The Devil's in the Detail: Design Considerations for Wise Interventions
The final article in our five-part behavioural science crash course, here we detail the principle behind wise interventions. These are the pillars which support nudges for the public good. Read on to learn how to effectively design your nudges for success and the practical considerations to keep in mind to avoid failure.
In the last post we took a deep dive into Semantic Prosody and mindless choosing. We’ve looked at how particular words, with the same meanings, can negatively or positively balance the tone of a sentence. And in reviewing the photocopier study, we’ve identified the importance of using reasons when trying to guide behaviour. Each study has demonstrated the powerful effects of our dual process system of thought, as well as provided some useful tips that you can enact within your organisation. Better still, you can start putting these principles into practice within your decision-making processes or personal life, right now!
The next step is to consider a ‘nudge’ more holistically – as its own project. If you were designing and implementing a targeted nudge for a process within your organisation, a customer journey or for some sort of social good in the public sector, how would you go about it? Let’s take a look at a real-life example of a carefully planned Wise Intervention – a nudge aimed at promoting social good!
These are described by Walton as brief, ordinary, and precise nudges, targeted at promoting some social good or benefit1. In this section, we’ll explore what is required for a nudge like this to be set up for success.
Bryan, Walton, Rogers and Dweck examined the effects of ‘invoking the self’ to positively impact voter registration and voter turnout within two state-wide elections in the USA. In three randomised, controlled experiments, the researchers showed that using limited and subtle linguistic cues the day prior to the event, they were able to produce a significant increase in voter turnout (11%)!
That’s an enormous increase in turnout, but how did the experimenters cause it? In Bryan et al’s study, all it took was a simple change to the grammatical structure of a pre-election survey the day before voting took place! The intervention condition of Bryan et al’s study was to frame survey items around nouns i.e. “How important is it to you to be a voter in tomorrow’s election?” Whereas in the control condition, survey items were framed using verbs i.e. “How important is it to you to vote?”
The predicted psychological process in action, was that using nouns would cause positive framing effects – noun wording would present an opportunity to become a valued kind of person, as opposed to verb wording which presents an ‘errand to be accomplished’. The results were significant, and turnout was substantially increased.
As surprising as these results seem, Walton identifies many different examples of long term and significant changes, resulting from sometimes very short and subtle interventions. These interventions ranged in scope from workshops on neural plasticity encouraging a growth mindset, which lead to long term improvement in junior maths grades, to random clean ups of crime hot spots, leading to significantly less 911 calls and arrests over the following 6 months! In each of these studies, and the many others you can find detailed in Walton’s excellent article, you can see a proven psychological process in action . Each demonstrating that subtle nudges can have profound and long-term effects.
The key points to remember when designing your own nudges are to carefully specify your measures and plan out your solution. The intervention condition must have a powerful sample, be tested against a control, be backed by a solid psychological theory or recursive process, and assessed using specific, reliable measures across a predetermined timeframe. We’ll go into some of these points in more detail…
Importance of a supporting theory
The study about improving maths grades was based on the theory that the growth mindset supports academic improvement. A growth mindset being the belief that one is able to improve one's own abilities through hard work. The second study was based on a well proven social theory known as ‘broken window theory’, which suggests that social cohesion and crime rates are lower in areas which do not present as disordered. Each of the studies Walton et al highlights in their article are based on well-developed psychological theory, and when designing your own nudges this is particularly important to keep in mind. Develop an evidence based, working hypothesis, and design your solutions accordingly.
So how do you go about designing nudges when you don’t have an existing pool of evidence to base your ideas on? Well there’s a few solutions to this problem. The first option, if available, is to do some analysis on what has worked well before internally. Examine the interactions you’ve had with customers. Try to understand what specific elements of your customer journey are most effective.
You can try to replicate this in an a/b test alongside a control as part of a future nudging project, and you can add this learning to your own evidence based nudging initiative.
Using what you’ve learned so far
You can also draw from existing social-psychological theory. We’ve already discussed semantic prosody, cognitive load, the photocopier study and several other psychological theories, each supporting potential nudges you can enact within your organisation. We take a mixture of these approaches, drawing on our learnings from successful journeys we have delivered, as well as cutting edge psychological theory.
Clearly, developing your nudges will require some precision, but you’ve already got most of the tools to start! Psychological precision means beginning with a well-established and specific theory – you could start with dual process theory, and to be specific - that using reasons alongside small favours should improve compliance rates via type 1 bias. This is a strong theory with a simple, easy to test relationship.
As relationships get more complex, so too does the level of expertise required to properly maintain psychological precision in your interventions. You’ll also need a plan to measure the effectiveness of your nudge. I suggest you try to keep reporting considerations in mind early on when you’re in the design phase. These will be enormously helpful later on.
Building an evidence-based approach
At ContactEngine, we’re constantly chasing improvement, not only as an artificial intelligence business, but also in terms of our linguistic expertise and journey design capabilities. To accomplish this, we try to follow these rigorous requirements: we meticulously design a/b tests according to sound psychological principles, and carefully measure the effectiveness of our solutions, before incorporating this knowledge into our staple journey design methodology. Each solution is backed by a proven and growing body of not only academic theory, but practical evidence and experience gathered through real life user testing at scale. Doing so helps us achieve agile and iterative improvements across the entirety of our solution designs.
We’ve briefly mentioned reporting considerations, but these must become an integral part of the design when implementing nudges. How will you measure success? What factors could this nudge affect and are you set up to report on these results accurately, in a timely fashion? There is a difference between the level of scientific rigour necessary in the academic and the business worlds, but a nudge must be set up to be measured precisely prior to going live.
Trying to keep the variance between the nudging journey and control as minimal as possible is also good practice. As well as determining when the experiment should be stopped and evaluated. Doing these things will ensure you are in an excellent position to work out the effectiveness of your nudge, and not keep it going longer than it needs to if it proves ineffective.
Finally, one last reporting consideration – keep in mind the null hypothesis, that is, what needs to happen to show that your nudge was not successful or there was no significant difference between the intervention and control. In most cases the answer is going to be – no significant change. This information is crucial to shaping future solutions - finding out what doesn’t work is just as important as finding out what does!
Long-term changes through recursive processes
If a nudge is to effect long term change, it’s important that the intervention targets recursive processes. Walton has demonstrated that wise interventions may have effects which persist for years, but this is only when the targeted psychological processes are of the kind that will compound over time – such as the study mentioned earlier based on the growth mindset.
Outside of academia and public policy, the benefits of customers undergoing long term change may seem sparse. But in the case of longer-term customer journeys, such as those within utilities and telecoms, or journeys aimed at existing business relationships, partnerships or customer retention, employing effective nudges focused on long term influence and change, may have huge business benefits.
The last thing to remember about wise interventions is that they’re context dependent. Even a carefully designed nudge, with sound psychological backing and a solid implementation plan, may not work across different segments, customer journeys or scenarios. This is because every intervention is dependent on the context within which they are applied. This is why getting a thorough testing plan together is essential to every implementation.
We know that every organisation we work with has a different cohort of customers, with different motivations, expectations, prior experiences, and needs. Part of why we rigorously test our designs, is to try to understand this context which varies across clients, so we can adapt accordingly. Within your organisation, you may find that what works at one stage in the customer journey, fails later on – and this can be for similar reasons – the context of the intervention has changed over time! When you’re designing nudges for business, test, retest and adapt accordingly.
Congratulations! If you made it this far and you’ve read our other articles, then you’ve finished our crash course in nudging and behavioural science. The goal here was to equip you with the principles behind setting up effective nudging solutions, and some of the more important business focussed elements to that. But we’ve covered a lot more across the series.
At this point we have covered an awful lot of ground! You should be comfortable with the dual process theory of mind, system 1 and system 2 thinking and error, understand designing choice structures with cognitive load in mind, and have an awareness of your own susceptibility to confirmation bias and poor mental accounting.
You should have a practical knowledge of semantic prosody and its implications and know how to use reasons in conversations to elicit type 1 responses and how to be prepared to offer more detailed reasoning when required. Finally, you should now understand what makes a nudge a wise intervention and have some understanding of how to implement them into your professional or personal life.
For more information about how and why we do what we do at ContactEngine, please don’t hesitate to get in touch with us. Good luck nudging!
Find the other articles in this series in our CX Insights Hub
 Walton GM. The New Science of Wise Psychological Interventions. Current Directions in Psychological Science. 2014;23(1):73-82. DOI: 10.1177/0963721413512856
 Christopher J. Bryan, Gregory M. Walton, Todd Rogers, Carol S. Dweck Proceedings of the National Academy of Sciences Aug 2011, 108 (31) 12653-12656; DOI: 10.1073/pnas.1103343108