In the solution phase, this is when you really want to think about that EAST. Is the solution easy, attractive, social and timely? And if it's not, why not? How can you make it easier for the user? And then finally, of course, you need to be continuously updating that mental model you have of the capabilities, opportunities and motivations of your users.

A practical example at Multiverse

[00:16:11] So we've learned about behavioral science and where it fits into the product life cycle. Let's take a practical example from my work. How might we encourage apprentices to retake knowledge checks?

[00:16:23] Today I work at Multiverse. We're an edtech company on a mission to create a diverse group of future leaders by providing an alternative to university and corporate training through tech apprenticeships. We provide apprenticeships to young adults and those looking to reskill. We focus on really in-demand skills, so not plumbing, or, plumbing is very much in demand, but more like software engineering, digital marketing and data analytics, that tech side of the business. And apprentices pay nothing for their training and earn a salary while they learn.

[00:16:53] So that's a little bit of what we do at Multiverse to get you situated. Let's focus on the problem that we were having a look at. This is broadly our learning ecosystem. Our apprentices engage in their learning over a 12-to-15-month course. This has a lot of detail, but in the middle here you can see what we would call the core learning areas that an apprentice is doing. They do some kind of content, they do workshops, assignments, and they do knowledge checks, which is what you see in green, and that's what we're really going to focus on in terms of the problem that we were solving.

[00:17:28] So we wanted our apprentices to revise and learn more to make sure they understood the content. Knowledge checks are these short quizzes that apprentices complete at the end of each learning module. So you'll have a bunch of learning modules as part of your learning experience. What we wanted to do is increase the number of apprentices retaking those knowledge checks until they scored 100%. We wanted them to, when they got 60% on the quiz, we wanted them to go back in, understand what they got wrong, and ultimately redo it until they got all the questions correct, so that obviously they were showing learning over time.

[00:18:05] The issue is we had a very basic interface previously. Think about essentially a Google Form with just the question, and you get to select the answers. So it's a very basic approach. And so we asked ourselves, how might we encourage apprentices to retake knowledge checks?

What we built, and the result

[00:18:21] So here is what we did. We wanted to make finishing the quiz attractive to people. We tested whether a split screen, seen here, the purple and then the gray, would work better than a typical approach where questions are often read vertically. So you read the question and then the answers are beneath it. And we actually ran this via an A/B test, and our users found this a much more attractive approach, so much more likely to finish the quiz and actually to retake it.

[00:18:47] We also gave our apprentices easy and timely review. That's the right answer of why they got that question right or wrong, which you can see here in the image. And therefore it was easy then to find, and it was also provided just at that timely moment they wanted to understand more information about the topic, particularly if they'd actually gotten it wrong.

[00:19:10] In our summary sections, once you finish the quiz, we then use coloring to make it attractive to retake knowledge checks and make it easy to revise topics you got wrong, by highlighting them to users in their knowledge check summary.

[00:19:23] So these are broadly the nudges we used. Really easy and attractive to make sure you kept on taking it. Really easy and timely places to review to make sure that you're learning on the way. But I'm sure after this presentation you could think of many more. For instance, you could do some social nudges around, are other people taking this at the same time, or what have other people gotten, in order to encourage that behavior. But importantly, the result of all this was a more than 25% increase in retaking behavior, which was our target, compared to the existing platform. It's a huge increase, helping our apprentices stay motivated and ultimately learn by themselves.

Summary and further reading

[00:20:03] So in summary, what have we learned today? First, behavioral science is a systematic study of how and why humans make decisions and take action. An entry point is nudges: if you want to change behavior, make it easy, attractive, social and timely. Make this part of your product development process. The highest leverage moments will likely be in your equivalent of the explore and solution phases. And finally, it works. We apply it every day at Multiverse, most of the big tech companies apply it, and there is good academic evidence, although it's always somewhat contested here and there, of examples that work from the literature.

[00:20:42] And of course there is a lot to read and understand more on. So some really easy entry points if you're interested in this topic: Michael Lewis has a bunch of books in this space, and Laszlo Bock[?], who used to be my boss, the Behavioural Insights Team, whose work I've used quite a lot in this presentation, and some great podcasts here as well, which I'd recommend. Of course you can always dive deeper. You can read Daniel Kahneman's Thinking, Fast and Slow, and actually read it, don't just put it on your bookshelf. And then there are some other great books as well that get you into that space.

[00:21:15] So finally, that's it. And if you like this presentation, if you're interested in this space, if you're interested in the work we do at Multiverse, of course, we're hiring. multiverse.io/careers. Please check out our pages and see what you think. So, thanks a lot.