The crucial component of behavioural design that no one talks about

Behavioural design, or the application of behavioural science in solving real world problems, has come along way in the last few years.

The approach has not only allowed practitioners to better understand the root causes of human judgement, decisions and actions, by turning their attention towards heuristics, cognitive biases, social drivers, and contextual factors, but has also enabled them to design products, services, programmes and communication that integrate prescriptive treatments grounded in these learnings. Some examples of these prescriptions include defaulting, social benchmarking, choice limitation, personalisation and the messenger effect.

What is more, behavioural design has been instrumental in showcasing the value of impact evaluation and experimentation. The field has shifted attention away from unproven outputs to a focus on the measurement of outcomes and the quality of work that a particular output achieves. Outputs that have prioritised these outcome-focused features, such as nudges, are often not the most elegant or aesthetically pleasing designs, but they are normally backed by sufficient evidence to suggest they will get a particular job done. They are good at what actually counts, that is, shaping user, employee or citizen behaviour. For this reason, it is necessary to build a organisation that appreciates impact evaluation as necessary and useful as well as shift practitioner mindsets to focus on the behaviour-driven outcomes that really matter.

An added benefit of integrating experimentation, and the rigorous assessment of an intervention’s impact, is that it creates a user-centric culture. Without measures in place that track the impact that a design output has on users, the success of the output is internally assessed by the designers or managers only. This is not to write-off expertise and the value of informed intuition, but in a rapidly changing digital world, where we do not have frequent, meaningful, valid, and relevant interactions with users, it would be a gamble to assume that the designer’s subjective assessment of a particular output is perfectly representative and predictive of its success.

The missing piece

As we have discussed, a thorough diagnosis of existing behaviour, evidence-based design, and systems in place to measure a design’s impact are all cornerstones of the behavioural design approach. However, there is one other component that has proven to be even more fundamental than the aforementioned three, and yet is still underrated and surprisingly easy to get wrong. It forms the backbone of any behavioural design project and, if carried out incorrectly, can often lead to difficult problems later on. I am referring to defining the desired behaviour that you are aiming to change. I like to think of this phase as the architectural blueprint that is used to build a house. If you get it wrong, it is very hard to turn back or change something once construction has started. The rest of this article is dedicated to helping practitioners ensure that they do not find themselves building houses based on poorly designed blueprints.

Setting up a behavioural design project:

Behavioural design projects should always start with an almost obsessive focus on identifying and narrowly defining the particular user behaviour that your team will be striving to change. As Kirsten Berman, co-founder of Common Cents, often says, it is important to get uncomfortably specific about the key action that you want the user to take (or not take).

Identifying this behaviour isn’t the sexiest part of behavioural design and may seem trivial at face value, but the amount of times that I have been exposed to practitioners who get it wrong has helped me to foster a deep respect for this phase of BD projects. The attention I’ve given it has helped me to identify some of the subtle traps that often crop up, which if left unchecked, can lead to catastrophes down the line.

Some of the traps practitioners can fall into:

• Thinking too big: Prioritising an outcome (e.g. increase in the total savings) without identifying and defining the user behaviours that drive it. Not being clear on the specific actions that you are trying to get users to take will create problems during both the diagnosis and design phase of your project.
• Thinking too small: Mistakenly focusing on a supporting action (e.g. users successfully opened the sent educational email) instead of the primary behaviour that you are ultimately aiming to change. Assuming that there is a positive relationship between these supporting actions and the primary user behaviour can lead to an unchecked sense of success.
• False consensus: Not spending enough time describing the target behaviour in detail and communicating that description to all major stakeholders. This can lead to conflict in teams due to misalignment, because everyone has assumed that they are on the same page when, in fact, they have subtly different interpretations in mind.

To avoid these traps, and many others, we use a structured process for identifying, prioritising and describing in detail the behaviour we are targeting. We have found this process to be crucial in setting up a behavioural design project and successfully achieving the outcomes we are aiming for. This process can be broken down into five clear steps, which have been unpacked below.

Step 1: Start with what is known and prioritise from there

Businesses usually have KPIs, OKRs, or something similar. These are typically the outcome measures that stakeholders use to assess the success and growth of an organisation, its teams, and individuals. They are familiar and typically agreed upon within the organisation, which make them a useful entry point for bringing stakeholders onto a behavioural design project.

If there is a specific problem you and your team are working on this can be a useful point to start at too. Just the problem statement doesn’t have assumed lines of relation to key measures. The problems relationship should be direct and assumption free.

To get going, start with the following:

1. Map out the team’s primary performance metrics (KPIs, OKRs, etc). These should relate directly to the core drivers of an organisation’s success.
2. Filter out the metrics that are not directly related to shaping human behaviour (you will be surprised to see how few there are).
3. Prioritise the performance metric based on business and user needs. (Assessing existing administrative and performance data can be useful in making decisions here).
4. Agree on a specific performance metric to focus on for this project.

Useful tasks to support this step:

• Analyse administrative data to identify problem spaces
• Build hypothetical outcome scenarios to assess impact possibilities

Some examples of primary business measures for behavioural design projects include:

• Monthly contribution by existing investors (% change)
• Amount under arears of existing banking clients (% change)
• Number of new unique account set ups for the quarter
• Daily sales % of product x online
• Monthly new user subscriptions (% change)

Output: A prioritised list of the user-driven business metrics (KPIs or OKRs) or business problems

A list of the key business metrics as well as the top priority metric for this particular project. The metric should be driven by user behaviour, have large effect on business performance, and be feasible to change.

Step 2: Unpack the behaviours that drive the key metric

This is a subtle but important exercise. In order to have a real chance of achieving an outcome, it needs to be unpacked so that the user behaviours that drive its success are revealed.

Behavioural design projects are about shaping key behaviours, which, in turn, drive business outcomes. Without a concrete understanding of these behaviours, the likelihood of having an impact is significantly lower.

Identifying the behaviours:

1. Map out the different kinds of behaviours (key actions) users can take that directly drive the outcome measure of interest.
2. Describe each of these behaviours in as much detail as possible. Who are the users that perform the behaviour? Are there prerequisites that the user must meet? Where do the actions take place? Are there different channels for the same action? What are these channels?
3. Prioritise the different behaviours. Which of them have the largest potential impact on the outcome measure of interest? What assumptions are we using to assess this? Which of these actions can realistically be improved on at scale? Which of them can be measured directly or via proxies?

Useful tasks to support this step:

• Analyse and segment the data driving the key metric
• Build hypothetical outcome scenarios to assess impact possibilities

Output: A prioritised list of the user behaviours that drive the key metric:

The list should include of the different kinds of user behaviours, as well as key details that provide further context on each of them.

Step 3: Define the behavioural objective

Once the priority behaviour has been selected, it should be defined as a behavioural objective. This strategic statement will form the backbone to the behavioural design project and something to return to when consensus dissolves or confusion relating to direction sets in.

Below are the key questions to ask in ensuring that the behavioural objective is set correctly:

Is the behavioural objective defined as achieving a particular action? A trivial, yet critical aspect, of setting a good behavioural objective is that the focus should be on actionable behaviour change, as opposed to changes in awareness, knowledge, or attitude.

Is the behavioural objective free of any assumptions relating to causes or solutions? It is important to make sure that there are no assumptions of causes or solutions implied in the definition of the behavioural objective.

Is the behavioural objective a very specific action within a certain context? It is important to ensure that the behavioural objective is as precise as possible. The objective should focus on a very specific behaviour and, if possible, within a certain context (i.e. be narrowly defined).

Can the behavioural objective be easily understood by all parties involved in the process? Practitioners will need to continuously refer back to behavioural objectives throughout the project. With this mind, it is crucial that the behavioural objective is clear in order to create the confidence that all team members have the exact same understanding of what they are trying to achieve.

Are we able to accurately measure the impact that an intervention will have on the target behaviour? In order to test the effectiveness of an intervention in achieving the behavioural objective, we need to be able to accurately measure its effect on behaviour in a clear and statistically unbiased way.

Output: Behavioural objective statement

A narrowly defined statement that details the specific action that we want users to take.

Step 4: Set the primary project outcome

The behavioural objective gives the team an understanding of the specific behaviour they will be aiming to change. This allows for clear direction and consensus, which are absolutely fundamental to a good behavioural design project. The behavioural objective is also used to set the primary project outcome — the specific, measurable, and time-based goal that the team is aiming to achieve.

The project outcome needs to be:

• Behavioural (specific to a defined behavioural objective)
• Measurable (how the behaviour is tracked and quantified)
• Targeted (characteristics of the target group that can be identified)
• Assignable (to participants in a control and treatment group)
• Time-based (a specific day by which the behaviour will be achieved)

Examples of behavioural design project outcomes:

• Increase the number of active debit orders with debt playing clients, who have more than R50,000 outstanding, by 8% by the end of December 2018.
• Improve on-time monthly loan payments among students who graduated from (xxx) programme in 2018 by 15% at the end of the first quarter of 2019.
• Improve the number of existing client who contribute consistently to their savings account on a monthly basis, and who also earn more than R20, 000 a month, by 10% from the beginning of 2019 to the end of the second quarter.

Output: Primary project outcome statement

A description of the specific outcome that the team will be prioritising for this project. The outcome must link directly to the target behaviour, be measurable, assignable, and time-based.

Step 5: Mapping out the desired behaviour

Behavioural mapping is used to get a better understanding of the specific action you want users to take and how to get there.

An important consideration is to think about the user’s decision as separate from the action they need to take. Constructing this dichotomy helps to better understand where the barriers are and what kind of interventions will be necessary to bring about the change.

This approach of separating out the decisions and actions is something many behavioural design teams use around the world. Specifically, the approach has been championed by Ideas42 as a crucial step in understanding the roots of a specific behaviour and how to change it.

There are some important aspects of the decision-action model that are worth unpacking. Firstly, it accommodates the possibility that users have not made a decision about whether or not to perform a particular action (this possibility occurs far more often than you might assume). This is important, because often when we see that users have not taken a particular action (through outcome measures tracking that action), we might assume that they have ‘decided’ not to act. This could very well be the case, but it also might be that they have not made up their minds, or that they intended to perform a particular action but were not able to follow through. The latter case is called the intention-action gap, a phenomenon that is well studied within the behavioural sciences, but is also a scenario that every human is familiar with too.

Think about it. Have you ever packed your gym kit in the morning with the intention to go to gym, but then, at last minute, skipped it? How about deciding to take a one minute shower, yet ending up with a five minute one? Or thinking badly of someone who is texting and driving, but then have done it yourself?

We are all guilty of having a gap between our intentions and our actions. The practitioner’s role is not to blame the user for this. Rather, it should be to identify these gaps, understand the causes, and develop ways to close them.

Mapping behaviour with the decision-action model:

1. Consider the key behaviour that you want users to take. Break that behaviour up into a decision (should I do x?) and a final action (do x).
2. List out all the factors that could influence the decision.
3. Use backwards induction to map out all the action points that connect with the final action that you are interested in measuring. The more granular, the better here. If there are multiple pathways, map these out too and categorise them.

Output: Behavioural map of decision factors and action pathways

A list of factors that stakeholders assume are likely to affect the user in their decision to perform a particular action, as well as the different action paths leading to the final action stakeholders want users to take.

Moving forward:

Getting down the assumed factors of influence and mapping out all the action points will provide a clear picture of the assumed journey that users ought to follow. This structure will be used to guide your approach as you move into the diagnosis phase. It can act as a map from which you can identify and track the barriers and drivers that are veer users off track, as well as help you identify the potential root causes of these behaviours.