Illustration by Mallory Haack

Guiding Principles for Choice Architecture

How to help users predict their future needs and wants — Designing for Choice Architecture Series — Part 2 of 3

Common interface principles focus on organizing information within an interface by hiding, displacing, organizing, and removing information to increase simplicity. [1] Let’s go beyond visual design principles and dive into principles for reducing decision fatigue, helping people make a decision, and how to frame options. These principles can help guide your user research strategy and improve your design concepts.

This is article 2 of 3 in the Designing for Choice Architecture series. Check out the previous post, Choice Architecture: Introduction to Designing for Decision Making, if you haven’t already.

Principles for Reducing Decision Fatigue

The way people approach a one-time decision is often different than how they will repeatedly use a tool that helps them make decisions. The following principles tend to work best for systems that are repeatedly used:

1. Cut
2. Concretize & Model
3. Categorize
4. Condition for Complexity
5. Compare
6. Context
7. Cause

Cut

Distill choices to a manageable number and differentiate each choice. The system needs to help people easily differentiate options through its information architecture and visual design. A drawback of cutting is that a large set of options are often appealing to people, even though it increases decision fatigue and reduces action. [2] Showing people that the small set of options are drawn from a wider pool of options conveys a sense of customization while making the system seem robust.

Concretize & Model

It’s helpful to provide people with a potential outcome based on the choices they have selected. The model you provide can instill confidence on the part of the user by sharing expert advice or by using insightful data based on past behaviors or behaviors from a similar group of people. Be transparent about how the forecasting works.

For example, adults who are on the fence about going back to school to improve their career prospects may feel more comfortable if the interface prominently displays student outcome information such as: testimonials, average salaries, employment rates, number of jobs in their city, etc. This kind of information will help them better envision their outcome and may reduce their reluctance to get more information about going back to school.

Categorize

Categorization based on highlighting desired attributes can help demonstrate expertise. Categorizing by positioning (sort, delineation, grouping) can trigger anchoring, the primacy effect, and even loss aversion. What a system first shows should attempt to be the most relevant when possible. Categorization based on classifications allows people to clearly see how a set of options differs from another and to understand the relevant qualities and components of each choice. People will have a bias towards what you show first, so the system should take this into account. More on this in the Nudge Principles below.

Condition for Complexity

Utilize progressive disclosure: start with simple questions/tasks, build momentum, and end the process with the more difficult choices. This allows the recommendation or tailored results to be perceived as trustworthy. At any point in the journey, a person should have a sense of what decisions they’ve made and what’s next.

Compare

People find making decisions easiest when they can compare options based on discrete attributes. Make sure the attributes they are comparing have a frame of reference. The purpose of some systems is to help a person plan for or achieve an outcome they haven’t experienced before. e.g. First time home buyer or trying to plan a vacation to a region the person have never been to before. In these cases the choice architecture needs to understand what the user’s expectation will be regarding the outcome of their decision and either help her reframe her expectations or to provide options to meet and exceed those expectations.

We should attempt to concretize data that is being compared. If we can highlight attributes that people value it may help guide them through the comparison. The way a person gauges the value they will receive from an outcome differs when they have previously experienced similar outcomes. A person who hasn’t made a similar decision before will have a harder time comparing.

A simple example is a scenario where a person wants to go out to eat and doesn’t know where to eat. Deciding between restaurants she has been to before isn’t difficult because she has experienced each and can predict the outcome of dining at any of her options. Deciding between trying a new restaurant versus going to one of her goto restaurants is often more difficult because she have to assess the value she’ll get from a restaurant she has never been to before.

For some diners, the novelty of a new experience provides so much value it’s an easy decision. For many other people the fear making a bad choice can be crippling. The opportunity cost of picking a new restaurant and not enjoying it often prevents people from trying something new. This is compounded when trying to pick a restaurant to go to with a person you don’t regularly dine with. Researching your audience segments to make sure the system supports different scenarios is key to making a successful and flexible system.

I’ve made example diagrams called task models of similar scenarios, in the follow-up article, Communicating Decision Making via Task Modeling.

Context

Frame decisions and provide contextual help to reduce information fatigue. Show the person how to get to the next step and how the micro-decisions they are making might affect the outcome. For example, tooltips which define terminology or when filters provide previews of the number of results they will return can be helpful. Understanding a person’s context is important. Letting people pause their journey and return to the decision later is huge. Sometimes a user needs to get additional information from outside sources or don’t have time to finish a process.

Cause

Motivate people by reminding them of their motivations. Encouragement and incentives can go a long way to help people move forward.

Nudge Principles

These principles focus on how to frame information to incentivize and assist a person to come to a decision.

1. Provide incentives
2. Map decisions to outcomes
3. Defaults should be purposeful
4. Give feedback
5. Expect and design for errors

Incentives

Tap into intrinsic (desire for mastery, need to belong, power of stories) and extrinsic (future success, desired perception) motivations. [3] This is similar to the cause principle above.

Understand Mappings

Forecast how a person’s decisions may affect his or her goals. This is similar to the “concretize & model” principle above. To be able to forecast you need to be able to show how any choices made in any part of the decision making process can affect the outcome.

Defaults

When possible and applicable set good defaults to reduce barriers. People will often interpret defaults as a subtle recommendations so make sure the defaults of the system align with the user’s needs. Depending on the choice being made there can be important ethical concerns for the decision maker and the public. [4]

Give Feedback

Show how a choice affects progress within the system.

Expect Error

Make sure that undoing selections is simple. Additionally try to use feedforward mechanisms when possible to reduce error.

Framework for Making Optimal Decisions

These principles are good starting points to help users better understand process and considerations they need to go through to achieve their desired outcome.

1. Help set goals
2. Value hierarchy
3. Display options
4. Assess options
5. Choose

Set Goals

Make the goal of each step within a flow is explicit at the beginning of the process. If done well you can inform and even prime users with a metric they could use as they compare options. Good onboarding should trigger a motivation (incentive) with the person.

Value Hierarchy

A helpful system can try to understand which factors a person cares about the most. e.g. If trying to explore new career options ask them about what their goals are: Salary? Job stability? Speed of graduation? Financial cost? etc. These preferences can be used to highlight relevant information throughout the journey. If done well the system can create a compliance effect wherein the person identifies with a goal they have chosen and they then subconsciously use that goal as a “measuring stick” (also known as the “escalation of commitment” heuristic). These effects not only make it easier for people to make decisions, but they may also feel more confident in their prior decisions.

Display Options

Allow people to survey their set of options. If the system combines its recommendations with cutting, categorizing, and good defaults it can reduce fatigue.

Assess Options

Help them evaluate how likely each option will support their desired goals. If we know the importance of certain criteria the system can give feedback on how well a result is matching that criteria.

Choose the Winning Option

Generally speaking a (rational) person would rather use the consequences of a choice to modify her future goals, adjust the importance she assigns them, and then modify the way she evaluates the future possibilities. Unfortunately certain decisions such as buying a home or choosing a career have too large of a feedback loop. In this case give supportive information about other people who have made the same choice and what their outcomes have been.

Research Required

Principles are good starting point to help you quickly reflect on the strategy you have, but they don’t help you realize your strategy into effective designs.

The following prompts are from the principles above:

• Tap into intrinsic and extrinsic motivations
• Utilize progressive disclosure: start with simple questions/tasks, build momentum, and end the process with the more difficult choices
• The system needs to understand what the user’s expectation will be and either help her reframe her expectations or to provide options to meet and exceed those expectations

You can’t answer those questions without understanding your users well. This means you need to conduct qualitative research and collect quantitative data. For complex domains, you may need to collaborate with domain experts to help you integrate expert domain advice into your designs. Principles don’t replace research, but they may help you realize you aren’t addressing a potential need.

Next Article: How to Visualize Decision Making

The final article, Communicating Decision Making via Task Modeling, gives examples of task models and how to utilize them in your design process. Stay tuned.

Footnotes

1. Colborne, Giles. Simple And Usable. Berkeley, CA: New Riders, 2011.
2. Iyengar, Sheena S., and Mark R. Lepper. “When Choice Is Demotivating: Can One Desire Too Much Of A Good Thing?”. Journal of Personality and Social Psychology 79.6 (2000): 995–1006. Web
3. Weinschenk, S. M. (2013). How to get people to do stuff: Master the art and science of persuasion and motivation. Berkeley, CA: New Riders.
4. Raihani, N. J. (2013, December 19). Nudge politics: Efficacy and ethics Retrieved from ncbi.nlm.nih.gov.

Additional Sources

• Iyengar, Sheena. The Art Of Choosing. New York: Twelve, 2010.
• Thaler, R. and Sunstein, C. (2015). Nudge. 2nd ed. Clitheroe: Joosr.