3 strategies to change user behavior
There’s no (ethical) way to avoid having the user consciously choose whether or not to act, but products can change the nature of that choice.
There’s no (ethical) way to avoid having the user consciously choose whether or not to act, but products can change the nature of that choice.
How can a product help its users pass all the way through the Action Funnel and actually take action? There are three big strategies that a company can choose from, to change behavior and help users take action. Two of them come straight from the research literature and from the difference between deliberative and intuitive actions. The third is less obvious, but immensely powerful—it’s called cheating.
The conscious, deliberative route is the one that most of us are familiar with already—it entails encouraging people to take action, and them consciously deciding to do it. Users have to pass through all five stages of the Action Funnel, and often spend considerable time on the conscious evaluation stage.
The intuitive route is a bit more complex. Recall from Chapter 1 that our lightning-fast, automatic, and intuitive reactions arise from a mix of various elements: associations we’ve learned between things, specific habits we’ve built up, our current mindset, and a myriad of built-in shortcuts (heuristics) that save our minds work but can lead us astray. Of these, habits are the most promising route to developing a sustainable path to behavior change because there are clear, systematic ways to form them.1 And once they are formed, they allow the user to pass effortlessly through two of the stages of the Action Funnel—the conscious evaluation and the assessment of the right timing for action.
The third strategy takes a lesson from Chapter 1, that our minds are usually on autopilot, to the extreme: it decreases users’ need to act altogether, so they simply need to give consent if they wish the target action to occur. This strategy—to “cheat”—I’ll argue is the most effective and desirable of all.
In the following sections, I’ll go into detail about each of these strategies, and when each of them is most appropriate, but in short, here are the guidelines:
If what you really care about is the action getting done, and it’s possible to all but eliminate the work required of the user beyond giving consent, then do it.
If the user needs to take an action multiple times (like eating better or spending less), and you can identify a clear cue, routine, and reward, then use the “habits” strategy. Also use this strategy if the user is fighting an existing habit—to cleverly undercut it, rather than using brute force to stop it directly.
If neither of the other two is available, then you must help the user consciously undertake the target action. There are ways to make this process nicer and easier, but it’s still the hardest path to follow.
In each case, the individual makes a conscious choice; what’s different is what is being chosen and what happens afterward. In the first strategy, the person chooses whether to give consent to the action occurring on her behalf. In the second strategy, the person chooses whether to set up the conditions for habit formation (or for stopping an existing habit), and chooses whether to repeat the behavior until the habit is formed (or broken). In the last strategy, the person chooses whether or not to take the original target action. If the action is to be repeated, so is the choice. There’s no (ethical) way to avoid having the user consciously choose whether or not to act, but products can change the nature of that choice by selecting among these three strategies.
These behavioral strategies provide high-level direction for how the product should be designed: how it accomplishes the process of behavior change. Behavioral tactics (such as comparing users to their peers, highlighting the pain of losing an opportunity, or priming them to think about particular topics) don’t provide much overall guidance on how the product should work. Instead, they can be slotted in at various junctions in the product to make each piece of the product more effective. Throughout this book, we’ll talk about both high-level strategies and lower-level tactics. This chapter is all about strategy, though, so let’s get started with the first one: cheat.
While you can make an action rewarding, easy, familiar, socially acceptable, or any of the other things we talked about in Chapter 1, the activity still involves work for the user. Ideally, the company should find ways to shift the user’s burden onto the product, by identifying clever ways to make active participation by the user unnecessary beyond giving informed consent. That’s what I call cheating—substituting the user’s nasty problem with a much simpler one: deciding whether he wants the product to take the action for him. As you’ll see, this strategy is only available in certain cases, but when it is feasible, it is immensely powerful.
Exactly how a company can “cheat” depends on whether the target action is undertaken once or infrequently (like buying running shoes) or repeatedly (like going running each morning). I’ll talk about each of these two situations in turn.
To default an action, the company first finds a way to take the action on the user’s behalf. Then, it gives the user a choice about whether the product should take the action on his behalf, where the default answer is “yes.” The user can say “no” if he so chooses.
Most defaults are invisible—you don’t even think about them as defaults; they just happen. In fact, we’re not used to seeing the defaults that are all around us, and so we rarely think of it as a solution. To that point, the most common reaction I get to proposing defaults is, “That’s great, but there’s no way that will work here. You can’t default this behavior.” Well, maybe. Here are some examples to show how defaulting works in real life:
Two of the greatest success stories in the recent history in helping users save money are 401(k) auto-enrollment and auto-escalation.2 (For non-US readers, 401(k)s are retirement savings plans provided by an employer to employees.) Under auto-enrollment, individuals who are eligible to participate in their company’s 401(k) plan are defaulted into contributing to the plan, but are given the option to not contribute if they wish. Similarly, auto-escalation automatically increases the contribution rate over time, but the individual can opt out at any time.
The initial action that users take is often quite minimal—signing their name on a package of new-employee documents—and afterward, contributions to the 401(k) plan are automatically deducted from their paychecks and placed into their retirement account on their behalf. Instead of requiring that an individual choose to contribute to the retirement plan each month (or choose to find the HR representative with the necessary paperwork required to enroll in the plan), this process effectively removes the work required by the user.
401(k) auto-enrollment is a powerful example of increasing savings, but it also can skirt the line between voluntary behavior change and trickery. Some employers strive to inform employees about their retirement plans and default contributions. In other cases, the employees don’t know about their accounts until they leave their job and get a check—which they quickly spend on non-retirement needs, since they weren’t informed and invested in the process in the first place.
The impact of defaults is significant in this case: defaulted (auto-enrolled) plans have nearly twice the participation of non-defaulted plans (Nessmith et al. 2007).
High-end camera manufactures have a problem: many consumers want lots of features, but those same features make the camera sensitive to user mistakes and result in bad pictures.
Good cameras have a simple solution that help people take quality pictures, but still provide power options (and a premium price): the cameras have default settings that are dirt simple and would provide a good picture in most scenarios. In addition, they have all of the fancy bells and whistles that make the product more attractive and expensive than a bargain-basement camera.
Similar defaults are common in computer software (“Would you like the standard install or the scary customized one?”)—the options are there, but the software makers have provided intelligent defaults so most people don’t have to worry about them and install the software without getting themselves in trouble.
Impact of the defaults: apparently, cameras still can’t help us take interesting pictures. More seriously, though: do any mass market cameras exist anymore that don’t have intelligent defaults for things like contrast, white balance, and F-stop?
If the action can’t simply be defaulted away, there’s another clever option—have the action come along for the ride with something else that users are going to do anyway. In other words, don’t have them think about doing the action at all. Make the action happen automatically when the user does something else—something that is inherently more interesting or engaging—but leave the option for the user to decline the action if he so chooses.
Here are two examples:
OK, before I go into the solution, what’s the most effective way to improve the amount of vitamins and minerals people get? Convince them of the benefits? Pay them to eat well? Run a public campaign with celebrities endorsing vital minerals? How about this: put it in the food people already eat—with their consent, and without removing other food options. For example, put iodine in salt.
Iodine deficiency is the leading preventable cause of mental retardation ([ref129]).3 It causes stunted growth, infant mortality, lower IQ, goiter (big lumps in the neck), and more ([ref93]). Two billion people suffer from insufficient iodine. Iodine costs virtually nothing to produce and add to salt.
The story of iodized salt also shows that defaulting can’t be allowed to turn into coercion, either practically or ethically. At various times, people around the world have objected to iodine being added to their salt without their consent, causing these iodization campaigns to fail (and iodine deficiency to continue). When there is no way to opt out (non-iodized salt) and no consent, it’s not “defaulting”—it’s just an ethically problematic mandate. There must be consent among the population first.
Impact of making iodine incidental: in many of the places where iodized salt has been used (with consent), the problem of iodine deficiency simply ceases to exist; that’s the ideal outcome of any behavior change strategy. In the United States, iodine deficiency is rarely an issue anymore, except where it hasn’t been made incidental.
One solution in this case is a savings lottery, aka prize-linked savings accounts. A prize-linked savings account is like a lottery in which a person can “buy” multiple tickets.4 Each ticket is a contribution to their savings account. Like any lottery, there’s a jackpot—a big pool of money that one (or more) winners get. Unlike a normal lottery, the participant doesn’t lose the cost of their ticket: it’s just deposited into their savings account ([ref175]). There’s a significant upside, but little downside, to participating.
Prize-linked savings makes contributing money to a savings account incidental because some users have a strong preference for playing lotteries ([ref64]); the fact that they don’t lose the money they use to play is a nice added bonus, but incidental.
Impact of making it incidental: prized-linked savings programs have been highly popular around the world for centuries, starting in Britain ([ref134]). They have recently gained traction in the United States through the tireless work of Doorways to Dreams, a Massachusetts-based NGO.5, 6
And there are many more examples that we rarely think about in our daily lives. If you want your toddler to take a pill, you crush the pill up and put it in some juice that he likes. The toddler doesn’t care or know (and if he doesn’t know, he can’t complain) about the pill; it’s incidental. The juice is what matters.
With repeated actions, you can use both of the last two approaches (defaulting and making the action incidental). For example, with prize-linked savings programs, the savings lottery can be repeated each month to encourage sustained savings contributions. Each time the person acts, savings is incidental. Similarly, each time the person uses an application, she can encounter the same (configurable) defaults.
In addition to these two approaches, another approach becomes possible with repeated actions: you can turn the repeated action into a one-time action by automating the act of repetition.
Taking an action repeatedly is inherently more difficult than taking that action once, even when the person learns how to do it better over time. So, why not turn a repeated action into a one-time action?
In this scenario, the individual takes a one-time action to set up or accept the automated process, and then the rest is handled without her intervention by the product itself. The principle is simple, and is very similar to defaulting a one-time action: use behind-the-scenes magic to shift the work from the user to the product.
Some great examples of automating repeated behaviors in the health space are exercise trackers that people carry with them throughout the day. These include devices such as the Nike+ FuelBand, Jawbone Up, and Fitbit One; and apps (e.g., RunKeeper) that use GPS or a phone’s accelerometer to accomplish this without a separate device (see Figure 1-2). These apps and devices automatically log and compare exercise against a user’s target. They’ve successfully taken something that is annoying but beneficial (logging exercise in a journal, comparing it to one’s daily goals), and made the work magically disappear. Once exercise tracking was automated away, companies could focus on more interesting (and user-beneficial) target actions—like helping users exercise more.
Another example of automating behavior comes from the personal finance space, with software that automatically categorizes transactions and tracks spending—such as HelloWallet (where I work), Mint, and numerous bank websites. In the “old” days (i.e., the 1980s), if you wanted to know how much you had in your checking account, you had to track your spending and balance your checkbook (remember checks?). When ATMs become popular in the 1990s, you also had to track your cash withdrawals. If you had a credit card, it would send you a monthly statement, but before that arrived, you were out of luck.
With personal finance applications, tracking expenses can happen automatically. Each individual transaction is automatically logged, categorized, and, where relevant, compared against a goal or budget item. As with many other forms of automation, once the action is automated for the user, the product team is then free to focus on more interesting and difficult-to-change behaviors—like helping users stay within their budget. But that wouldn’t be feasible for most users if they are wasting their time tracking their spending first.
The most powerful combination of all is to combine automation with defaulting—automation makes it a one-time action, and defaulting makes it little more than an acceptance of that automation. I didn’t go into detail about this earlier, but 401(k) auto-enrollment is such an example—the savings contributions are automatically deducted, and the default is to enroll in the program.
Before I move on to other behavior change strategies, I’d like to confront an implicit assumption that I’ve seen in many do-good products—that doing good requires making our users work hard. If we, as people designing for behavior change, want to help people take an action, we should be pushing people to tackle that hill! We know it’s hard, they know it’s hard, and that’s what makes it worthwhile, right?
If the goal is to make people healthier, and the action is consistent with that goal (say, by making the food that people already eat magically become healthier but taste and cost the same), and automation doesn’t have nasty side effects, does it inherently matter if the user doesn’t have to work hard for it? This probably raises all sorts of hackles. In the case of magically healthier food, I can hear my own inner do-gooder say, “Well, that misses the point—we want people to make wise choices, learn about the wonders of nutrition, be grateful for all the energy we put in to help them, etc.!”
That’s why it’s vital to be clear about the end goal of the product. For example, educating people about health is a laudable goal. But do we really only want to educate people? Or, do we educate in order to help people change their eating habits, which then makes them healthier in the long term? If we could jump ahead, solve this very particular problem, and move on to something else, wouldn’t that be a good thing? Maybe making food healthier helps with the goal of decreasing vitamin deficiencies, but it doesn’t solve the issue of cardiovascular disease. Great—once the food solution is in place, then you can devote your energies to the next problem: helping people decrease cardiovascular disease.
Any product will have multiple aims. But there should be one clear thing that you gauge its success against—a final outcome or goal (that one thing can be a composite of multiple smaller things). I’ll talk about how to identify and fine-tune the product’s goal in a later chapter, but let’s assume you know what it is. When you’re clear about what exactly is being sought, go for it. Even if it feels like cheating because it doesn’t make people suffer. There are no martyrs in beneficial behavior change. The point of making work magically disappear is that you can move on and help your users with other, more intractable problems.
There’s good behavioral science behind this point too. In short, our self-conceptions are constantly adapting based on our own behavior. We often forget or ignore the reasons why we do things and develop a story of who we are based on what we observe about our own behavior ([ref188]).7 For example, if we successfully contribute money to a retirement plan, even if we were defaulted into it, we suddenly feel that that’s something we can do—we’re savers! The pride that people feel at saving money through automatic enrollment is real, and should not be discounted. That self-conception as a saver then has knock-on effects for other related behaviors—we’re prepped for future action.
A classic study in this field is [ref73], in which the researchers started by asking homeowners to put a small sticker in their window encouraging safe driving. Weeks later, this randomly selected group was far more likely to accept a large lawn sign about safe driving than other homeowners; a whopping 76% of them accepted the large sign, compared to 17% who hadn’t been asked to show the small sticker. In other studies, homeowners were also more likely to accept other non-driving related lawn signs. The homeowners started to see themselves as people active in their community, which had broad effects on their behavior.
There are cases when this doesn’t work, of course. When people don’t know the action occurred at all, then the self-conception doesn’t change—but in that case it isn’t a voluntary behavior change at all: it’s behind-the-scenes trickery.
Remember the Create Action Funnel from the last chapter? It’s difficult for a user to pass all of the way through the funnel from the initial cue to a conscious choice to act with sufficient urgency. The cheating strategy takes the funnel, and changes its meaning. With a conscious choice to take a hard action, “success” occurs when the user passes through the funnel. When the product cheats, “success” occurs when the user agrees to the action occurring, but doesn’t pass through the funnel to stop it from occurring.
Habits are widely used in products that change behavior. For example, the Nike+ FuelBand builds the habit of checking progress throughout the day with a simple cue (the band itself), routine (checking NikeFuel points), and reward (seeing the points increase). That habit of checking progress is an essential part of the feedback loop used to encourage additional exercise. The mobile phone application Lift builds user-selected habits by cueing people each day to take their action, and then recording it and rewarding them.8 Before talking about how to build them, let’s look at the role they play in behavior change.
Our minds are built to form habits because they are essential—they free our conscious minds from handling mundane details, for everything from thinking about how to prepare our breakfast in the morning to how to properly greet a friend we meet frequently. Without habits, our conscious minds would be overloaded with the minutiae of daily decisions—it would be like we were always moving to a new house, a new job, a new city, and trying to consciously maneuver our way around, every single day of our lives.
Before a habit is formed, however, the user still needs to choose to act. Just like with automation and intelligent defaults, using habits doesn’t remove the need for the mind to think; it simply shifts the task to a simpler problem. With habits, the simpler problem is to help the person “get into the groove”—to start taking the action, so that the mind can make it habitual. I’ll talk about how to consciously start taking an action in the next strategy; for now, we’ll focus on how to build the habit given the user’s initial commitment to act.
From a behavior change perspective, habits are both a boon and a bane. If a product helps the user form a habit, then the person can act on autopilot. The product team, and the user, can move on to tackling harder problems—just like the “cheating” strategy discussed earlier. But “bad” habits work on autopilot just as much as “good” ones. The next two sections talk about how to form (hopefully “good” habits) and clever ways to attack existing “bad” ones.
Chapter 1 described the two basic types of habits: habits created out of simple repetition (cue-routine, cue-routine, etc.), and habits that have the added feature of a reward at the end (cue-routine-reward) that drives the person to repeat the behavior. Your product’s users could form habits by simple repetition, but then the burden of work and willpower is all on their side. When designing for behavior change, add a reward at the end to help bring people back while the habit is forming.
The cue-routine-reward process is depicted in Figure 1-3. Charles Duhigg popularized the process in the Power of Habit (Random House, 2012), building on an old tradition in applied behavioral analysis.9
To build habits with a product, here is a straightforward recipe:
Identify a routine that should be repeated dozens of times, without significant variation or thought each time.
Identify a reward that is meaningful and valuable for the user.
Make sure the user knows about the cue, routine, and especially, the reward.10
Make sure the user wants to and can undertake the routine (i.e., the user needs to make the conscious choice to act).
Deploy the cue.
Either facilitate the routine or, at least, seamlessly and immediately track whether the routine has actually occurred.
Have the product immediately reward the user once the routine has occurred. That allows dopamine in the brain to reinforce neurons associated with the cue and routine before the memory fades.
Repeat steps 6–8, tracking completion times and rates, and adapting the process until it’s right.
There’s a lot of nuance there, of course.
First, the cue really needs to be single-purpose and unambiguous (i.e., after the habit is formed, the cue is linked to the specific routine and nothing else), because you want to avoid the mind having to think about what to do when the cue occurs. [ref71] argue that the triggers (i.e., cues) can be:
Directly tied to another event (e.g., looking at the bathroom mirror first thing in the morning is connected to picking up your toothbrush)
At a specific time of day every day or every week
The trigger/cue can be “internal” (boredom or hunger) or “external” (seeing the clock first thing in the morning, or getting an angry email). Internal triggers are great, since they are inherent in the human condition; however, lots of other things in one’s life compete for the same triggers (which makes them not single-purpose and thus ambiguous). External triggers can be just as effective, if wisely constructed.
Second, while the routine must be structured so that it can occur effectively without thought, it need not be “stupid” or “simple.” Good driving, for most people, is a (complex, impressive) habit. Remember how hard it was to learn to drive? Remember all of the thought that was required just to start the car and get it going? Yet, after learning, we avoid getting too close to other cars while on the road, we coordinate what our eyes see with what our hands do to steer, and so on. The reason is that driving uses a set of hierarchical habits—large, complex habits built out of thousands of small, routinized behaviors that are cued from the environment and linked to one another in succession. Each piece is structured so that it can be consistently executed after the cue without conscious thought.
Routines that can be made into a habit often will have a strong and clear feedback loop (i.e., after the action is taken, the reward is immediate and unambiguously tied to the success). Habit formation is not a conscious event, though we can consciously put ourselves in situations where we’ll learn them.
Third, the reward need not be offered every time, as long as it is still clearly tied to the routine. Random rewards are quite powerful in some circumstances. In the operant conditioning literature, habits with random reinforcement take the longest to form but also take the longest time to extinguish once the reward is no longer given. Gambling provides the ultimate random reward—and once you have the bug, it’s difficult as all heck to get rid of. One reason that random reinforcement is so powerful is that our brains don’t really believe in randomness. We look for patterns everywhere. So, part of the desire driving a random reward is our brains trying to find a pattern (ever talk to a gambler who has “a system”?).11
And finally, a key part of using products to build habits is experimentation and fine-tuning. Your product is probably going to get it wrong the first time—the cue won’t be clear or won’t grab the user’s attention, the user may stop caring about the reward, or the context for the routine might change, and conscious thought is required.12
This book is about helping users take action. Sometimes, though, that may require intentionally stopping a habit, instead of just adding new behaviors. For example, at some point, improving fitness through exercise means not just exercising more, but also sitting less. And that means overcoming an existing habit.
Unfortunately, it can be extraordinarily difficult to stop habits head-on. Brain damage, surgery, even Alzheimer’s disease and dementia sometimes fail to stop habits, even as other cognitive functions are severely impaired ([ref56]). BJ Fogg, for example, argues that stopping existing habits is the hardest behavioral change task to undertake (2009b).
Why are habits so difficult to change? First, it’s because habits are automatic and not conscious. Our conscious minds, the part that would seek to remove them, are only vaguely aware of their execution (see [ref44]); we often don’t notice them when they occur, and we don’t remember doing them afterward. Across dozens of studies on behavior change interventions, researchers have found that the conscious mind’s sincere, concerted intention to change behavior has little relationship to actual change in behavior ([ref183]).
Second, it’s because habits never truly go away—once a habit is formed (i.e., the brain is rewired to associate the stimulus and response), it doesn’t normally un-form. It can remain latent or unused, but under the right circumstances, that circuitry in the brain can be activated and cause the habitual behavior to reappear.13
Another way of thinking of habit cessation is this: if stopping bad habits were easy, we wouldn’t need so many darned books on everything from stopping smoking to dieting.14 Nevertheless, one can draw lessons from the literature on habit formation and change—which can save product teams needless pain and suffering. There are five main options that product teams can take to handle an existing habit:
Avoid the cue.
Replace the routine.
Cleverly use consciousness to interfere.
Use mindfulness to avoid acting on the cue.
Crowd out the old habit with new behavior.
In each case, the person doesn’t engage in a direct confrontation to simply suppress the habit. That takes constant willpower, which is finite and unsustainable in most cases.
The cue signals the brain to engage in the problematic behavior; one way to stop a habit is to avoid the cue ([ref186]). For example, in addiction counseling, counselors advise addicts to change their environment so that they don’t encounter the things that remind them to act. If you always stop for a drink when you see the bar on the way home, then change your route home so you don’t see the bar anymore.15
Designing a product to help people avoid cues is especially tricky. First of all, most cues for bad habits are, by definition, outside of the behavior change product. People use the product in order to change the habit—the product didn’t cause the bad habit. So, the product must help the person avoid the cues themselves: the product must provide guidance and instruction. And the individual must first know what the cues are—and be able to successfully avoid them.
Second, because the routine is outside of the product, the application usually won’t know if the person has engaged in the behavior. It’s up to the user to report falling off the wagon—which is doubly difficult. External monitoring systems are required—like the breathalyzers that alcoholics install in their cars to stop them from driving drunk. Much more is required in the case of chemical addictions like alcoholism, but we can learn from these efforts as we design products to stop less intractable habits.
While this route is clearly challenging, there are products that have successfully done it. One example is Covenant Eyes16—software that helps people who are struggling with sexual addiction or want to avoid the temptation before a habit is formed (see Figure 1-4). It helps users avoid cues (by filtering out sites with explicit content) and/or automatically monitors web usage to inform accountability partners of when the person does access pornography.
The other strategy that products can use to change a bad habit is to transition an existing cue and reward to a different (more beneficial) behavior. In The Power of Habit (Random House, 2012), Duhigg describes two elements that are needed: routine replacement and a real belief that the habit can change.
Routine replacement works by hijacking the cue and the reward, and inserting a different routine between them. He uses the example of taking a snack break when you’re not really hungry. The cue may be that you’re having a down moment at work or watching a commercial on TV. The reward would be the relief of (momentary) boredom and the pleasant crunching sensation of the snack. To hijack this process, one needs to:
Identify the trigger, and the reward (if appropriate).
When the trigger occurs, consciously engage in a different routine that provides a similar reward (like doing a crossword puzzle when bored during commercials).
The process of consciously replacing routines is also known as “competing response training.” It is used in the treatment of people with Tourette’s syndrome (involuntary tics), and has shown dramatic results in experimental testing ([ref149]; [ref44]).
For especially difficult habits, like smoking and drinking, swapping in a new routine isn’t enough, though. The new reward is never quite like the old one. Swapping can handle everyday behavior, but when times are tough, people can be immensely tempted to “fall off the wagon.” Something else is needed to get through those dark times and back to the day-to-day humdrum that they can handle. That something else can be faith that the hard times will pass. It can be a religious faith, a personal faith in themselves, or a faith in others that pulls them through. Either way, it’s an internal narrative that things will get better.
How does routine replacement work in practice? One of two ways. First, you can ensure that the product itself is present at the moment when the cue normally occurs. At that moment, it would remind or entice the user to do the new routine instead of the old one. After the routine is done, it would reward the user—or encourage him to reward himself.
The other route is trickier and is needed when the product isn’t present when the user encounters the cue. As with avoiding the cue (described in the previous section), the product must advise and prepare the individual for the moment of temptation, and find some way of tracking what action the person took. ChangeTech.no has an intensive program of support and tracking that accomplishes this, with over 400 points of contact with individuals during their smoking cession program. And, its method has shown positive results in randomized control trials ([ref24]).17
An example of in-the-moment hijacking of habits that we’re all familiar with is shopping in brick and mortar stores with a smartphone:
Cue. See a camera, computer, or something else you like.
Old routine. Pick it up, go to the cash register, buy it.
New routine. Look it up on the phone, compare price (usually lower), and buy it.
This habit hijack is killing brick and mortar stores. It’s not a “beneficial behavior change,” but it’s the same underlying process.
Our big brains are really good at blocking our own autopilot; properly deployed, they can interfere with habits in progress without requiring direct willpower to overcome the action. Thinking = bad, for a habit at least. In sports, masters of their game sometimes “choke” because they consciously cut into a process that normally runs on autopilot, and this happens in any field of mastery ([ref13]; [ref74]). To interfere with a habit: think about it. Look especially for what triggers it. Then closely examine the routine that’s normally automatic—just by thinking about it (consciously), we can interfere with its smooth execution.
Products that do this should be present at the time of action and can grab the user’s conscious attention to his or her behavior. The Prius is well known for functioning this way. The car’s consumption monitor provides ongoing, immediate feedback about the car’s gasoline consumption. This in-the-moment feedback can break people out of their existing driving habits by making them consciously aware of what’s going on, causing them to use less gasoline, aka “the Prius Effect.”
In order for this approach to work, like all habit intervention (and habit formation) approaches, it must be voluntary. If someone doesn’t care about mileage or finds the car’s consumption monitor annoying, he won’t listen to it. It starts with the conscious choice to act.
Another, subtle way to overcome bad habits is by employing mindfulness. Mindfulness is a concept used in Buddhism to refer to awareness of the present moment and its experiences, without judging or trying to control them. It’s a mental state of openness and acceptance of events and sensations as they occur. Mindfulness-based therapies are increasingly popular in the treatment of mental conditions, such as acute stress, anxiety, and depression ([ref92]). Similar to mindfulness meditation in Buddhism, these therapies entail an intentional focus on the present moment without interference or judgment ([ref162]).
By bringing the cues that trigger habitual behavior into conscious awareness, it’s possible for one to be aware of the trigger without acting on it. For example, mindfulness has been shown to limit undesired, but habitual, binge drinking ([ref29]). A number of apps, such as Headspace,18 support mindfulness to reduce stress or increase focus, though do not target habit change in particular.
Another way of approaching habit change is to crowd out the old habit with new behaviors—a method that combines option 1 and 2 (and sometimes 3). In this method, you focus on doing more of what you want instead of less of what you don’t want.
For example, think about someone who is in bad shape, spends lots of time watching TV, and has bad eating habits. The person starts to go to the gym to exercise more (creating a new habit). As the person goes to the gym, he meets new people, and enrolls in exercise and cooking classes with them. Slowly, the amount of time available to watch TV decreases. The person simply isn’t at home as much, which leads him to avoid the old cues to watch TV. Also, because of the cooking class, and new ways of eating and cooking, there simply isn’t the hunger and opportunity to use his old eating habits; they are slowly being replaced.
Naturally there are multiple forces at work in his life, such as changing self-identity, and changing social norms. However, as the structure of his daily life changes, the old habits fade away—not through a direct assault, but because other things are taking up his time and satisfying his hunger pangs. This only works if he gets far enough down the path of habit change—and doesn’t quit going to the gym soon after signing up, as so many other people do. The initial choice to push ahead, before the habit is formed, is a conscious one.
You probably noticed that both of the previous two strategies involved removing user work and simplifying the problem. But the simplified problem still requires a conscious choice to act. That’s unavoidable, and it’s even a good thing. The conscious mind must be engaged at some point. Ideally, that interaction entails informed consent—and the product automates or defaults the rest. Or the conscious choice can be to start down the path of habit formation or habit change.
The explicit strategy of making a conscious choice over each part of the action should generally be avoided because it requires additional effort on the part of the user. All else being equal, more effort means less chance of acting. But sometimes, a head-on approach is used (or required). That means passing through all five stages of the Action Funnel (cue-reaction-evaluation-ability-timing), in order to execute the action itself.
The conscious choice to act is a primary focus of this book, and Chapters 6–8 cover it in great detail. They describe how to support users to take conscious action—whether the action is the original behavior that the company sought to address, or a restructured action like giving consent to an automatic action taken on their behalf.
This chapter discusses the three primary strategies for behavior change. In each case, there is a choice that individuals make, but the nature and subject of that choice changes. Here’s a quick recap of when to use each one, and how they fit into the larger picture of behavior change:
What it is. Help the user avoid the work of the action altogether by making the outcome occur by default when the user interacts with the software or when the user takes a different action, or fully automating a repeated behavior after consent is given.
What is consciously chosen. Whether or not to give consent to the action occurring on the user’s behalf.
Examples. 401(k) auto-enrollment; substituting healthier ingredients into the food people already eat.
Use this strategy. When you can replace a hard action with informed consent. This is not appropriate for overcoming ingrained habits, nor is it appropriate for cases in which the user needs to personalize the action to specific needs—that requires conscious, active involvement.
What it is. Help the user avoid conscious effort and thought by making the desired action an automatic response to a trigger. Or when changing habits, cleverly attack the habit’s structure to hinder it from occurring.
What is consciously chosen. Whether or not to set up the conditions for a habit to form (or be broken).
Examples. At the supermarket, go down the produce aisle before the canned foods aisle; walk once a day.
Use this strategy. Whenever the user wants to undertake a behavior that is done multiple times, in a consistent context. Also use this when trying to overcome existing habits, with the tricks described as part of the strategy, rather than using brute force to consciously override habits.
What it is. Help the user think about the action, and take the necessary steps (consciously) to make it happen.
What is consciously chosen. Whether or not to take the target action.
Examples. Educating people to get a good mortgage; encouraging people to sign up for (and attend) a yoga class for the first time.
Use this strategy. Whenever the first two strategies aren’t feasible, especially when the action is complex, novel, and requires the user to make numerous small choices that can’t be defaulted.
Look for technical solutions to remove user work wherever possible; it’s often much more effective to engineer a solution than it is to change behavior.
Three technical solutions are automating the action behind the scenes, using intelligent defaults, and making the behavior a side effect of something else the user is doing.
Habits are immensely powerful ways to lock in repeated behaviors. They require an unambiguous cue, unvarying routine, and a meaningful, immediate reward.
When possible, avoid trying to stop an existing behavior. The product should build new ones instead.
If stopping a habit is required, help users avoid cues, replace the routine, crowd it out, or draw conscious attention to the trigger and routine via mindfulness.
No matter what, some type of conscious choice is required for voluntary behavior change—the strategies presented here are all means to simplify the choice and the work required of the user.
1We will talk about the other two elements (associations and shortcuts) later, but they are really tactics that can improve any other behavior change strategy—assuming you know enough about the population you’re working with.
2For many Americans, the behavior change isn’t what our policy makers and companies intended—it’s become a short-term savings vehicle. But the impact on savings is still amazing. See [ref60] for a discussion of the downsides of auto-enrollment and auto-escalation.
3From [ref129]: http://www.nytimes.com/2006/12/16/health/16iodine.html?fta=y&_r=0. Though, to be fair, I found that citation from Wikipedia: http://en.wikipedia.org/wiki/Iodised_salt.
4Legally, prize-linked savings is usually structured as a sweepstakes in the United States; no need to get into the legal definitions here though.
5One could also point to state lotteries as examples of making a behavior, contributing money to schools, incidental. In California alone, they have funneled $24 billion to “education funding” since 1985 ([ref168]). But state lotteries are also a great example of how different the mind is from a conscious budgetary process. In our minds, school contributions are a side-effect of our lottery purchases. No extra work. In reality, state budgeters consciously know that a dollar is a dollar, and move the money around from one budget category (schools) to another. Changing lottery participant behavior doesn’t mean that you’re also changing the behavior of accountants!
6Another great example of making savings behavior incidental comes from the IDEO/Bank of America “Keep the Change” program. The program rounds up purchases made on debt cards to the nearest dollar, and takes the different between original cost and and the rounded version and automatically deposits it into the person’s savings account. The person does nothing differently—savings is incidental. See http://www.ideo.com/work/keep-the-change-account-service-for-bofa/ and https://www.bankofamerica.com/deposits/manage/keep-the-change.go.
7The related tendency to create consistent (non-dissonant) stories about our own behavior has been in everything from North Korean gulags to the “foot in the door” technique of successive commitments in sales. See [ref33].
9This cue-routine-reward is a clearer presentation of the antecedent-behavior-consequent (ABC) model used in Rational emotive behavior therapy and applied behavioral analysis. See Miltenberger (2011) for one application of the ABC model.
10In studies of classical conditioning with animals, you actually don’t need to link the cue, routine, and reward beforehand. You can build a habit around simple trial and error. However, with us humans, and especially with voluntary behavior change, you can skip the trial-and-error part and tell people what’s going on.
11But where we expect a strong pattern, and don’t get it, we’re angry. Would you be happy if you went to Starbucks for your morning coffee, and some days the coffee was terrible, and other days it was great? That would be random reinforcement.
12There’s much more that one could say about designing habits, but my goal is not to exhaustively cover them here. Numerous books have been written about forming and breaking habits in different contexts; [ref55], [ref44], and [ref59] are three good places to start. In addition, BJ Fogg has developed a new hands-on approach to creating habits in one’s own life; see http://tinyhabits.com/. For now though, my goal is to give enough of a foundation that a product team can make a solid product plan, and then learn what really works for them in their particular context.
13And, in cases of chemical addiction, there are added layers of difficulty that make defeating addiction beyond the scope of this book. For example, drugs can cause brain’s receptors for key neurotransmitters to change, requiring additional levels of stimulation to receive the same experiences that were had before the drug was used. While many of the techniques described here are also used for addiction, I don’t try to cover the extensive research on addiction.
14They’re difficult to change on their own, and, of course, there are also other factors associated with the habit that make it more difficult to change—like peer pressure, chemical addiction, etc.