Here’s a quick overview of the ICAR framework I am using to both analyze existing situations and plan behavioral changes. It’s called the ICAR framework because it contains the following 3 dimensions: Incentives, Choice Architecture, Repetition You must use the three dimensions if you want to analyze the dynamics of a behavior or plan an intervention aiming at changing this behavior. Think of it as examining behaviors using three different Lenses that focus on different aspects of any behavior. More details on all three below. You can then use these lenses to analyze what is happening to users/consumers at different levels of granularity. You can use them to analyze in depth each individual touchpoints, or you can apply them at the behavior level or break it up into intermediary steps. It depends of your goals.
Lens 1: Incentives
Behavioral economics results focus on the situations where agents don’t act rationally according to their interests. It’s very useful, but it shouldn’t make you ignore the incentives at play. In most situations, incentives will still outweigh any cognitive bias impacting the agents. When mapping out the incentives, first distinguish between Positive and Negative incentives, and Certain/Uncertain incentives. That gives you 4 different incentive types indicated below:
Rewards (Positive & Certain)
Straightforward, but can be also Non-Monetary and Private, don’t forget those.
Potential Benefits (Positive & Uncertain)
Potential Benefits are the positive incentives that agents see as possible but not certain. Similarly, people will pool them with the certain incentives with an importance varying from person to person. They are also critical to uncover, but in some situations may be real to outsiders but not perceived by agents.
Costs (Negative & Certain)
Straightforward, but can be also Non-Monetary and Private, don’t forget those. Change is a huge non-monetary cost often underestimated.
Risks (Negative & Uncertain)
Risks are the negative incentives that agents see as possible but not certain. However, since most people are risk averse, they will be pooled by the agents with the certain incentives. Their importance will vary from person to person but will always be there. You need to find out what the perceived risks are in a situation, and there are almost always perceived risks in a behavioral change situation.
Drilling down within each type
Then for each Type, you can further refine by distinguishing between Monetary/Non-Monetary and Public/Private incentives.
Monetary / Non-Monetary
Monetary incentives are of course financial incentives. Paying people double to do behavior B compared to B will in all cases increase the proportion of people switching from A to B. Non-monetary incentives comprises positive social externalities like increased status or prestige. They also comprise non-monetary goals, especially in organizations, that result in increased power or long-term benefits. For example, if your company’s processes are changed from C to D, you have a non-monetary incentive (which we could call monetary in the long-term) to adopt D and not get in trouble.
Public / Private:
Public incentives are incentives that are clear for everyone involved. Double pay publicly announced is an example. Private incentives are incentives that are not clear to other actors in a situation, mostly because they don’t care enough to try and understand what is happening. People eligible for social benefits but not claiming them are a good example of implicit incentives. It may seem puzzling and irrational when looking at the situation as an outsider. The easy explanation, thankfully used less and less, is that these people are just stupid and not understanding they could get those benefits.
The reality is that when you analyze the explicit and implicit incentives for them, you begin to understand that for a number of reasons, including the complex process of claiming those benefits, the rational choice for them is to not claim them. If we want to change their behavior and make them claim then, the key is in understanding these incentives and changing them. Not trying to make them understand why they should assess the situation differently. If you want to work through each category systematically to be sure not to forget anything, then you can use the Detailed Incentives Analysis Matrix to assist you.
Lens 2: Choice Architecture
For each touchpoint or situation, you then look at the choice architecture elements shaping people’s behavior and that’s where behavioral economics come into play. To analyze a situation in terms of choice architecture lens 2, you actually have to break it down in 5 dimensions
Data & Information presentation
This dimension is looking at the well-known biases and human tendencies at play when looking at quantified data and other structured information. This is where framing effects, anchoring bias, etc. are at play. If the touchpoint or behavior you’re analyzing presents some sort of data or structured information, you need to look at this dimension.
Contextual persuasion and copy
This dimension looks at the contextual persuasion elements used to influence behavior. This could be the addition of an arrow to a form, shifting the order of questions, and all copy changes are also in that category. These elements are mostly medium-independent, i.e. they are valid whether you use the copy changes on digital or printed media.
Choice ergonomics and usability; user experience
This dimension is closely linked to the actual medium used within a particular touchpoint or behavior. Digital media will have different rules and impact than printed. Mobile is very different than web. Wearables are yet another category, etc. Some tools here will translate to other medium, some won’t. This category should also look at the user experience best practices existing in each field, and which are not linked (formally anyway) to behavioral sciences. Looking to influence behaviors through a mobile app? Well, you need to know the state of the art best practices of designing user interactions on mobile. Trying to nudge people entering hospitals into using hand disinfectant? Better know the best practices in terms of spatial architecture, etc.
Options and possibilities structuring
This dimension looks at the actual choices offered to agents when they have to make a decision. Do they have to actually choose an option (forced choice) or can they postpone the decision? Do you offer plan A and B or decide to add plan C as well? Do you offer them a choice between existing behavior and targeted one only or throw also an easier-to-adopt intermediary behavior into the mix available?
Finally, the last dimension to examine is to look at the ways to help people stick with their choices and actually follow through with them. If the intended behavior in a situation is not immediate, then you need to examine this.
Lens 3: Repetition
The final lens examines the elements that contribute to a repetition of a behavior. Not all behaviors are meant to be repeated though, so this is only when behavior designers aim at shaping a repeated behavior. Recycling is a good example of a behavior that should be repeated each time and is not meant to be a one-time behavior change. There are two elements to examine in that lens: triggers and reinforcers.
Triggers are the elements that will make a person repeat a behavior after the first run through it. Generally, the goal would be to start with Prompts, move on to Cues and then to Habits/Routines. Not every successful behavior change needs to progress to become a habit though. Annual health screening for example are most often prompted by a letter. If you then attend a screening, then the behavior change has been successful. You don’t need to build a habit to attend annual health screenings, all that matter is that you follow the prompts when it comes.
Prompts are triggers explicitly aiming at getting a person do an action. It can be a notification, an email, a letter, another person asking you or recommending you to do something, etc.
Cues are triggers present in the environment or as a need, that will make a person repeat a behavior. For example, if you generally use Uber for your transportation needs, if you need transport (cue) you will most likely use Uber. Or if you use Instagram and you see something you think you could post (cue), you will snap it and post it.
Habits are trigger to repeat a behavior, obviously. They are the most powerful trigger but also the most difficult to adopt and maintain over time.
Reinforcers & Deterrents
Reinforcers and Deterrents are elements that come in play after a person first complete a behavior. They affect the probability that an agent repeats this behavior by modifying his/her expected evaluation of the Incentives and Choice Architecture dimensions. Reinforcers increase the probability that a person will repeat the same behavior by increasing the attractivity of repetitions after the first run. They provide more (perceived) value in one of two ways:
- increasing the net value provided for repeating a behavior (positive reinforcers)
- imposing costs if a behavior is not repeated, i.e. increasing relative value provided for repeating a behavior by opposition to switching to an alternative (negative reinforcers).
Deterrents decrease the probability that a person will repeat the same behavior by decreasing the attractivity of repetitions after the first run.
Positive reinforcers increase the probability of a repeat behavior by increasing the expected value users will get from repeating a behavior. Personalized recommendations on Netflix or other products are a typical example of getting more value out of a behavior after the first repetitions. Experiencing how easy a behavior is if you thought it was difficult is another. Partially randomized payoffs also are a positive reinforcer, though they may also be characterized as addictive reinforcers.
Negative reinforcers also increase the probability of a repeat behavior, but they do so by imposing a cost to not repeating a behavior or switching to an alternative. Having locked-in data or data in a proprietary format is the typical example. Network effects is another, as you will suffer should you switch to an alternative.
Deterrents decrease the probability of a repeat behavior by decreasing the expected value users will get from repeating a behavior. For example, having a terrible customer experience will decrease probability of buying again the same product/service. Going through a very lengthy bureaucratic process to complete an application for a governmental service and experiencing the full complexities of doing so will decrease probabilities of a repeat behavior. For digital applications, binge-watching cat videos on Youtube instead of working may decrease the perceived self-esteem of an agent afterwards and may decrease probabilities of doing this same behavior again (which of course will compete with the effects of powerful Reinforcers…).
BEHAVIOR CHANGE PLANNING
The 3 lenses (Incentives, Choice Architecture, Repetition) of the ICAR framework are useful to analyze an existing situation or even a planned hypothetical one when discussing with policy planners, executives, etc. But when trying to plan changes to be made within a behavioral situation to actually change agents’ behaviors, then you need to map out the details of each envisioned change.
First, complete the analysis above. Then look at which dimension at which touchpoint could yield a desirable behavioral change. Finally, looking at the whole situation holistically, create a change you think will shift people’s behaviors and analyze each suggested change carefully by going through the elements below. The easiest way to do this is to look at the worksheet picture included here or better to download my Behavioral Orchestration Toolkit to get all the fillable templates.
Explain why you think such a change would impact the behaviors at play and in which sense.
Explicitly define the expected results you’re anticipating and quantify them if at all possible
Backfiring Risks / Negative Externalities
Always try hard to anticipate what possible negative externalities could arise from such a change. You can’t catch them all of course, but make an effort to really anticipate the most obvious ones.
For each backfiring risk, try to include in your change a mitigation element that will preserve the impact of the change while preventing possible backfirings.