What’s in the Behavioral Design Toolbox?

5 min read

behavior design tools

Behavioral Design offers ideas and techniques for persuading user behavior.  Some of these techniques encourage behavior, others discourage it. Some work best on behaviors that happen once, while other techniques work best on behaviors that happen often. For each tool and technique in the Behavioral Design Toolbox, there is a purpose — and for each purpose, a tool.

Reinforcement Learning

Reinforcement Learning increases the frequency that someone performs a behavior. RL focuses on how a habit can be induced by carefully controlling the rewarding consequences of an action.  RL predictably improves how much and how often people engage with an app or product, and for how long they’ll retain using that app. It works on behaviors that people already do, as well as behaviors they do not yet do.


A Trigger (also known as a Cue), is a prompt to perform an action. When used together with Reinforcement Learning, they increase the frequency of a behavior (it’s the T in the TAFR Model). Triggers work because, often, users automatically perform certain behaviors in response to signals from their internal and external environments.

Chapter 1 explores how the association of a particular trigger with a particular action is the fundamental unit of a habit. Chapter 2 teaches you and your team how different types of Triggers work, and how you can use them to change user behavior and make your app habit-forming.

Optimal Challenge

Optimal Challenge prescribes that at a given moment and for a given person, there exists an optimal amount to challenge or push, someone to perform an action. Challenge less, and you may not help the person achieve their goals. Challenge more, and you may induce fatigue and user burnout.

Optimal Challenge proposes that not only does this “sweet-spot” exist, but that it is detectable, predictable, different between people, and changes with respect to time. Optimal Challenge can be used to induce both a behavioral increase or a behavioral decrease.

Stimulus Devaluation

People learn, through experience, that certain actions will lead to positive outcomes: that’s the core of Reinforcement Learning and the TAFR Model. But how do you un-wire that learning?

Stimulus Devaluation provides a technique for destroying a learned Trigger-Action association by introducing delay and friction between an Action and its associated Reward. The resulting experience still allows a user to perform their desired action while destroying the habit-forming nature of the experience.  Stimulus Devaluation works best to decrease behaviors that people already perform, especially those for which they’ve been positively reinforced using the TAFR Model.

Stopping Rules

An object in motion tends to stay in motion, and our behavior works the same way. Our nervous systems naturally look for “stopping cue” signals from our environment that tell us we’ve completed an action.

Stopping Rules prescribes that stopping cues can be controlled to increase or decrease certain behaviors. One commonly seen example of Stopping Rules is to remove all of the normal stopping cues from an environment. Without stopping cues, users consume much, much more. Ubiquitous infinite scroll social feeds, pioneered by Pinterest, but now seen on Twitter and Facebook to leverage this effect: the more users view content, the more content is loaded for their consumption.

Choice Architecture

When presented several options of how to behave, people often perform the default behavior. It’s no mark of weak-will or laziness, it’s part of how our brain takes shortcuts to help us make decisions efficiently.

Choice Architecture prescribes that, as a Behavioral Designer, you can design someone’s environment as to create specific, intentional default actions they will take. Popularized by Sunstein, Thaler, & Balz, Choice Architecture (also known as Nudge Theory) pervades our daily life and is one of the most successful and widely-used Behavioral Design techniques. It’s best used to guide users towards a particular desired choice when faced with several possible actions they could take, and it can be used to both increase and decrease behaviors.

Ambient Communication

Often, you’ll need to present complex or dense information to someone. Ambient Visualization prescribes that using non-text communication, such as color, size, texture, pattern, motion, sound, vibration, or time can help you communicate complex content to users quickly and intuitively.

It relies on the brain’s other, non-verbal information processing streams to understand the environment faster than reading alone might afford. It can be combined with other Behavioral Design techniques, such as Reinforcement Learning, Optimal Challenge, or Triggering, to increase their efficacy. It’s often used successfully to persuade people to perform an action they need to only do once, such as a signup form or completing a purchase.

Optimal Information Flow

People process information differently from one another. For example, steps in a signup workflow that might be intuitive to one user might be wildly disorienting to another user. Someone might be able to mentally juggle many complex ideas simultaneously. Another person might focus best on one idea at a time.

Optimal Information Flow proposes that there exists, for each person, an optimal ordering of steps for a process and an optimal density of information they can be presented with before they struggle to perform your desired action. This technique works best on actions that you need people to perform once or infrequently, such as a signup form or completing a purchase.

Soft Incentives

Where Reinforcement Learning increases the frequency of recurring behaviors by variably providing a reward immediately after someone performs the desired action, Incentives work by promising a future reward for an action taken today.

Soft Incentives describes how your App may use the future promise of emotional or community-based value to encourage a single, challenging, one-off behavior today. Soft Incentives largely rely on our desire for personal accomplishment, congruence with our narratives of identity, or the approval of our friends.

Sunk Cost

When we begin to financially, emotionally, or logistically invest in things, we grow to value them more: even beyond how much we actually like them or how actually useful they are to us.

The Behavioral Design technique of Sunk Cost leverages the cognitive bias of Sunk Cost Fallacy, for which people inaccurately over-value experiences, relationships, or products because they’ve already given them time, money, information, or opportunity. Sunk Cost is a critical feature of Behavioral Designer Nir Eyal’s “Hook”\cite{eyal2014hooked} Model (which shares many commonalities with the TAFR Model) and is captured in his “Investment” phase of the Hook Model. Sunk Cost can be a particularly effective way to increase the frequency of a behavior that users are already performing by introducing a “ratchet” effect in which they become more likely to keep using you Product because they over-value the importance of their previous use and investment.

Optimal Group Structure

If your App connects people to one another, how do you know the best way each user would prefer to experience this social component?

Different users exhibit different preferences for their interactions with each other: some feel most motivated alone, others in small groups of peers, and some in the full gaze of the public.  Optimal Group Structure proposes that there exist predictable, optimal scopes and natures of interactivity between app users that will best motivate them to change their behavior.

You may have heard of the hypothesis that students have “learning styles,” and that students learn best when the material is adapted to that style (That hypothesis has been soundly repeatedly falsified). But there is a larger truth to that myth. Not everyone thinks the same way or wants the same thing. Many design teams use personas to help design for different user groups, skill levels, or user needs. New machine learning techniques are making it possible for designers to produce even more adaptive products.

Cognitive Load Balancing

By and large, the human mind is capable of juggling only a few tasks at once and capable of remembering up to about 10 things at once.

If your App requires the user to simultaneously interact with many different events, people, concepts, or ideas, Cognitive Load Balancing prescribes techniques for how you might best display only fragments of the whole information set to a user at once and effectively switch between those fragments. Limiting how much mental work the user must do at any time can improve their quality of experience and increase their ability to properly perform the target action you need them to do.

Why Start with Positive Reinforcement?

As we’ll explore in Chapters 2 and 3, Reinforcement Learning isn’t new. Beyond our intuitions of why and how positivity should influence behavior, Reinforcement has been proven effective for inducing behavior change for decades. Since its first academic explorations in the early 1950s, teams have published thousands of reports exploring the underlying brain anatomy, neural circuitry, and behavioral implications of Reinforcement. This rigorous exploration – and its continued utility in product design – grants Reinforcement a “Best Practice” status amongst Behavioral Design interventions and those interested in changing behavior.

While the Behavioral Design Toolbox contains many techniques, we’ve decided to constrain this first Volume to a dive deep on Reinforcement – and how you and your team and can leverage it to build habit-forming products.  If you would like to read more about Behavioral Design you can signup HERE to get a FREE copy of our eBook when it’s released.

Matt Mayberry