Designing for Behavior Change: Applying Psychology and Behavioral Economics (2013)

Part VI. Putting It into Practice

Chapter 16. Conclusion

Four Lessons

We can design products that help people take action. We’ve seen it done—in exercise, in energy usage, in personal finance, and in many other domains. It requires four things:

1. An understanding of how the mind works

2. Clarity about the intended outcomes of product, and about its users

3. A detailed plan for how the product will help the user act

4. The willingness to accept that the product won’t be perfect, and needs to be tested and refined over time

Each of these maps onto one of the major processes of designing for behavior change, and the corresponding parts of the book: understand, discover, design, and refine. Throughout this book, I’ve tried to provide the background information and the practical tools needed to make this happen in your own business, government agency, or NGO.

Here’s a quick recap.

Understand: How the Mind Makes Decisions

When we’re awake, we’re always doing something—whether it’s reading, running, or resting. We tend to think that the things we do are intentional: I run at 7 p.m. each day because I really want to; I read the newspaper every morning because I want be informed about world news. We believe that our conscious minds think carefully about what to do and decide on a course of action. But the reality is a little more complex:

§ Most of the time our behavior is on autopilot, driven by the automatic, reactive part of our brain. Habits are a key type of autopilot behaviors and work in predictable ways, with cues, routines, and, sometimes, explicit rewards. But habits aren’t the only type. We intuitively react to situations based on our past experiences and the mindset we’re in—the particular facet of self has been activated to respond to the current situation.

§ Even when we’re not on autopilot, our minds are constantly looking to save work. We have terribly limited attention, limited short-term memory, and limited and easily depleted willpower. That’s not to say we’re stupid or weak—instead, our minds find creative ways to work around our biological limits. We substitute simple problems for hard ones. We look to our peers to guide us when we’re not sure what to do. We look for solutions that are “good enough” rather than trying to find the absolutely optimal answers.

§ The simple, obvious stuff matters. Designers have known this for a long time. We like products that are easy and beautiful. We’re more comfortable with familiar things and familiar actions than new and strange ones. We like good experiences, and we really don’t want to fail. We try to do urgent things before things that aren’t urgent.

Given this quirky decision-making process, why do we end up taking one action rather than another? We’re each unique, and there’s no simple formula that says person X will always do Y. But we can extract lessons from the research literature to identify the preconditions that need to be in place for a person to take a specific action:

§ Cue. The thought of taking the action needs to somehow cross the person’s mind. This can occur because of an external cue (the phone rings) or an internal one (the stomach growls).

§ Reaction. In the blink of an eye, the automatic, intuitive self reacts to the idea. It checks whether the action is relevant and interesting at all. It generates an emotional response. It activates and starts to consider other possible actions that the person might take.

§ Evaluation. If the action isn’t intuitively discarded, it can bubble up to conscious awareness for a cost-benefit analysis: how hard will the action be to take, what’s the action’s value for the user, etc. The mind would evaluate other alternatives as well and determine the most promising action.

§ Ability. If the action is worthwhile (benefit > cost, and more than the alternatives), the person must actually be able to act. Ability has many levels: the person must know logistically what to actually do, have the resources to do it, and must not be dissuaded by the assumption of failure. If the user faces obstacles, often they can be resolved, but that process takes time and leaves the person open to distraction.

§ Timing. The person needs to have a reason to act now, rather than doing something else that is more urgent. That reason can range from external emergencies (the car is about to crash) to internal disequilibrium (hunger, thirst).

If all of these are in place, then a person can execute the action. Together, they form the C-r-e-a-t-e Action Funnel. For any given action, you can think about people leaking away at each step, either intentionally deciding not to take the action, or becoming distracted by other things. Designing for behavior change entails helping people pass through all five stages, from inaction to action.

There are three main strategies that products can take to help people through the Create Action Funnel:

1. They can use the brute force method, supporting the user’s conscious decision to act, each and every time the user should take the action.

2. They can build habits, which rely on the automatic, intuitive part of the mind. With habitual behavior, the evaluation and timing parts are usually skipped over, making it more likely that the person will take the action.

3. They can “cheat”—by shifting the burden of work from the user to the product, so that after the user gives consent, the user doesn’t need to act. That inverts the Create Action Funnel and makes inaction work to the user’s benefit.

Discover: How to Clarify Goals and Understand Your Users

Designing for behavior change builds upon a clear understanding of the target outcome, action, and actor:

§ The outcome is what will be different in the real world when the product is successful;

§ The actor is who will cause that change (the actor is usually the user of your product); and

§ The action is how the actor will do it—the behavior the actor will undertake.

You elicit and clarify each of these three things in order to find problems early on in the development process: problems where the team disagrees on the actual purpose of the product, or who the target audience is, or how the product will be judged for success or failure.

Ideally, the outcome should be a tangible, real-world impact. There’s a natural tendency to think about the impact on users’ emotions, awareness, or knowledge—things that are inside users’ heads. We want to move beyond that, into the practical impact on people’s actual behavior and the world around them. Often, it’s the application of a user’s new-found knowledge that companies really care about, and not the knowledge itself. For example, if a product taught people about the dangers of smoking but did absolutely nothing to change smoking behavior, that would be a failure. It’s the real-world impact that matters.

Next, the company should clarify exactly how the actors (usually the users of the product) are going to make that outcome happen. If the outcome is improved health, then are users supposed to exercise more or eat better? If the company isn’t sure what the best action is for their users, this process can help:

1. Generate a set of ideas. The team can brainstorm ideas on its own, or use questions like these to spark thought: “What does someone do right before the outcome occurs?,” “What actions do users already take that are related to the outcome?,” “What’s stopping users from making the outcome happen?” Direct observation of users to learn how they are already making the outcome happen (e.g., how they are already losing weight) is invaluable.

2. Learn the about the target users. What’s their existing motivation to act? What blocks them? Do they have prior experience with similar actions or with similar products?

3. Generate user personas. For each potential action, segment the user population based on who is going to respond differently to an appeal to act. For example, people who exercise regularly are going to respond differently than couch potatoes to an appeal to run a half marathon.

4. Evaluate the potential actions. Rate each action according to four criteria: impact (how effectively it will drive the outcome, like health), motivation (how much users currently want to take the action), ease (how familiar the action is to users), and cost (how costly it will be for the company to build a product that supports the action). Then select the best action given the particular resource constraints of the company and the realities of the users.

If the company already has a target action in mind, that’s great—all it needs to do is specify what the action is, learn about its target users, and generate the personas.

Design: How to Design for Action

Designing a product to support behavior change can seem like a daunting task. There are countless choices that you need to make, and hundreds of different cognitive mechanisms at play. You can think, in turn, about the three different parts of the decision-making context that come together to shape how you influence behavior, and the Create Action Funnel tells you what needs to happen here.

Action

Start by breaking the target action into discrete steps that the user will need to complete. For example, in order to eat healthy dinners at home, the user will need to decide what to eat, know how to prepare it, plan out the ingredients, buy those ingredients, budget enough time to cook, and cook. Given this list of small, discrete steps:

1. Tailor each step to the user’s level of expertise and interest.

2. Simplify each step to reduce the amount of work required of the user. Ideally this means removing steps altogether.

3. Make each step easier to complete by making it understandable to the user, ensuring the user can see it as feasible, and providing feedback of progress during long-duration actions.

Environment

Shape the user’s decision-making environment in two senses—by designing the product itself to support action and by helping the user change the broader environment, outside of the product. In each case:

1. Increase motivation by reminding users of their existing reasons to act and, potentially, by providing additional incentives.

2. Cue action by directly asking users to act and encouraging them to set up a cue in their environment.

3. Generate a feedback loop by providing ongoing status updates about how users are doing so they can adjust their behavior accordingly.

4. Remove competition for users’ attention and behavior by removing distractions within the application, coopting and building upon existing behavior, or competing head-on for users’ interest.

5. Remove obstacles by ensuring that the product is accessible and meets the particular constraints (e.g., handicaps, reading ability, platform access) of the users.

User Preparation

Now the behavior has been tailored to the users, and the environment has been set up to support that behavior, look at how to best prepare the user to take action. There are three tactics:

1. Narrate. Change how users see themselves. Help users see the action as a natural extension of who they are by having them recall similar things they’ve done in the past and by linking the action to their self-identity.

2. Associate. Change how users see the action. Make it clear how this action builds upon, and uses knowledge from, other things they’ve done before. Make a strong connection between the action and other positive things they enjoy—like smelling the fresh cut grass after mowing.

3. Educate. Change how users see the world. Show them logistically what’s required to complete the action. Make sure they have all the information they need to act. However, it can be ineffective to use education as a tool for motivation (people probably already know the action is important).

Use this three-step process, focusing on the behavior, then the environment, then the person, to develop a behavioral plan—a narrative that describes what the user might do to progress from not taking the action to taking it. That plan can be articulated in a customer experience map, journey map, textual narrative, or a simple list of bullet points with annotations.

The behavioral plan provides the core functional requirements for the product’s behavior-changing aspects. It’s a conceptual design for the product. The product team converts that plan into the raw material for the team to consume and act upon. In agile development, that is a set of user stories; for sequential development, it’s a functional spec.

The interface designer and other members of the product team then develop wireframes or mockups of the product. If the product team is stuck and not sure how to structure the application flow itself, they can use one or more design patterns for behavior change—such as trackers, gamification, or how-to tutorials.

While the wireframes are being developed, you can deploy specific tactics from the behavioral social science literature to increase the probability action—ranging from social proof, to reminders, to competition, to loss aversion. Table 10-1 provides a list of two dozen, high-impact tactics, organized by the part of the Create Action Funnel that they affect.

Finally, the team should test the concepts with potential users, using either wireframes, or ideally, prototypes. If everything checks out, it’s time to build the full product.

Refine: How to Iteratively Improve the Product

Whenever we design a product that interacts with the vagaries of human behavior, we’re going to get some things wrong. The goal of the design process is to make the initial product less wrong.^[160^] But there will still be significant room for improvement. Improving the product proceeds like this:

1. Define clear metrics for the outcome and target action.

2. Measure the impact of the product on the outcome and action. If the outcome isn’t directly measurable within the product, develop a data bridge—a statistical relationship between something that can be measured in the app and the outcome the company cares about. The data bridge can come from existing research or a small pilot study conducted by the team or third-party academics.

a. If possible, run an A/B test or another type of controlled experiment to measure impact. That’s the gold standard, and there are numerous tools to help companies run them with online products. Statistical knowledge is not required (though it can help).

b. If experiments aren’t feasible, the team will need to measure overall impact, then formally or informally control for other things that might influence the users’ behavior. Statistical techniques such as matching and the analysis of panel datasets can mitigate these issues but require statistical expertise.

3. Find obstacles to impact by directly observing your users and checking the data for places in the app where users get stuck. A causal map, showing the various pathways by which users can decide to take the target action, can help the team figure out what users are doing and how to best interpret the data.

4. Identify potential solutions, either through user feedback or a set of small, quick tests to diagnose why the problems are occurring. The team can use the Create Action Funnel to target and interpret those quick tests.

5. Prioritize proposed changes to the product designed to improve behavioral impact alongside other proposed changes meant to support business goals, engineering goals, etc.

6. Test each major change. Regardless of whether everyone on the team thinks it’s a good idea, check the impact of that change on behavior. Human behavior is just too complex to accurately forecast, and a culture of testing can help check everyone’s assumptions.

Themes

Stepping back from these detailed lessons, there are a few basic themes that I’d hoped to express here. These themes provide some additional color and guidance on how to design for behavior change.

We’ve learned a great deal about how the mind makes decisions, but our knowledge is still limited

The behavioral social science literature provides the foundation for this book, and there are numerous field experiments that illustrate how our decision making works. The research on biases and heuristics is well established, but research on how products can help change behavior is still in its infancy. This book pulls together the best of what’s known currently, but our knowledge is certainly not complete.

Understand your users

Changing behavior is a highly personal affair. Our products should meet users where they are in their daily lives and help them take action—that starts with understanding their needs and interests, constraints, prior experiences, and levels of expertise.

There are no magic wands for behavior change

There’s no sure-fire way to make people take an action. Even if we fully understood the decision-making process, we each have unique personal histories and environments that shape our behavior. While we can’t dictate behavior, we can facilitate it. If someone wants to act, well-designed products can help him do it.

Be intentional

If the goal is to help people change behavior, then do it. Don’t hide behind wishy-washy recommendations or “raising awareness” among users; that leads us to design and build products that are internally conflicted and ineffective. Be up front and proud of products that help users take action.

A behaviorally effective product must first be a good product

The product must be well designed, pleasant to use, and solve a user need. Any considerations about changing behavior need to build on top of that foundation—otherwise, people will simply choose not to use the product.

Avoid user work

Look for technical solutions to avoid user work; it’s much easier to engineer a solution than it is to change behavior. We can and should celebrate user triumphs—but those triumphs should occur where hard work was unavoidable, and not where the product was poorly designed.

We don’t need to get fancy

Look for the Minimum Viable Action that a user can take to reach their goal, and remember simple, obvious lessons about the psychology of design: we like beauty, simplicity, familiarity, and following our peers.

We should assume we’re (partially) wrong

No matter what we design and build, we’ll get some things wrong. A dose of humility in face of the vast complexity of human behavior is a good thing. It spurs us to test the real, quantifiable impact of our products, and test every major change we make to them.

Looking Ahead

Some pretty cool things are happening right now: Rapidly increasing power and resolution on smartphones and on tablets. Wearable computing. Smart thermostats and other gadgets. New research in behavioral economics and psychology on how our minds work.

Within this flurry of activity and innovation, we’re seeing the emergence of a new generation of products that apply behavioral social science to the practical problems of people’s everyday lives: helping people take action and change behavior. A few of these products already touch millions of lives and have the potential to profoundly change society for the better: by helping us get in shape; by helping us save for the future; and by helping us become active in our communities and in politics on the issues we care about. Other products shape our lives in more mundane, but still powerful, ways: by helping us manage our email better or organize our days.

No one knows what will come next. There’s a wide-open field ahead of us. I look forward to what each of us discovers, designs, and builds.