Understanding behavioural design: barriers and enablers

Abstract

Behavioural design has emerged as an important domain of design research and practice. However, there is a need to better distinguish behavioural design and subsequently map its unique features and challenges. In answer to this need, this work examines the evolving role of behaviour in design and contrasts this with eleven in-depth behavioural design cases. This has resulted in our identification of behavioural design as a new paradigm of design, with a number of unique characteristics. Based on this, we propose a model of behavioural design. Furthermore, we identify three critical barriers to behavioural design in practice, and suggest ten mitigating enablers. Together these findings provide implications for further development of theory, practice, and education in behavioural design.

KEYWORDS:

• Behavioural design
• design theory
• user behaviour
• user-centred design
• design process

1. Introduction

Behavioural design, which ‘actively’ aims to change problematic behaviours, has emerged as a new and important domain of design research and practice (Cash, Hartlev, and Durazo 2017; French et al. 2012). Effective design for positive behaviour forms a critical component in addressing major societal challenges (Michie, Atkins, and West 2014; Tromp, Hekkert, and Peter-Paul Verbeek 2011), including health (Kelders et al. 2012; Taylor, Conner, and Lawton 2012) and sustainability (Abrahamse et al. 2005; Bhamra, Lilley, and Tang 2011). Here ‘positive’ is considered to be beneficial to the individual, society, and environment, balancing individual, collective, and ecological concerns (Tromp, Hekkert, and Peter-Paul Verbeek 2011; Bhamra, Lilley, and Tang 2011). Behavioural design achieves these effects by ethically evoking, nudging, persuading, or motivating people to behave in ways that are desirable for both the individual and society (Tromp, Hekkert, and Peter-Paul Verbeek 2011). However, major questions remain in how best to understand and support behavioural design.

Current research on behavioural design, and related works in persuasive design (Fogg 2009; Wendel 2013) and design for social (Tromp 2014) and sustainable behaviour (Bhamra, Lilley, and Tang 2011), have delivered a number of major insights. These include an array of design process models (Cash, Hartlev, and Durazo 2017), lists of behavioural interventions (Lockton, Harrison, and Stanton 2010; Michie et al. 2013) and strategies (Dorrestijn 2017; Daae and Boks 2018; Cash et al. 2020), ethical guidelines (Berdichevsky and Neuenschwander 1999), and recommendations for novel design practices (Niedderer, Clune, and Ludden 2018b). Interventions describe campaigns, artefacts, systems, or other means used to shape behaviour change (Tromp, Hekkert, and Peter-Paul Verbeek 2011; Francis et al. 2009). However, this profusion of research across disparate bodies of literature has led to the emergence of two major research needs in behavioural design. First, behavioural design is conceptualised as the ‘active’ (Tromp, Hekkert, and Peter-Paul Verbeek 2011) or ‘overt’ role of design in shaping behaviour (Niedderer, Clune, and Ludden 2018a, 5). This suggests a need to understand how ‘behavioural design’ is distinct from prior views of behaviour in design (Khadilkar and Cash 2020). These include the technical behaviour of a product (Gero and Kannengiesser 2004), user behaviour in support of technical function (Sun et al. 2013; Hubka and Ernst Eder 2012), infrastructure effects on social behaviour (Winner 1980), unintentional product-driven changes in behaviour (Verbeek 2008), or unsustainable purchasing behaviours associated with products designed for obsolescence (Papanek 1984). Thus, there is a need to distinguish behavioural design and subsequently map its unique features and challenges. Second, prior empirical studies have typically focused on specific challenge areas, such as sustainability (Lilley 2009) and health (Michie and Abraham 2004), or technical contexts, such as human computer interaction (Hekler et al. 2013) and product design (Bhamra, Lilley, and Tang 2011). As such, while engineering and behavioural design perspectives are fundamentally connected in the blending of human and technical solutions to societal challenges (Kouprie and Visser 2009; Hoyle et al. 2011), there is little research on the general challenges facing practitioners in this context. Thus, there is a need to evaluate what aspects of behavioural design are general verses context-specific. Together, these gaps hamper efforts to develop effective design support (Daalhuizen 2014) and stymie theory building in this domain (Cash 2018).

In this work, we address these two needs using a two-part approach. First, we examine how the role of behaviour in design has developed over time in order to provide a framework for differentiating behavioural design from other engineering-focused design approaches. Second, we report on eleven in-depth cases drawn from across behavioural design contexts in order to propose a refined definition and domain for behavioural design. Furthermore, we draw these findings together in a conceptual model of behavioural design. Finally, we identify critical enablers and barriers of behavioural design work in practice. As such, we develop a number of contributions to behavioural design theory and practice.

2. The evolving role of behaviour in design

Understanding how the role of behaviour has evolved in design is crucial to understand how it affects current design practice. This requires a framework to first define behaviour, and then to understand how design has addressed it. In this section, we first present a research framework for understanding the role of behaviour in design and then trace its evolution through different design paradigms.

2.1. Structuring understanding of behaviour in design

The design literature describes a number of technical theories for understanding the subject of engineering design (Vajna et al. 2005; Cross 2018); however, similar theories of human behaviour in design are missing. Therefore, we adopt a widely used theory of behaviour from psychology in order to structure our understanding of behaviour in design. Specifically, we build on the four components defined by Fishbein and Ajzen (2010): ‘the action performed, the target at which the action is directed, the context in which it is performed, and the time at which it is performed’. This definition is drawn from the Theory of Planned Behaviour (Fishbein and Ajzen 2010, 18), which is widely operationalised in research on design-driven behaviour change (Michie et al. 2014). As such, this forms a robust basis for evaluating aspects of behaviour in design following a similar approach by Michie et al. (2014). The major components of action, target, and context/time are defined with examples in Table 1. In addition, it is crucial to understand how designers deal with user behaviour during the design process. Here, we focus on the design goal and object (Cross 2000; Roozenburg and Eekels 1995; Ulrich and Eppinger 2000), as well as how behavioural theories are used during the process and how user psychology is utilised in achieving the goal (Michie et al. 2014). These are again defined in Table 1, together with examples and sources.

Table 1. Research framework with seven components of behaviour in design.

	Component	Definition	Example
Related to user behaviour	Action	The specific behaviour performed (Fishbein and Ajzen 2010)	Derived from the user's role in the phases of product lifecycle – Accept/purchase, use/maintain, or discard/recycle/reuse (e.g. Ulrich and Eppinger 2000; Hubka and Ernst Eder 2012)
	Context/time	The time/place in which the action is performed (Fishbein and Ajzen 2010)	Industrial, public, and private (Otto and Wood 2001; Niedderer et al. 2016; Morelli 2007)
	Target	The target at which the action is directed (Fishbein and Ajzen 2010)	Product, process, service, interface or system (Ceschin and Gaziulusoy 2016)
Related to design process	Goal	The purpose of the design activity (Blessing and Chakrabarti 2009; Roozenburg and Eekels 1995)	To achieve Industrial efficiency (Hubka and Ernst Eder 2012) or reduce ecological impact (Lilley 2009)
	Object	The thing that the designer develops to fulfil the goal (Cross 2006, 15)	Products (Roozenburg and Eekels 1995), Product-Service Systems (Morelli 2002), Systems (da Costa Junior, Diehl, and Snelders 2019)
	Use of behavioural theories	The use of scientific insights in dealing with goals and objects (Michie et al. 2014)	Not used, implicitly used, or explicitly used (French et al. 2012)
	Use of psychology	The utilisation of user’s psychology in achieving the goal (Michie et al. 2014)	Used as a static reference, used as a dynamic means (e.g. refer axis of influence in Lilley et al. 2018, 45)

Based on this framework we are able to distinguish both how designers understand behaviour and how it has been treated as part of the design approach. The components in Table 1 allow us to identify seven paradigms of design, which we describe in the following section in a quasi-chronological order following similar analyses of other design domains (see: Ceschin and Gaziulusoy 2016 in design for sustainability; Bijl-Brouwer and Kees 2017 in human-centred design). The seven paradigms are not mutually exclusive and are chronologically overlapping. Essentially, the paradigms share features related to e.g. design process, yet are distinct with respect to the critical elements related to the role of behaviour (Table 1). For example, a design approach aiming to deliver an engineering object, might formulate the goal as ‘democratic involvement of stakeholders in the design process’, and as a result start using participatory design methods and knowledge. This would be distinct from an engineering design paradigm, because the considered actions include ‘acceptance’ in addition to ‘use’, and the considered context are ‘social’ rather than purely ‘industrial’ (e.g. see the differentiation between ‘things’ and ‘Things’ in Bjögvinsson, Ehn, and Hillgren (2012)). Thus, the purpose is not to delineate the history of design, but to categorise conceptually based on the role of behaviour in design. The seven paradigms are summarised in Figure 1.

Figure 1. The seven paradigms of behaviour in design.

2.2. Understanding the role of behaviour in seven paradigms of design

The first paradigm reflects the initial application of design in modern industrial production to increase efficiency. The main goal of this ‘engineering’ design was to solve a technical problem ‘within the requirements and constraints set by material, technological, economic, legal, environmental and human-related considerations’ (Pahl et al. 2007, 1). The objects developed were machines to be manufactured by non-artisans and operated by non-specialist workers (Jones 1992, 20–24). Most methods, processes, and tools during this paradigm focus on machine development and operation (Pahl et al. 2007; Hubka and Ernst Eder 2012; Cross 2000; Otto and Wood 2001; Lockledge and Salustri 1999; Brace and Cheutet 2012). Thus, the predominant action in this paradigm was use/maintain because the context/time was predominantly Industrial production and the targets were the products, and mechanical interfaces on machines and assembly lines. Behavioural theories was not widely used, nor was user psychology explicitly utilised in achieving the goal.

The second paradigm of design highlights the importance of understanding human limits, both physical and psychological, during the design of technical systems. The goal of this ‘human factors and ergonomics’ paradigm was to design ‘productive, safe, comfortable, and effective human use’ of objects like ‘tools, machines, systems, tasks, jobs, and environments’ by the discovery and application of ‘information about human behaviour, abilities, limitations and other characteristics’ (Sanders and McCormick 1993, 5). Research on accidents during World War II highlighted how technologically advanced, complex, powerful, and faster machines surpassed operators’ abilities (Wickens 1992). This necessitated designing the technologies, and their interfaces, to accommodate human’s physical and cognitive limits (Norman 2013, refer Chapter 5). With the advent of computers, this ethos was extended to human computer interaction (Grudin 2005). This paradigm still focused on the action of ‘use/maintain’ as the humans are still primarily considered as operators. The context/time was predominantly ‘Industrial’; however, the application later continued in personal and public contexts. The targets were typically products, such as computers, industrial machines, tools, programmes, and interfaces. Finally, while this paradigm used the behavioural theories both ‘explicitly and implicitly’, user psychology was treated as a static reference while designing.

The third paradigm emphasised the ‘human-centred’ nature of design. In the second paradigm, the designer acts as an expert and ensures that the machines fit humans’ abilities. In this paradigm, the stakeholders actively participate as designers along with professionals. The goal of this participatory and co-design paradigm was treating the weaker or invisible stakeholders as equals, taking decisions in the situations of actual action, respecting multiple visions about technology, and being democratic in taking decisions (Luck 2018). The objects of design were the interventions, such as products, processes, and work-environments that combined social and material factors (Bjögvinsson, Ehn, and Hillgren 2012; Luck 2018). This paradigm continued to focus on the action of ‘accept and use/maintain’ with added emphasis on human agency. The context/time was mainly industrial and social with technology, products, tools, and interfaces as the target. This paradigm explicitly does not build on behavioural theories, instead focusing on co-creation and user involvement (Bjögvinsson, Ehn, and Hillgren 2012; Ehn 2008). Similarly, utilisation of generic user psychological was typically downplayed, in favour of a focus on the situated actions of stakeholders through methods that allow participation of common users, known as ‘infrastructuring’ (Bjögvinsson, Ehn, and Hillgren 2012; Ehn 2008).

The fourth paradigm focused on the personal consumption of products, where user’s affective experiences were as important as technical functions. The prominent design approaches in this paradigm were design for emotion and experience design, e.g. the Philips’ Wake-up Light that simulates the sunrise and bird chatter, in order to wake up with positive emotions and experience (Hassenzahl 2013). The goal of this paradigm was to provide competitive advantage to companies through affective experiences (Desmet and Hekkert 2007) ‘with a focus placed on the quality and enjoyment of the total experience’ (Norman 2013, 5). The objects of design were mainly ‘products, processes, services, events, and environments’ (Norman 2013, 5). The considered actions were both acceptance/purchase and use/maintain. The context/time was mainly private, expanding to experiences in public and industrial contexts. The targets were the aesthetic and functional interfaces, products, and systems. Finally, while the use of behavioural theories was typically not explicit, user psychology was widely used as a static reference against which to design affective objects.

The fifth paradigm dealt with responsibility and end-of-life considerations. The goal of this paradigm was to design sustainably for responsible consumption, production, and end-of-life of products (Bhamra and Lofthouse 2007, chap. 4). The objects of this sustainability paradigm expanded from isolated products, services to interconnected systems that linked technology and people (Ceschin and Gaziulusoy 2016). The objects of the majority of the design for sustainability approaches, such as cradle-to-cradle design, product-service system, and systemic design, are artefacts with reduced ecological impact (Ceschin and Gaziulusoy 2016). In addition, approaches, such as design for sustainable behaviour, and systems innovation and transitions, achieve the goal by changing unsustainable behaviours (Ceschin and Gaziulusoy 2016; Manzini 2014; Norman and Stappers 2016). However, the projected objects of design in these cases were also artefacts, e.g. ‘Design for sustainable behaviour: Using products to change consumer behaviour’ (Bhamra, Lilley, and Tang 2011) and ‘Design interventions for sustainable behaviour’ (Lilley et al. 2018). This paradigm started with the action of discard/recycle but expanded to include acceptance/purchase, use/maintain, and reuse, in industrial, private, and public context/time. The targets are products, lifestyles, systems, and societies. The approaches that focused on reducing the lifecycle impact of artefacts did not typically use behavioural theories or user psychology. However, design for sustainable behaviour explicitly uses behavioural theories, as well as making use of user psychology as a means to design the desired behaviours.

The sixth paradigm emphasised synergy between individuals and social context to define and achieve social goals. The goal of this paradigm was the application of "design thinking and design methodologies to social issues to create innovative solutions" (Tromp, Hekkert, and Peter-Paul Verbeek 2011) achieved through the "process of change emerging from the creative re-combination of existing assets (from social capital to historical heritage, from traditional craftsmanship to accessible advanced technology)" (Manzini 2014). Here, conflicts between individual and collective concerns are addressed by attempting to synthesise mutually beneficial goals that shape positive change in an efficient and sustainable way (Margolin and Margolin 2002; Tromp, Hekkert, and Peter-Paul Verbeek 2011). Common design objects were systems of change combining products, processes, and social movements. While Tromp, Hekkert, and Peter-Paul Verbeek (2011) highlight pro-social behaviours as a critical goal, they retain a focus on artefacts as objects. This paradigm focused on actions related to accept, use/maintain, and recycle/reuse as well as targets such as social movements, infrastructure, lifestyles, policies etc. The context/time was predominantly social but indirectly affected choices in personal contexts. Current approaches to social design do not explicitly build on behavioural theories, but do use key aspects of user psychology, such as motivation, as a means to achieving design goals.

Finally, the seventh paradigm focuses on communication, and has roots older than the first paradigm due to its origin in the arts. The main approaches of this paradigm were graphic design and communication design. A specific instance of communication design: advertising, focused on persuading action (Richards and Curran 2002). Furthermore, the emerging ubiquity of computers also links aspects of persuasive human computer interaction to this paradigm (Fogg 2002). Cutting across these perspectives, the overall goal of this paradigm was to codify and direct communication, deliver information, persuade action, or manage perception (Barnard 2005, 13–17). The objects used to do this were ‘diagrams, sketches, charts, photographs, video, and animation’ (Agrawala, Li, and Berthouzoz 2011, 60). The actions in this paradigm are accept/purchase, use/maintain, and discard/recycle/reuse, in industrial, public, and private context/time. The targets were typically specific products, systems, and interfaces. While the use of behavioural theories is not widespread in non-persuasive communication design, it is ubiquitous and explicit when persuasion is the goal. Furthermore, user psychology is widely utilised as a means to achieve design goals, via concepts such as Gestalt Theory (Behrens 1998).

Bringing these paradigms together it is possible to develop a timeline of the evolving role of behaviour in design, as illustrated in Figure 1. Here the seven major paradigms are shown (denoted as P1, P2, etc.), as well as more specific approaches within them. The arrows also show the interrelation between them, e.g. Paradigm 5 is the response to industrial (originating in Paradigm 1) as well as personal (Paradigm 4) unsustainable practices. It is notable that while the goals of each paradigm change in response to the behavioural challenges of the time, the focus on artefact objects remains more consistent. Engineering design has also evolved to these changing roles of behaviour in design, by giving special consideration to the human actions in using a technical product (Sun et al. 2013; Sun et al. 2019), adopting to participatory processes for design of new technologies (Bjögvinsson, Ehn, and Hillgren 2012), to focus on methods for selecting using environment-friendly materials in sustainable design (Eddy et al. 2015), and by understanding how personalisation of product form affects emotional bonding (Mugge, Schoormans, and Schifferstein 2009). This underlines the importance of understanding the role of behaviour in engineering design. Although this paradigm does not explicitly target a behavioural object, use behavioural theories, or change user psychology, it still deals with human behaviours that are necessary for technical function of an artefact object (this has also been referred to as behaviour or behavioural design (Sun et al. 2013; Vermaas and Dorst 2007)). Furthermore, a number of paradigms make use of user psychology either as a reference or as a means to achieving a goal, but few currently make extensive, explicit use of behavioural theories in directing both the design process and object development. While these paradigms all involve user psychology, they differ in their goals. For example, Paradigm 4 uses experience or emotions to improve the sales of products, while Paradigm 5 alters behaviour for sustainable use, or even stops purchase of unnecessary products. Similarly, Paradigm 4 is anthropocentric in nature in contrast to the socio-technical nature of Paradigm 5. We leave behavioural design blank in Figure 1 because it is the aim of this work to determine if this does, in fact, denote a distinct new paradigm or not.

3. Methodology

In order to address the research needs outlined in the introduction we adopt a theory building approach (Cash 2018; Eisenhardt and Graebner 2007), building on a qualitative, multi-case research design (Robson and McCartan 2011; Handfield and Melnyk 1998). Specifically, there is a need to distinguish behavioural design and understand its unique features across contexts (Fogg 2002; Lilley 2009; Tromp, Hekkert, and Peter-Paul Verbeek 2011). Thus, we formulate two research questions:

RQ1: How does behavioural design differ from prior design paradigms with respect to the role of behaviour?

RQ2: What are critical enablers and barriers of behavioural design in practice?

In order to answer these questions, we evaluate design cases in practice through the perspectives of expert behavioural designers (Yin 2014, 4). In doing so we strive for theoretical generalisability (Robson and McCartan 2011) by contrasting cases that exemplify different perspectives on behavioural design and its application across contexts (Yin 2014).

3.1. Case selection

Cases were framed at the project level and selected based on three main criteria. First, all cases dealt with the explicit design of a new user behaviour (Tromp, Hekkert, and Peter-Paul Verbeek 2011; Niedderer, Clune, and Ludden 2018a). Second, cases included all design stages from overall aim to testing as elaborated by Cash, Hartlev, and Durazo (2017). Third, the majority of the critical decision makers, such as project managers, designers, and behavioural analysts etc. were accessible for interviews. The authors contacted each case company and presented a brief about the research project. After clarifying doubts, the case companies communicated their willingness to collaborate, and presented a list of projects that enabled identification of specific cases using the three-selection criterion listed earlier. This resulted in the identification of eleven case projects from four of the largest and most successful design consultancies dealing with behavioural design in Denmark. This allowed for comparison and replication both within and across companies (Yin 2014, 56–61).

The cases were selected from a list of cases suggested by the case companies using the seven-component framework as a reference. In all cases, the design teams identified the project goal about changing a specific behaviour. Ten out of eleven cases had evident problematic behaviour or target behaviour, where the behavioural designer developed interventions for them. Furthermore, the cases covered a range of objects including individual products, as well as systems of products, series, and other interventions. The cases covered mainly acceptance and usage-related actions. Four cases dealt with social context, three with industrial, two with personal, and three cases had multiple contexts. As such, we conclude that the sampled cases reflect a range of behavioural design in practice and thus provide a basis for further analysis.

3.2. Data collection

A semi-structured interview guide was formulated based on the research questions and research framework (Section 2.1). Questions targeted the seven components outlined in Table 1 as well as the wider context of the design work in the company, as summarised in Table 2. These contextual questions were added in order to understand the enablers and barriers to behavioural design in practice.

Table 2. Sample interview questions.

Component of research framework	Example questions
Action	Can you talk me through the behaviour you worked on in this project? What were its characteristics and range of behaviour?
Context/Time	How general was the behaviour and demographic? What were the causes of this behaviour? Why do you think users behave in this way?
Target	How did the intervention get the person from current behaviour to target behaviour? What other interventions did you consider during development?
Goal	How did these considerations [related to behaviours] inform your project formulation? Did you make a brief, if so what did it include?
Object	How did these considerations [related to behaviours] inform your project formulation?How did you arrive at target behaviour? How did you evaluate the success of the project?
Use of behavioural theories	Can you talk me through any behavioural theories you use during the design process?
Use of psychology	How did the intervention get the person from current to target behaviour? Why do you think users behave in this way?
Wider context	Example questions
Company context	Did you make a brief, if so what did it include? Do you differentiate projects with respect to the target behaviour? If so how?
Process	Can you sketch a project timeline? and talk me through the process from idea to final solution?
Project Management	How did you set up the project and team organisation? How did you keep track of the project?

The first author conducted twenty-four face-to-face interviews with the key decision makers in the eleven cases. The interviews lasted for approximately one hour. Multiple interviews were carried out for each case until saturation was achieved (Guest, Bunce, and Johnson 2006) i.e. no new information was revealed by further questioning. Importantly, in seven out of eleven cases small teams of two or three designers/analysts carried out the project. In two cases, all members were interviewed, and in four cases, two out of the three members were interviewed. The non-interviewed person was only involved in initiating the project with the client. Hence, in most cases we were able to interview all project members. Additional documents such as internal reports, project briefs, presentation decks, final project reports, blogs etc. were also collected.

3.3. Data analysis

All interviews were recorded and transcribed. The transcribed and coded interviews were then used as the basis for thematic analysis. Open coding was used to understand the role of behaviour in the design process, the seven-component theoretical framework was used flexibly to move back and forth between abstract concepts and specific details (Neuman 2011, 511). Axial coding allowed for the identification of themes in the behavioural formulation and its broad process and practice implications through different project members’ perspective (Neuman 2011, 512–513). Three final themes were developed via a number of iterations between the authors. This process helped to revisit the data to substantiate the themes. Finally, analysis was presented back to each company in order to gain feedback and confirmation regarding the results as well as consent for publication.

4. Findings

The thematic analysis revealed three main themes, which we examine below. The first related to the unique role of behaviour (RQ1) while the second and third related to critical enablers and barriers (RQ2), respectively.

4.1. Designing behaviour

This theme reveals two unique features of behavioural design. First, the goal of behavioural design is the explicit, ethical design of positive behaviour desired by both the individual and society. To this end, behaviours in themselves are the primary objects and the artefacts and interventions are secondary objects, and treated later in the design process. Second, behavioural design explicitly uses behavioural theories as a basis for both design process and method considerations, as well as object development. Finally, user psychology is explicitly considered as a key means of achieving the design goals.

In terms of the goal, three main features stand out from the cases. First, the goal was typically framed in terms of a problematic and/or target behaviour. This shaped the tone and language used in each project, with success described in terms of behavioural outcomes. Project Manager and Lead Designer in Case 6 said: ‘we set out to find out from the beginning what are the target behaviours’ while the Lead Designer in Case 10 commented: ‘To begin with, the target was very clear? we needed them to use this system instead of the other system’. As such, the purpose of artefacts or interventions in these projects was to enable the target behaviour. Thus, artefact and intervention development typically came later in the process as explained by Lead Designer in Case 11: ‘everything that we do for designing interventions has to be with an understanding of, how this one person is functioning? what drives this person's behaviour? what's this person's logic?’

Second, the objects typically linked to multiple different actions. This is illustrated by the Lead Designer in Case 1: ‘the behaviour we had to change was the most important one … because there are many behaviours that we bump into along the way’ and the Project Manager and Designer in Case 5: ‘for some, they would [act] but it would not be [the desired action, and] for others they did not really want to [act in the first place], and I think that was the two main behaviours’. Similarly, the behavioural analysis documents in Cases 1, 2, 6, and 7 each show a list of different actions and possible psychological constructs that were addressed by the different objects.

Third, even when the client initially framed cases as technical problems, the behavioural design teams invariably transformed this into target behaviour during the project. This happened in five projects as illustrated by the Lead Designer in Case 11: ‘[The project] started with saying, we need a new [technical support] system, … . now it has changed … ., I think we have 38 behaviours defined’. Similarly, the Project Manager in Case 1 stated that ‘[The project] was not initiated from a behavioural perspective’, yet the final output was a behavioural intervention.

In terms of use of behavioural theories and user psychology, except Cases 3, 4 and 5, designers explicitly used behavioural theories in both informing the design methods used and the development of the final design objects. Furthermore, all cases explicitly used user psychology as a means to achieving the design goals. The Lead Designer in Case 6 highlighted this process: ‘we constructed an [intervention] that utilized a mental model of [a psychological construct]’. Similarly, insights from behavioural theories were used in a number of cases to reframe a desired behaviour from a conscious action to an unconscious habitual action. The Project Manager and Lead Designer in Case 7 elaborates this: ‘[if] you have to do it every day, what's the best way to do it? Try to attach it to an existing habit’. Notably, all of the project teams included members with either formal degrees in psychology or training in behavioural theories, and experience in human-centred design.

Bringing these results together, the reframing of goals in behavioural design opens a new perspective on the solution space. Specifically, multiple objects, such as behaviours, artefacts, technical, and social systems to actively shape user psychology, are designed in unison to achieve a single behavioural goal. This is in contrast with prior paradigms of design where a specific artefact or system has typically formed the objects of design. Behavioural designers are thus ‘freed’ from preconceptions regarding the means by which a desired behaviour will be achieved. This is similar to the conceptualisation of health interventions, where multiple interventions of various types can be deployed in order to address a single health outcome (Michie et al. 2014). Furthermore, the terminology and means used to discuss these problems and solutions are fundamentally based in user psychology, which is explicitly understood through scientific insights and models drawn from behavioural theories. Thus, ‘scientific’ understanding and evaluation of behavioural outcomes is central to behavioural design, similar to clinical usages in health (Michie et al. 2008). This feature makes behavioural design distinct from prior approaches which have focused on ergonomics (Moray 1995), user needs (Norman 2013), emotion (Desmet 2008), or experience (Hassenzahl 2013). Together, these features distinguish behavioural design from the prior paradigms identified in Figure 1.

4.2. Behavioural complexity

In addition to the features identified in Section 4.1, Behavioural design also has an unusually high degree of complexity. This is due to the behavioural goal and the typical need to coordinate the design of multiple interventions of various types, often combining artefacts, technical systems, and other means e.g. educational campaigns or social movements. This combines two main features: (i) the complexity associated with behaviour in context and (ii) the technical complexity of coordinating an array of interventions, potentially over a long period.

First, behaviour in context is inherently complex and typically resists simplifying assumptions, such as rationale decision making. The Project Manager and Lead Designer in Case 6 explained: ‘[the behaviour is] very broad and very complex and very abstract for most people’, while the Project Lead and Designer in Case 5 also highlighted: ‘the problematic behaviour could have many facets’. Designers deal with multiple of such behaviours as noted in Section 4.1, where the Lead Designer in Case 11 stated: ‘we have 38 behaviours defined’. This is further compounded by the interaction between behaviour and the temporal, technical, and social context. The Lead Designer in Case 7 highlighted the many factors they need to consider: ‘there are seasonal fluctuations to [the environment under consideration], so we know it tends to be worse when it's really cold, we are less prone to [behave in the desired way]’. Similarly, Lead Designer in Case 9 stated: ‘in the morning it's very difficult to change people's behaviour’.

Second, due to the typical utilisation of multiple linked interventions there is a high degree of technical complexity. The Lead Designer in Case 11 stated: ‘we found that there are too many things in this [project] … we need many interventions’. Furthermore, Lead Analyst in Case 6 stated: ‘In terms of actually achieving [desired behaviour] there are so many thing you need to do’.

Bringing these results together, behavioural design combines behavioural and technical complexity. This makes it distinct from prior paradigms that have typically focused on technical complexity in e.g. automobile design and development (Novak and Eppinger 2001). Here, the results align with insights from behavioural theories, where behaviours are dependent on multiple factors that are context, time, and person specific (Fishbein and Ajzen 2010, 20–27). Furthermore, the challenge facing behavioural designers in resolving this complexity is illustrated by the shear variety of explanations available regarding behaviour. For example, Michie et al. (2014) identify 83 theories with 1695 constructs related to health behaviour alone. Generally, behavioural design involves multiple problematic and target behaviours (Michie, Atkins, and West 2014). Furthermore, as the examined projects typically deal with multiple actions linked to multiple interventions, the connected behavioural/technical complexity is substantial. This, in turn, adds to the procedural complexity of behavioural design, with each artefact and intervention requiring an own design cycle coordinated with all the others (Khadilkar and Cash 2019). Eight out of the eleven cases developed multiple artefacts as part of an overall solution. This aligns with research by Bhamra et al. (Bhamra, Lilley, and Tang 2011) and Lilley et al. (2018) working on sustainable behaviour, who also highlight the complexity of the design process in this context.

4.3. Challenges in practice

Behavioural designers currently encounter three main challenges in practice. First, typical cases deal with social behaviours and/or contexts that are difficult to change, making solution development highly constrained. The Project Manager and Lead Designer in Case 7 exemplify this: ‘we can't just [change the user environment], how much can behaviour design actually move the needle and improve [the situation] through the targeting behaviours?’ This fundamental constraint is also commonly compounded by clients who approach behavioural design as a ‘last resort’ and thus also constrain the technical solution space. The Designer in Case 10 stated: ‘they made so much resistance, so our new task was actually not going from [physical document] to [Information technology based system] but, having the new system adopted by [the users]’. Similarly, Designer in Case 10 suggested: ‘90% of all projects with behavioural design that I have been in, or 95%, are working with a solution already developed’.

Second, there was a general lack of awareness regarding behavioural theories among the client companies defining the cases projects, and a perception of artefacts as the primary objects of design. Thus, challenges were often encountered in convincing clients of the behavioural nature of the problem and the need for a behavioural design approach. This is highlighted by the Project Manager in Case 4: ‘in [the client’s] minds they have a great idea … they just want us to build it so they can sell it. and … we try to sneak the [behavioural] analysis and investigations in from the back door’. Furthermore, clients often lacked understanding of the behavioural theories at play, and thus challenges were often encountered when new problem behaviours were identified during the project. The Lead Designer in Case 1 stated: ‘it’s always a problem to tweak something that has been sold to a customer, and we want to do something else’.

Finally, due to the behavioural design being a relatively new field of design practice, the resource and process requirements are often unknown to non-behavioural designers and clients. This is illustrated by the Communication Designer in Case 10: ‘we often get challenged on the [need for] interviews, …  [the client says] ‘we know what is going on, or we already did the interview’’. Furthermore, the Lead Designer in Case 1 highlighted that while there is appreciation about the value of behavioural design, the awareness regarding its resource intensiveness is low: ‘people are quite interested in behavioural design … but it becomes an issue … . because you use quite a lot of time gathering data and in data analysis … my colleagues would [question], why are you going to use that much of the budget only on the analysis of the problem’. This, in addition to the lack of theoretical understanding noted in the second challenge, adds to the complexity of managing relationships with clients and collaborators.

Bringing these results together, a number of barriers to practise emerge regarding constraint, lack of key capabilities, and difficulties in managing relationships with those not familiar with the demands of behavioural design. Challenges of constraint have been hinted at in prior literature. For example, Meadows (Meadows 2009) highlights the challenges of changing a small sub-system of a large functioning system. Furthermore, in the public policy context, behavioural insights are often applied to ‘fine tune and improve implementation and compliance’, fundamentally constraining any behavioural interventions (OECD 2017, Pg. 15). However, challenges of capability and management associated with behavioural design remain unexplored. Finally, while some of these are more fundamental, such as constraints on changing the environment of an action, others are distinctly linked to the lack of widespread understanding and research on behavioural design in practice, such as knowledge regarding capability and resource requirements.

5. Discussion

Contrasting our results with the seven paradigms of design outlined in Section 2.2, Figure 1, we are able to answer RQ1: the observed behavioural design cases differ substantially in a number of fundamental ways from prior design paradigms; specifically, in terms of goal and object framing, use of behavioural theories and user psychology. A number of prior approaches, such as human factors (Sanders and McCormick 1993), design for sustainable behaviour (Lilley 2009), social design (Tromp, Hekkert, and Peter-Paul Verbeek 2011), persuasive technology (Fogg 2002), and advertising (Richards and Curran 2002), actively or overtly state changing behaviour as a goal. In addition, the analysis of cases shows that dealing with behavioural objects presents unique challenges to the designer and design process. Similarly, although human computer interaction (Hekler et al. 2013), design for emotions (Desmet 2008), and experience design (Hassenzahl and Tractinsky 2006) address user psychology as a reference they do not use it as a means of design or explicitly build on behavioural theories. Thus, we present the unique characteristics of behavioural design.

5.1. Understanding the unique characteristics of behaviour in design

The analysis of the cases highlights the crucial importance of positive behaviour in achieving successful resolutions to technical and social challenges. Bringing the defining elements identified in Section 4 together, we propose the following definition as an encapsulation of the observed design approach, formulated with respect to our research framework:

Behavioural design has the goal to explicitly and ethically realise positive behaviour, desired by both the individual and society; the object of design is behaviour itself, which is explicitly understood and designed for using behavioural theories and brought into effect by actively changing user psychology with the help of artefacts.

The most distinguishing aspect of behavioural design appears to be the formulation of the primary object to be designed in terms of behaviour. Designers conceptualise both problems and solutions with respect to behaviour. This means that the development of artefacts or other interventions – associated with specific user actions and targets – starts much later, as secondary objects of design. These secondary objects – often combining products, systems, communications, and social interventions etc. – serve the primary object and design goal. This contrasts prior paradigms which have typically focused on artefacts (Roozenburg and Eekels 1995; Pahl et al. 2007) or systems (Morelli 2002; da Costa Junior, Diehl, and Snelders 2019) as primary objects. These traditional design approaches, like product design, may also specify, study, and evaluate human behaviours to ensure that the technical functions are achieved. Here, the artefact takes precedence, and human behaviours are analysed in relation to the technical behaviours, and physical structure of the artefact, together with other socio-technical factors (Sun et al. 2013, fig. 1,2,3). In contrast to this, in behavioural design, the design of the behavioural objects leads to the design of the artefact object(s). Furthermore, we introduce the concept of a system of ‘secondary objects’ explicitly designed in service of achieving a behavioural goal. While this has not previously been described it does highlight parallels between behavioural design and approaches, such as systems engineering (Haberfellner et al. 2019), where multiple objects of various types must be coordinated. Thus, despite its unique features a number of insights from prior paradigms could be relevant to behavioural design.

Following the features outlined above, behavioural design is highly complex. Specifically, we describe how behavioural design combines the – often difficult to conceptualise – complexity of contextual, human behaviour in a social system (Fishbein and Ajzen 2010; Kahneman and Egan 2011) with that of a multi-component technical system (Weck, Roos, and Magee 2011). While these dimensions have been individually addressed to some degree in, for example, social design (Tromp, Hekkert, and Peter-Paul Verbeek 2011), within complex industrial product/machine design (Sun et al. 2013; Sun et al. 2018), and systems engineering (Haberfellner et al. 2019) respectively, they have not previously been so closely connected in a single design approach. Furthermore, the interaction and evolution of these systems must be managed in the long term, in order to account for changes in behaviour, both planed and unexpected, and across contexts (Michie et al. 2008, 2014). This places a significant complexity burden on the designer and makes the use of behavioural theories critical to understanding, developing for, and evaluating behavioural design projects and goals. In particular, behavioural theories are an essential element in helping behavioural designers work responsibly with user psychology as a means to bring about behaviour change. This mirrors many prior approaches that have built on technical theories in order to develop technical and material means (Hubka and Ernst Eder 2012; Altshuller and Altov 1996).

Bringing these insights together, we propose a tentative conceptual model for understanding behavioural design as an emerging paradigm in Figure 2. This illustrates the key features identified in this study and provides a basis for further work in this area.

Figure 2. A proposed conceptual model of behavioural design.

5.2. Enablers and barriers of behavioural design

In answer to RQ2 we identify three major barriers and ten enablers encountered across the observed cases. For each we identify the barrier and its corresponding enablers below, and summarise them in Table 3.

Table 3. Three key barriers and 10 enablers in behavioural design.

Barriers	Enablers
Joint behavioural and technical complexity	1. Use of behavioural theories to align analysis, development, and evaluation activities 2. Dividing overall behavioural goals into linked sub-behaviours 3. Explicit division between behavioural objects and artefact objects as sub-projects
Capability	4. Teams where members combine design and behavioural science expertise, by cross-training and skills development between design and behavioural science 5. Compilation of case compendia for training, reflection, and lessons learned 6. Collaboration with universities to structure knowledge via frameworks and methods 7. Project-specific business collaboration to provide specific capabilities
Management	8. Highlighting the behavioural competencies and challenges through e.g. social media 9. Formulation of projects with a clear emphasis on behavioural aspects and distinct process steps in project budgets 10. Communication of specific behavioural outcomes, such as behavioural analysis reports, and evaluation of outcomes through behavioural Key Performance Indicators

The first major barrier to behavioural design highlighted across cases was the high degree of complexity. In order to counter this, we observed three main enabling strategies. First, designers built extensively on behavioural theories, which were used to align analysis and development activities, simplify prototype testing, and guide evaluation and success criteria. This aligns with similar use of theories in the technical domain as noted in Section 5.1, and has been highlighted in a number of prior works on behavioural design (Francis et al. 2009; Cash, Hartlev, and Durazo 2017). Second, the behavioural objects can be divided into multiple sub-behaviours, which could be analysed and specified independently and then synthesised as a whole (Cash, Hartlev, and Durazo 2017; Michie et al. 2014). This is similar to the approach taken in complex engineering design (Pahl et al. 2007). Third, designers explicitly divided the projects into two distinct sub-projects, where the first sub-project dealt with understanding, specification and design of behavioural objects, and the second sub-project designed the artefact. As such, artefact development was temporally separated from a first analytical and theory development phase. This helped mitigate the increased complexity in behavioural design by separating major analytical and developmental foci.

The second major barrier encountered across cases was the diverse capability set required to understand psychology, social science, and technical systems, as well as to carry out effective development work. Here, the first enabler was the integration of design team members with training in both design and behavioural science throughout the design process. This aligns with calls for the inclusion of behavioural insights in university as well as continuing design education (Gemser et al. 2012; Norman and Stappers 2016), and highlights the growing interaction between design and the cognitive and social sciences. However, it was also emphasised in all cases that design expertise was essential in coordinating the project, ideating effectively across possible intervention types, and synthesising multiple interventions to foster real-world behaviour change. As such, it is essential that skills sharing and education flow both ways between the behavioural sciences and design in this context. A second enabler here was that most of the companies kept compendia of prior project so that patterns between problems and solutions could be identified and reflected on. This aligns with similar efforts to establish open compendia in this domain (OECD 2017; ‘The European Nudging Network’ n.d.). Such compendia were critical in on-boarding new designers, fostering continuing education with the organisation, and capturing lessons learned from resource-intensive data collection and testing activities. The third enabler is the association of the design consultancies with universities that help in generating frameworks and methods situated in practice. Two consultancies out of four in this study have active collaboration with design researchers who work with behavioural design. The fourth enabler is project-specific consultation from other businesses. The consultancies in this study have engaged other consultancies that deal with a number of specialists in order to compensate for the wide variety of possible behaviours and contexts encountered. These include software specialists for the development of persuasive technologies, prototyping specialists, and so on. This allows the companies, with limited resources, to address the wide variety of artefact development challenges found in the cases.

The final observed barrier was managing the projects within the constraints of lack of understanding about underlying logic and resource requirements. This was addressed in three ways. First, all companies made substantial efforts to communicate the logic of behavioural design to their clients, and provided extensive informational materials early in the design process. This is complemented by a more general strategy of market education supported by social media and popular sentience articles produced by several of the case companies. Second, behavioural design companies typically elaborate the behavioural component when formulating the project, by highlighting the necessary process steps required, such as observational studies and interviews. These projects highlight the behavioural component of the project and recruitment of behavioural design specialists as part of budget. Third, the management of the client relationship is enhanced through explicit communication of behavioural outcomes. In most cases, the consultancies create behavioural analysis reports and separately communicate these to the clients during the design process. Here, target behaviours are treated as explicit ‘Key Performance Indicators’ and used to illustrate the impact of the behavioural solutions.

6. Limitations and implications for design theory, practice, and education

Prior to considering the implications of this work, it is important to understand its limitations. First, the focus on understanding the emerging characteristics of behavioural design demanded the selection of cases based on overt focus on behaviour in design. As such, these results provide in-depth insight into a select domain and further work is needed to establish the generalisability of the findings. This could also be supported by comparative studies of a wider range of cases reflecting other modern design paradigms. Second, the cases reflect a limited set of designers and design companies. However, the commonality of the findings across cases suggests recurring patterns and insights that do provide a basis for furthering the discussion of behavioural design practice. In particular, the sampled companies reflect some of the leaders in this area in Denmark, which is itself a leader in this domain. As such, further work is needed to examine the generalisability of these insights across countries and design educational systems.

Despite these limitations, this work holds implications for theory, practice, and education. In terms of theory, the proposed conceptual model (Figure 2) and evolving understanding of behaviour in design (Figure 1 and the characteristics presented in Section 5.1) provide a basis for theory building. Specifically, behavioural design appears to have a number of characteristics that explicitly differentiate it from prior design paradigms. As such, it is possible that behavioural design might reflect the emergence of a new design paradigm. This points to the need for further research on how this might impact understanding of the design process, designer capability development, and design education, as has been the case with other paradigms. Specifically, we highlight a number of unique features of behavioural design that have not been previously described and thus require further research in order to better understand and support in practice. Further work is required to examine how the conceptualisation of behaviour, as the primary object of design, impacts the design process, and might interact with other more technical design process models. The model in Figure 2 provides a basis for these investigations and distinguishes behavioural design from prior paradigms and approaches.

In terms of practice, we offer a number of concrete enabling recommendations and highlight key barriers (Table 2). Behavioural designers should focus on developing strong multi-disciplinary capabilities within their teams as well as investing in managing client and collaboration relationships. This is particularly critical, as the current state of general understanding regarding behavioural design is typically low outside of specialist behavioural design companies.

Finally, in terms of education, this work highlights the need to further develop education of behavioural science skills in design, whilst also introducing relevant design skills to behavioural science. Further work is needed to examine the most effective means for educating behavioural designers in the long term.

7. Conclusions

The aim of this work was twofold. First, to examine how the role of behaviour in design has developed over time, and second, to identify critical enablers and barriers of behavioural design work in practice. In answer to this, we have proposed a conceptual model of behavioural design that highlights its unique features, which could lead to elaboration of a new behavioural paradigm in design. Specifically, we emphasise a focus on behavioural goal framing as well as behaviour being the primary object of design. We further elucidate a strong focus on the use of behavioural theories and user psychology in both the design process and proposed solutions. Finally, we propose three key barriers to behavioural design in practice, and ten corresponding enablers. The explicit formulation of behavioural design in this work sets a foundation for further research, practice, and education in this area.

Acknowledgements

The authors thank all the companies that supported us in collecting the data used in this study. They thank team members who participated in the case interviews. The authors also thank the reviewers for helping develop the final manuscript. Pramod Khadilkar received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 754462.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant: [Grant Number 754462].