This is a controversial topic, so it's vital to clarify one very important detail from the start: influencing user behavior is only ethical if it produces lasting and tangible value to the user.
Customers that feel used will usually never come back, but if you are capable of generating a relationship of trust between you and your customers, you will find that they will gladly share high quality data in order to improve their experience.
In this post, we'll explore how to create this type of trust relationship with your customers and different non-intrusive options to influence their behavior.
The first step in changing or influencing user behavior is to understand what it is we want to change and how we can cause it to change.
One powerful definition of behavior is “the way in which an animal or person behaves in response to a particular situation or stimulus”. This definition is powerful because it establishes a cause and effect relationship between a stimulus or situation and the reaction of an individual. For example, the stimulus could be an in-app notification and the reaction could be clicking through the notification to get more information about a product or service.
Behavioral change is a trickier concept because it relies on the notion of causality. There are three main criteria for causality: covariation, temporal precedence, and control for “third variables”.
- The first criteria implies that something (event A) actually happened and something else changed (event B).
- By the second criteria, event A has to happen before event B.
- Finally, one of the simplest ways of explaining the third criteria is the idea that event B would not have happened or would be different if event A never happened. In other words, there is no third variable (event C) causing both event A and event B.
With this context, we can now establish that our goal is to be able to perform some action or event that causes our users to react in a way that wouldn’t happen without our intervention. On top of this definition, we also want the user to be satisfied with his/her reaction, for ethical reasons.
Examples of behavior change
Now that we know what we want to achieve, we can go through a few examples of tools that are generally used to produce behavioral change:
According to Richard Thaler and Cass Sunstein, a nudge is “any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives”.
It is important to note that nudges are not mandates or prohibitions. Users have access to all potential options, but are consistently “nudged” towards one desired action due to some factor that affects how the user makes decisions, like the way in which the options are presented.
The core of the nudge approach lies in the idea that we all have cognitive and motivational deficiencies and biases that can be harnessed for our own benefit.
The most common nudge is the usage of defaults. These are the predefined options that are applicable unless the decision-maker actively switches to another option. Humans have a natural tendency for inaction, which makes this a very powerful nudge. This is also the reason why websites have to present the non-acceptance of cookies as the default option according to data privacy policies. Users have to explicitly accept privacy policies, so this nudge helps users protect their data privacy.
Nudges are very powerful and cost-effective, but also potentially dangerous. People tend to have negative reactions when they know they are being nudged, even if the nudge potentially adds value. For that reason, it is important to make sure that nudges feel like they are part of the general product flow.
In essence, nudges leverage people’s existing cognitive biases to affect behavioral changes, while boosts train people in how they can use existing or new decision heuristics. In other words, the aim of boosts is to help users improve how they make decisions through the use of heuristics.
This means that users usually don’t know when they are being nudged, but they can consciously choose when to use the competencies learned from boosts. For this reason, boosts tend to be a better option from a normative and ethical perspective.
One example of a boost is training customers of a financial institution in using rules of thumb without providing extensive accounting and financial education like “Save at least 10-15% of your take-home income for retirement” or “Use separate drawers/buckets for business and household proceeds”.
Another example is training people in temptation bundling, which helps them overcome self-control problems. The idea is to couple instantly gratifying activities like watching a TV show or eating an indulgent meal, with behavior that provides long-term benefits like exercising.
An explicit prompt corresponds to directly and explicitly asking the user to perform the desired action. One example is the use of a popup as a classical lead capturing tool, asking a user to subscribe to a newsletter when they are about to leave the site.
This option is transparent to the user but triggering an explicit prompt that is not relevant or with poor timing can be seen as annoying.
Sustained human behavioral change is difficult
Unfortunately, knowing about these tools is not enough to actually accomplish sustained behavior change. Changing the choice architecture or presenting a popup usually does not lead to behavior change unless we also take into account internal motivation and other external factors.
BJ Fogg, PhD (author of Tiny habits and founder of Stanford’s Behavior Design Lab) has developed one of the most important models in behavior change literature: The Fogg Behavioral Model.
This model argues that three main elements have to converge at the same moment for behavior to occur:
Behavior, as explained by these three elements, can be explained by the following graph:
Behavior happens above the action line. For example, the harder an action is to do, the higher the motivation the user needs to perform the desired action.
Prompts can also be thought of as cues, triggers, call to actions, etc. An alarm triggers us to wake up in the morning and walking through the kitchen can potentially trigger us to open the fridge.
This model gives us the theoretical background to design our nudges, boosts or other triggers to influence behavior change. For example, users have different motivators and we have to take this into account when designing our product and messaging.
This information can be obtained through product analytics and consumer science. For example, we can evaluate how good a trigger is by measuring user response and the lag time between a trigger and the corresponding expected behavior. If there is a high click-through rate, but no effect on the expected behavior, it’s probably not a good trigger.
A/B testing with different messages can also help us identify our customers’ motivators and the data collected from these experiments also allows us to group users in different segments according to their main motivators.
Here is a summary of different behavior models with examples on how to use them developed by Joanne Rodrigues in her book Product Analytics: Applied Data Science Techniques for Actionable Consumer Insights:
Self-actualization / self-competency
Description: Visualizing yourself completing a task; thinking that you would be good at a task.
Web product example:
Give your users a brief description of the process so they can more easily visualize what they should be doing.
You could prime some other behavior that they may be good at; make it like a game or a puzzle to complete an action.
Positive / negative reinforcement (learning models)
Description: A response that is positive and reinforces a behavior, or that is negative and causes an adverse reaction to continuing a behavior.
Web product example:
You could award points for completing a task such as writing a post and display points on a leaderboard.
You could take away an account if a user acts in a disrespectful way to other users.
Goal setting / goal pursuit (APA)
Description: To achieve behavior change, it takes two phases: setting a goal and then striving to reach that goal.
Web product example: You could offer up daily goals to improve behavior: “Remember to clean out one of your email spam folders today.” Then assess users’ achievement of that goal.
Five stages: precontemplation, contemplation, preparation for action, action, and maintenance
Description: Changing behavior takes recognition, understanding how to change, planning for change, making a change, and maintaining that change.
Web product example: Make your users aware of a behavior: “You have not visited the site recently. We miss you.” Help users plan how to change by showing them the stages of behavior change.
Fogg Model: Motivation
Description: There are different types of motivators (pain vs pleasure, hope vs fear, and social acceptance vs rejection). Touching on one of these core motivators can spur behavior.
Web product example: Users often fear social rejection. Lessen the pain of social rejection by not showing user’s rejections. Make it difficult for users to see rejected members’ profiles. Use social support as a tool for conversion. Have reviews or support of a service from friends. Determine which triggers are the most powerful for a given user from past behavior. Have you users reacted to call-to-actions that emphasize social acceptance, hope for the future or fear of missing out? Tailor your call-to-actions to what they have responded to.
Fogg Model: Ability and triggers
Description: Always make behaviors as easy as possible from the perspective of time, money, mental and physical effort, etc.
Web product example: Make it easy to change behavior. For instance, make purchasing quick and easy.
Armed with a model for understanding behavior and different tools to produce behavior change we can use product analytics and consumer science to design powerful prompts and triggers that add value to our customers.
In general, when we design products we want to try to make desired behaviors as easy as possible, reinforce positive behavior, help them set and accomplish goals, as well as generate prompts that prime users to change certain behaviors. With time this can develop into lasting patterns and habits.
Still, the effectiveness of these tools depends on our understanding of our customers. They expect us to use their data in ways that generate value for both parts. Transparency also increases the probability of them sharing high quality data with us. By learning more about their motivations and goals we can design products with nudges, boosts and prompts that are positive for our users.