Lost in the diffusion chasm: Lessons learned from a failed robot project in the public sector

Abstract

Public sector organizations increasingly engage in robotic innovation projects to assist or substitute for humans in service delivery. However, transitioning small-scale development projects into a large-scale context is a notoriously difficult task that often fails, with many promising robotic projects becoming lost in the diffusion “chasm.” We investigate a failed robotic diffusion project to analyze what went wrong and what can be learned from it. Despite an increased interest in learning from public sector digitalization failure, little attention has been paid to how and why seemingly successful service robot initiatives fail to move beyond the pilot stage. We identify three types of explanations for diffusion failure using an in-depth case study of a service robot initiative in the Danish eldercare sector that had a high degree of management support and commitment from key stakeholders. Our analysis demonstrates how the failure was caused by interrelated and context-specific reasons regarding the lack of technological maturity of the service robot (technology-oriented explanations), inadequate problem-solution fit in the conceptual design (scope-oriented explanations), and misalignment between the robot company and public sector organization mindsets (competing logic-oriented explanations). We outline the lessons learned for public sector digitalization and discuss the paradox between the hype surrounding robot innovations and their slow diffusion.

Keywords

Public sector digitalization service robots diffusion diffusion chasm learning from failure

1. Introduction

It has long been acknowledged that digitalization has a transformative potential for public sector organizations (Danziger et al., 1982; Dunleavy et al., 2006; Fountain, 2004; Meijer, 2015; West, 2004). In the recent discourse on digital transformation in the public sector (Clifton et al., 2020; Mergel et al., 2019), robots such as surgical, cleaning, medicine-dispensing, and rehabilitation robots are increasingly sought and introduced as smart ways to transform service delivery (Beane, 2019; Čaić et al., 2018; Nielsen et al., 2016; Oborn et al., 2011). Aging populations and new technological developments have paved the way for robots to move out of industrial contexts and into public sector service areas, enabling new opportunities for home-based aging and independent living (Frennert et al., 2020; Mettler et al., 2017; Robinson et al., 2014). Nonetheless, public sector research has only considered this development to a limited extent, and scholarly calls have been made to pay much more “attention to robots in what [has] been called the second machine age” (Nielsen et al., 2016, p. 139).

Despite high expectations, few robot projects in the public sector have made it from the prototype stage to real-world practical use (Blackman, 2013; Glende et al., 2016; Van Aerschot & Parviainen, 2020), resulting in many initiatives becoming lost in the diffusion “chasm” between experimenting visionaries (early adopters) and mainstream pragmatists (the early majority) (Moore, 2014). This diffusion challenge represents a significant problem since the potential values of robot innovations are rarely obtained, resulting in missed efficiency gains for public sector organizations, missed market potentials for robotic companies, and eventually, missed opportunities for improving care recipients’ quality of life. Nevertheless, scarce research attention has been paid to why service robot innovations fail to move beyond the pilot stage and learning points have not been amply explored.

In this article, our purpose is to examine a significant example of a failed robotic diffusion1

¹
Following Fichman (2000), we define diffusion as “the process by which a digital innovation spreads across a population of organizations” (p. 1). This differs from the concept of adoption, which concentrates on the uptake of a digital innovation by a single organization (or individual).

project in Danish eldercare to identify key learning points. Although failures are, of course, not wanted in public sector digitalization, they also offer opportunities for learning as we demonstrate in this article. We draw from an in-depth case study of the rise and fall of a service robot for independent living, which we refer to as BathroomBot. BathroomBot was well on its way to becoming a successful public sector digitalization initiative – promoted as a “revolutionizing solution” by politicians (D1),2

All directly cited documents are referenced using the format (D#), where # stands for the document number (Appendix A).

“the future of care” in national newspapers (D2), a “business adventure” by the industry (D3), and celebrated for its user-centered design approach (D4). Despite the support and commitment from key stakeholders, BathroomBot failed to move beyond being a small pilot project and never became diffused on a larger scale (D5). Though limited to a single-case setting, our study provides a unique opportunity to learn from failure. Our study is guided by the following research questions: Why do seemingly successful robotic innovations fail to be diffused on a larger scale? What key lessons can be learned from diffusion failure in public sector services?

While current research mostly explains robotic diffusion failures from end-user perspectives, such as the lack of user acceptance (Papadopoulos et al., 2020; Van Aerschot & Parviainen, 2020), our BathroomBot analysis examines how key actors from both adopting organizations and robot company developers experienced and explained the diffusion chasm they encountered. In doing so, we identify three types of explanations for diffusion failure; (1) lack of technological maturity of the robot (technology-oriented explanations), (2) inadequate problem–solution fit in the conceptual design (scope-oriented explanations), and (3) misalignment between the robot company and public sector organization mindsets (competing logic-oriented explanations). Based on these insights, we identify seven lessons learned for crossing the chasm in diffusing robotic innovations in public-sector services.

The remainder of this paper is organized as follows. In Section 2, we review the related literature on the diffusion chasm (Moore, 2014) of service robots in the public sector before presenting our research design and method (Section 3). The analysis is structured into the three suggested types of explanations for the failure of BathroomBot (Section 4). We conclude by discussing our study’s practical and theoretical implications, limitations, and recommendations for further research (Sections 5 and 6).

2. Background

2.1 Robots in public sector service delivery: Diffusion challenges

Robots can be defined as programmable machines capable of – autonomously or semi-autonomously – carrying out complex series of actions (Mettler et al., 2017; Wirtz et al., 2018). Robots that operate in the public sector services exist in various forms and serve various functions, including physical, cognitive, medical, and psychosocial support (Frennert el al., 2020). The idea of using robots to assist in public sector service delivery in order to free up resources and support independent living has been around for many years. More than twenty years ago, Engelberger (2000) argued that the technology was already sufficiently developed to produce “a household robot for the elderly or disabled, that is able to undertake a wide range of domestic chores and supply 24-hour care and companionship” (p. 176). However, realizing this vision in practice has so far largely been unsuccessful (Van Aerschot & Parviainen, 2020).

In contrast to the widely diffused industrial robots that predominantly work in controlled settings and with limited human interaction, service robots in the public sector need to operate in close contact with people in less controlled settings (Mettler et al., 2017). This poses new design challenges; these robots’ success lies not only in their technical functionality but also in their ability to adapt to the user’s living conditions, especially regarding social interaction and appearance (Cresswell et al., 2018; Martinez-Martin & Pobil, 2018). It also raises several ethical concerns related to loss of human contact, feelings of being objectified, and control over one’s own life (Sharkey & Sharkey, 2012). Consequently, the use context for service robots is far more complex, and many initiatives are characterized by “failed attempts to scale up locally, spread distantly, or sustain the innovation long term at the organization or system level” (Greenhalgh et al., 2017, p. 1).

With limited uptake of robots in public-sector services, research tends to rely on imaginary scenarios or small-scale prototype tests (Van Aerschot & Parviainen, 2020). In this context, current research mostly explains this diffusion challenge through adopter-centric factors, such as the lack of acceptance among end users (Papadopoulos et al., 2020); different technological frames among end users, caregivers, and managers (Frennert et al., 2020; Mettler et al., 2017; Nielsen et al., 2016); or the motivations of early adopters influencing later diffusion processes (Compagni et al., 2015). Other scholars have stressed the appearance of robots, especially regarding humanoids being perceived as either “too human-like” or “too robotic,” as an important diffusion challenge (Cresswell et al., 2018).

While providing valuable insights into barriers for diffusing robots into public sector services, these studies only paint parts of the picture of the diffusion challenges. Rather than being primarily rooted in negative user attitudes or ethical concerns, Van Aerschot and Parviainen (2020) points to problems in manufacturing to be critical elements in the failed diffusion of service robots in the public sector. Accordingly, the field would benefit from including viewpoints from both adopting organizations and robot company developers to explore why robotic innovations in the public sector often fail to be diffused on a larger scale.

2.2 The diffusion chasm

To frame our investigation of service robot diffusion failure, we turn to Moore’s (2014) “diffusion chasm” approach as an appealing starting point for understanding the challenging gap that can occur between the initial hype surrounding robot innovations and their slow diffusion. Expanding the diffusion of innovation theory (Rogers, 2003), Moore (2014) has introduced the notion of “crossing the chasm” between visionary early adopters and the pragmatic early majority that developers and technology suppliers need to overcome in order to accelerate the diffusion of new disruptive innovations across each adoption segment in the market place. Moore argues that “visionaries” (early adopters) and “pragmatists” (early majority) have very different expectations and reasons for adopting new technologies and that failing to address such differences can have fatal consequences for diffusion. For instance, while early adopters are likely to be pioneers “driven by a dream” (Moore, 2014, p. 42) and willing to engage in high-risk, high-visibility projects, the early majority tend to be more risk-averse and seek more predictable changes; they are therefore able to patiently wait for the product to become more established in order to see how other adopters have fared with it.

To help cross the diffusion chasm, Moore (2014) outlines a set of general advice regarding how developers should 1) carefully target a specific user segment, 2) transform the pilot product into a “whole product” (i.e., providing supplementary services and products to ensure that users can benefit from the value proposition), 3) position the innovation in relation to existing solutions for comparative evaluations, and 4) select the most appropriate distribution channel.

While Moore’s (2014) work essentially focuses on marketing high-technology products in the early startup phase, the focus and the gap identified in our case study has a different focus (a service robot innovation that ended up as a diffusion failure) and a different context (public sector services). Cho and colleagues (2009) provide an example of applying Moore’s chasm approach in the context of healthcare as they studied the path from invention to commercialization and diffusion of a telehealth innovation in hospitals. In another study, Krishnan and colleagues (2019) draw on the notion of the diffusion chasm to examine the adoption and acceptance of surgery robots. While these studies come closer to our empirical context, they report successful examples of crossing the diffusion chasm. Our study aims to analyze what goes wrong in the robotic diffusion process and to detect learning points in the context of public sector digitalization.

3. Method

3.1 Research setting and case

We report an in-depth interpretive case study (Walsham, 1995) of the rise and fall of BathroomBot – a cutting-edge service robot in Danish eldercare – to examine and learn why seemingly successful service robot initiatives fail to move beyond the pilot stage. Danish eldercare services are publicly financed and delivered by 98 municipalities (local governments). Besides nursing homes, eldercare consists of homecare services, including assistance with practical errands (laundry, cleaning, food delivery, etc.) and personal hygiene and health (bathing, medication, toilet use, etc.) (Nielsen & Andersen, 2006). There have been attempts to outsource eldercare services to private providers, but to date, most services are provided by municipal nursing homes and homecare organizations (Nielsen et al., 2014).

Along with other Scandinavian countries, Denmark is a front-runner in the digitalization of public sector eldercare services. For more than ten years, emerging technologies, such as virtual homecare, tracking technologies, and robots, have comprised a strategic focus area for digital innovation in Scandinavian public sectors and are anticipated to play a critical role in finding solutions to demographic challenges (Frennert, 2019). Labeled as “welfare technologies” (Aaen et al., 2018), these solutions are expected to assist clients in their daily lives, reduce service delivery costs, and create a better work environment for service providers. This strategic focus has resulted in a myriad innovation projects, with over 1,450 registered in Local Government Denmark’s (2021) – the interest organization for all 98 Danish municipalities – database of welfare technology projects in Danish municipalities by 2021.

In this context, the highly acclaimed BathroomBot stands out. BathroomBot was regarded as a welfare technology flagship project, both at its launch and in subsequent years (D6–7). It was featured on national media (e.g., D8–13), highlighted in political strategies (D14–17), and showcased at international exhibitions (D18–22). Moreover, the idea for the service robot originated at the work practice level in eldercare, and the initiating municipality expected the robot to assist several hundred clients in need of help (D23).

Figure 1.

Conceptual design for BathroomBot, which assists users in dressing and undressing when using the toilet. Source: D78.

The robot was designed to help the elderly and people with disabilities, who currently depend on personal assistance when going to the bathroom (D24–25). Currently, clients who have reduced mobility in their arms, hands, or upper body can receive homecare nurses’ assistance in toilet use 5–6 times a day. The purpose of BathroomBot was to address this problem by creating a robot – mounted on the wall behind the toilet – that could help users undress and dress (see Fig. 1). The robot would thereby improve the users’ quality of life by providing them with more freedom and independence. In return, this would reduce both the need for assistance (fewer visits) from homecare nurses and the number of needed daily home visits, making it economically attractive to the municipalities. Despite being appraised as having “disruptive” potentials among visionary early adopters, the robot failed to cross the diffusion chasm and be implemented on large scale. Ultimately, the company developing the robot filed for bankruptcy in 2019 (D5). Table 1 provides an overview of BathroomBot’s lifecycle.

Table 1

Project timeline

Early 2015	Mid- 2015	Late 2015	Early 2016	Late 2016	Early 2017	Mid- 2017	2018	Early 2019
Municipality identifies a problem when implementing wash toilets	Contract between initiating municipality and robot development company	High expectations and media attention on the robotic innovation	Two additional municipalities buy the robot	The robot is the “talk of the town” at several international trade fairs.	More municipalities buy the robot	The robot company attracts new investors	The adopting municipalities are unable to implement the robot in practice	The robot company files for bankruptcy

The case, the robot, and the stakeholders involved have been anonymized. Furthermore, all quotes from interviews and documents have been translated from Danish to English.

3.2 Data collection

The data were obtained from both primary sources (interviews) and secondary sources (publicly available archived documents) (Van de Ven, 2007). The data were collected in 2020 in three steps, as summarized in Table 2. The first step of the data collection process was to systematically assess publicly available documents to map out a timeline of project incidents and stakeholders. The included documents amounted to 39 news articles, 24 webpages, and 6 videos (comprising 4 television news reports and 2 product demonstration videos). As the BathroomBot product webpage was terminated (due to bankruptcy in 2019), we reconstructed and retrieved 17 otherwise inaccessible and deleted documents, including subpages and project newsletters, using the Internet Archive Wayback Machine (web.archive.org). The second step of the data collection involved chain sampling by pursuing links and references found in the included documents, searches on participating municipalities’ websites, and documents received by the interviewees. These amounted to 35 additional documents, such as city council meeting logs and appendices describing the municipality’s involvement in the project, business cases, project plans, and product evaluations.

Table 2
Summary of data sources

	Source	Quantity
STEP 1:	Media search (infomedia.dk)	39 documents (89 pages)
Mapping project incidents and	Web archive (developers webpage)	17 documents (64 pages)
stakeholders	Web searches on “BathroomBot”	24 documents (121 pages) 6 videos (11 min)
STEP 2: Chain sampling for additional	Searches on participating municipalities’ websites (e.g. city council meeting logs)	20 documents (67 pages)
documents	Chain referral (following links and references in included documents or received by interviewees)	15 documents (92 pages)
STEP 3: Interviews with key	Developer (1 founder, 2 employees, 1 investor)	4 interviews (239 min)
stakeholders	Distributor (1 sales manager)	1 interview (42 min)
	Municipality #1 (2 project managers, 1 nursing home manager)	3 interviews (77 min)
	Municipality #2 (1 project manager)	1 interview (47 min)
	Municipality #3 (1 project manager, 1 test coordinator)	2 interviews (119 min)
	Municipality #4 (1 project manager)	1 interview (57 min)
	Municipality #5 (1 project manager)	1 interview (46 min)
		Sum 115 documents (433 pages) 6 videos (11 min) 13 interviews (627 min)

Finally, by assessing the collected documents, we were able to construct a timeline of events and map key actors (from all organizations involved in BathroomBot) whom we would contact for interviews. These included respondents among the robot company (the founder, employees, and the investor), the adopting municipalities that had bought one or more BathroomBots (primarily project managers in charge of testing, implementing, and evaluating BathroomBot), and the external distributor (sales manager in charge of the initial marketing). The interviews were conducted in the spring of 2020. Due to the COVID-19 pandemic, these interviews were held over Skype or on the phone, following Lo Iacono and colleagues’ (2016) recommendations for online interviewing. The respondents were asked to (1) describe their role in the robotic innovation project, (2) reflect on how the project unfolded and the challenges they encountered, (3) discuss strategies and interventions taken, (4) give reasons for the project failure, 5) offer some potential lessons learned and discuss this project’s impact on future projects in the public sector. Two interviews conducted in 2018, one with a project manager and the other with a nursing home manager from the initiating municipality, were also included.

As suggested by Van de Ven (2007), the variety of sources and stakeholders we used enabled us to identify and analyze a diverse range of (sometimes inconsistent and contradictory) reasons for BathroomBot’s unsuccessful diffusion.

3.3 Data analysis

In accordance with recognized principles for thematic data analysis in qualitative research (Miles & Huberman, 1994), our data analysis focused on how BathroomBot developers and adopting organizations experienced and explained the diffusion chasm they encountered. Following the advice of Gioia and colleagues (2012), the initial coding iteration closely matched the empirical accounts found in the vast number of documents and the interviews with key informants. This first-order analysis resulted in a myriad of observations concerning the challenges surrounding BathroomBot’s diffusion (see Table 3). In the second coding iteration, we grouped these various observations into conceptual themes of different reasons why the robot failed. Next, we distilled these themes into three aggregated types of diffusion failure explanations.

Table 3
Data analysis

Empirical themes	Conceptual themes	Aggregated themes
• Low usability • The robot only works with specific types of clothing	Technological immaturity	Technology-oriented explanations
• Practical barriers, such as the size and design of the robot, excluding potential users • Restrictions for mounting the robot and regulations in terms of electricity near running water in the bathroom, excluding potential users	Technological misfit in use context
• The solution did not match the needs of a user segment • The robot only solves an isolated problem and overlooks others in this process	Inadequate problem-solution fit	Scope-oriented explanations
• The potential user segment and business case were heavily exaggerated • The potential users’ cognitive and physical abilities deteriorate too quickly to effectively sustain usage	Complex user segment
• Slow sales and implementation processes in municipalities • The company was too focused on achieving early sales	Colliding mindsets	Competing logic-oriented explanations
• Care workers were needed as gatekeepers to effectively identify and screen potential end users • Care workers as showstoppers because of fear of replacement	Gatekeeper bottlenecks

The first type, which we refer to as technology-oriented explanations, concerns technological issues and whether the robot needed further product development and user testing. While these explanations praise the conceptual idea for the robot, they explain diffusion failure as the lack of technological maturity of the final product offered to the clients. Consequently, BathroomBot could succeed through incremental refinements and adjustments to improve usability. In contrast, the second type of explanations points to substantial flaws in the innovation idea that cannot be incrementally fixed through additional testing and development iterations. According to these scope-oriented explanations, BathroomBot did not solve the problem it was trying to address, and its developers would need to radically rethink the scope of both its conceptual design and business case to succeed. Finally, while the two previous types of explanations point to limitations in the product and its fit with the use context, the third type of explanations for BathroomBot’s failure notes factors in the innovation process and the collaboration between the robot company and public-sector organizations. These competing logic-oriented explanations indicate conflicting interests and misalignment of expectations between the inventing startup company and the adopting municipalities. Table 3 provides an overview of the data analysis and the identified explanations.

Going back and forth between our background literature and our empirical findings, we draw on these conceptualizations of different explanations for the failure to construct a set of managerial lessons for diffusing emerging robotic innovations in public sector services.

4. Analysis

Our analysis of BathroomBot’s demise is structured into four segments. First, we set the scene by analyzing the launch, the project setup, and expectations surrounding BathroomBot. Then, we unfold the three types of explanations for BathroomBot’s failure that we encountered.

4.1 Setting the scene: A daydream turning into a nightmare

The idea for BathroomBot originated in a municipal eldercare department that installed automated washing toilets for clients living at home and then realized that many potential users still needed assistance with taking their pants off and putting them back on. Seeing the potential in developing a technical solution to aid clients with undressing, the municipality approached a local robotics company with the idea. The company primarily specialized in making industrial robots but had previously collaborated with the municipality on a robot for patient rehabilitation. After the initial market research, screening for existing solutions, and preparing a business case, a collaboration was announced in 2015 concerning the development of this concept and a prearranged purchase of several robots (D6). It was predicted that 700 clients in homecare could be included in the target group in the initiating municipality. The business case estimated annual savings of 16,800–24,500 €per robot, meaning that a unit price of 13,500 and could be repaid in less than a year (D115).

BathroomBot was announced in a press release, with photos of the executive city councilor and the company founder signing the contract (D6). In the announcement and the following media coverage, the visionary innovators of BathroomBot emphasized a bottom-up approach, close collaboration between developers and the municipality, and a high economic potential (D27, D28). In the following years, BathroomBot was celebrated on mainstream media as a disruptive innovation (e.g., D2, D11, D29), demonstrated as a successful digitalization project at Parliament (D30), and nominated for a national startup award for an innovative solution that “addresses a significant need in the market that has not yet been met” (D22). Thus, BathroomBot was well on its way to become a prosperous public sector innovation. It had early success in terms of product sales to several municipalities, political and managerial support, nationwide media publicity, and financial investments from venture capitalists. However, despite several municipalities having purchased BathroomBots, it was difficult to find suitable clients in the homecare system who would and/or could participate in testing the robot and subsequently having it implemented (D23). Thus, what looked like an appealing context for widespread diffusion turned into a disappointing failure and ultimately bankruptcy within a few years.

4.2 Technology-oriented explanations for failure

The first type of explanations for BathroomBot’s failure, and perhaps the most obvious, concerned technological issues. From this perspective, the idea for the robot was good, but it lacked technological maturity and suffered from low usability for many potential users. An often-mentioned usability issue among the informants concerned the physical and the cognitive requirements for using the robot (D31). For instance, to ensure that the robot’s arms were able to pull the user’s pants up and down, the users should be able to stand up straight. This was a requirement that many potential users had difficulty fulfilling, as “people who have so many problems that they can’t even put their pants on typically also have a hard time standing up” (Project manager, Company).

It is a classic engineering error, in which you identify a need and ask some experts if the demand is there. ( $\ldots$ ) then we have two really talented engineers who go home to the workshop, delve completely into it, and figure out how this thing should work. And then they resurface and say, ‘Now we have a finished prototype or product that can satisfy the need properly.’ The problem is that you have just forgotten to test whether the clients can actually use it. (Project manager, Company)

Another technology-oriented explanation concerned BathroomBot not fitting the use context, including the fact that due to its physical size, the robot would not necessarily fit in many clients’ bathrooms, or limitations due to restrictions on altering the bathroom and using electricity near running water in the bathroom:

In our municipality, we found potential clients who were able to use the robot but were excluded as the bathroom was too small or designed in a way that was unfit for the robot. (Project manager, Municipality #5)

There were also barriers related to aesthetics, in terms of both the robot’s design and limitations on users’ clothing, as the robot only functioned properly on elastic pants and skirts without buttons or zippers:

I think it frightened people that it was so big. The robot looked monstrous. And when they became aware that it also restricted their style of clothing, many people declined the robot. (Project manager, Municipality #2)

Informants pointing to technology-oriented explanations for the failure of BathroomBot suggested the need for more user testing and more product development iterations:

My advice to the developers is that they should simply do some more testing. For instance, people struggling with unilateral paresis do not stand straight up. The idea is good. But they should have included many more clients for testing. (Project manager, Municipality #2)

In sum, technology-oriented explanations stressed that BathroomBot had the potential to cross the diffusion chasm if the robot could be further developed and improved, following a “user-driven process, where the target group is continuously involved in a development process” (Evaluation report from Municipality #1, D23, p. 25).

4.3 Scope-oriented explanations for failure

While the technology-oriented explanations pointed to issues that could be mitigated by additional development that would improve usability and practical feasibility, other informants mentioned much more substantial problems for the robot that could not be fixed through additional testing and development. In the scope-oriented explanations, the robot itself might be technologically sufficient but did not satisfactorily solve the users’ problem of being dependent on assistance from care workers, resulting in an inadequate problem-solution fit:

The robot only performs an isolated task and overlooks other activities in this process. For instance, what about the many clients who wear panties or diapers? I think half of all women age 80+ are incontinent. So, the robot wouldn’t be of any use to those if it couldn’t change the diaper in the process. They have looked at an isolated problem – and have not looked at the problem as part of a toilet visit process. That is why it did not succeed. (Project manager, Municipality #3)

Consequently, in order for the robot to create value in terms of supporting independent living, potential users (i.e., clients in need of assistance with toilet use) not only had to meet the physical and the cognitive requirements to operate the robot but should also not be in need of further assistance related to the service:

Generally, our experience is that when a client asks us for assistance with going to the toilet, they pretty much cross the last frontier in their personal sphere of intimacy. They ask for assistance with bathing and many other types of services a long time before asking for help with going to the toilet. (Test coordinator, Municipality #3)

Thus, the scope-oriented explanations also highlighted the complexities in specifying the user segment, as the users had multiple individual needs that were addressed simultaneously when care workers visited them. This also meant that the initial expectations for the robot’s business case (and ultimately, the potential user segment) were highly exaggerated as the municipalities needed to find clients who could and would use the robot – and ensure that this would result in fewer visits from care workers:

Perhaps the business case was a little too optimistic and took the wrong approach. You need to find clients who receive care worker assistance for the specific task, which corresponds to at least the same cost as the robot. And the higher the price, the fewer the potential clients. (Project manager, Municipality #1) We made a fatal mistake in believing the municipal consultants who said, ‘We know that we have 300–700 people who need something like this.’ But the consultants do not know that. It is assumed that the municipalities have a full overview of the needs of their clients. But there is a long way from consultants to what’s going on in practice. (Project manager, Company)

Overall, informants pointing to scope-oriented explanations for failure tended to be very skeptical about whether BathroomBot could ever be diffused on a larger scale without radically rethinking the scope of both its conceptual design and business case:

The robot did not fulfill a need that was actually there. We cannot implement it if there are no clients for the product – no matter how creative we are. (Project manager, Municipality #3)

4.4 Competing logic-oriented explanations for failure

The third type of explanations redirected attention away from the technological aspects and scope of the robot solution toward the clashes between a sales-driven startup company and long-term collaboration-oriented, public sector organizations involving multiple stakeholders. As pointed out by informants from the robotics company, the way from vision to actual use of the robot could be very long. A slow and complex sales cycle in the municipalities, in combination with the relatively high burn rates for a startup company, led to colliding mindsets and eventual financial disaster for the project:

When you hire four employees and have a burn rate of, let’s say, DKK 400,000 a month – then you need to make some sales. But the sales processes to municipalities have so many stakeholders. So many interests that need to be addressed, and so many who have something to say. So, it may well be that there is a great potential for selling many robots, but before they get to actualize the purchase, it takes – let’s say, two years. And if you cut the staff and expenses, you cannot come out and make sales or support the customers you already have. It’s quite a pickle. (Founder, Company)

Even when a sale was completed, ensuring the implementation of the robot in practice could be equally challenging and demanding:

The implementation process – especially when you have such a radical product as ours – takes a really long time. You meet a lot of skeptical people, and you have to involve the whole organization before you can actually see if the product will function in practice. This is also something that works against a startup business. I think we assumed that we just delivered the product at the front door, demonstrated how it worked, included a user manual – and then waved goodbye. But that’s just not how it works in municipalities – they need a lot of help with implementing the things they buy. (Project manager, Company)

Thus, having spent a lot of money on product development, in combination with a high burn rate to facilitate sales, as well as the costs of implementation and technical support activities, the company was faced with quickly depleting cash reserves and thus applied a sales-driven strategy to cross the diffusion chasm at the expense of further development:

The mechanical engineer and I started to raise several red flags that could be fixed if we made some significant adjustments to the product. But the board was simply not willing. Instead, the strategy was to try to sell the robots to whoever could use them, seeking to repay some of the deficit. Only then could we begin talking about further development. We were only allowed to make small adjustments, which was far from enough to make it an optimal product. (Project manager, Company)

While the sales-driven approach might have been necessary for financial reasons, it fitted poorly with the municipalities’ intention. Instead of a transactional relationship, going from one sale to the next, the municipalities bought the robots with the idea of engaging in long-term collaboration to further develop the robot:

By paying for the robot, we sought to engage in more equal collaboration, which I often find to produce better results as long as the developers use our inputs. However, because there was money on the table, they believed that the big sale was halfway there. They seemed to lose focus on what was needed to move on. It’s a classic mistake of the company sending too many salespeople and too few developers. (Project manager, Municipality #4)

Another clash in logic between the startup company and the municipalities concerned gatekeeper bottlenecks, as the care workers were tasked to identify and screen potential users. However, as the economic rationale for the robot revolved around reducing the need for homecare visits, the company speculated that this ultimately hindered diffusion by creating resistance and fear of replacement:

We assumed that since the municipality had bought the robots, they would, of course, also use them. And this is where we were wrong. It is crucial to understand where the resistance to using these technologies came from ( $\ldots$ ). We talked a lot about how we could handle the resistance and convince the care workers that the technology would not be a threat but something that would improve the service and at the same time, provide better working conditions. There were a lot of frustrations and discussions about how we could overcome care worker resistance. But we did not find the answer. (Investor, Company)

Regardless of whether it was due to care worker resistance, technical limitations, or inadequate problem–solution fit, the municipalities struggled to find suitable users. In this diffusion chasm, a negative attitude spiraled among the pragmatic adopter segments, effectively stopping BathroomBot from transitioning beyond the expectations from visionary innovators and into real-world practice:

The negative attitude appeared as soon as we were introduced to BathroomBot. That is, when we received the robot, the project group was assembled, and we were informed about all the requirements for clients’ eligibility to use the robot. We started to see the challenges, and as we found it difficult to find suitable clients, the attitude became more and more negative (Project manager, Municipality #3)

Overall, competing logic-oriented reasons for failure pointed to the need for better alignment of expectations between the robot company and public sector organizations.

5. Discussion

The BathroomBot case demonstrates that even with initial success, support among key stakeholders, and obtained financial investments, crossing the diffusion chasm is a treacherous process with a high risk of failure. Without a close examination of such “critical incidents,” we miss valuable learning opportunities (March et al., 1991, p. 2). The analysis of BathroomBot’s demise reveals three aggregated explanations for diffusion failure in robotic innovations in public sector services. Based on these insights, together with an inspiration from Moore’s (2014) diffusion chasm framework and extant literature on service robots, we provide a set of managerial lessons to help diffuse robotic innovations in public sector service delivery. As recently called for by Van Aerschot and Parviainen (2020), these lessons combine perspectives from a broad range of key actors involved in BathroomBot to understand better the diffusion chasm for robotic innovations and how to maneuver it.

5.1 Lessons learned for crossing the diffusion chasm

Moore’s (2014) diffusion chasm framework stresses the difference between visionary early adopters willing to embrace promising but incomplete technologies and risk-averse pragmatists waiting to adopt proven and ready-to-use solutions. The technology-oriented explanations for BathroomBot’s failure demonstrate the long way from visions of disruptive potentials to a feasible robot in a complex use context. While the journey from an idea to a functioning prototype can take years, transforming this prototype into a product that fits users, surroundings, and organizational contexts can be equally demanding, if not more so. From this, we can learn two technology-oriented lessons for transitioning robotic innovation from small-scale development projects into large-scale contexts in public sector service delivery:

1.
Design robots for difficult-to-control use contexts. While the use context in industrial settings can be standardized and adapted to a greater extent to fit specifically designed robots, this is not the case for service robots in clients’ homes, operating with human agents rather than on the assembly line (Mettler et al., 2017). In the BathroomBot case, we observed how the service robot turned out to have so many physical, cognitive, and practical requirements for the users and their homes that the municipalities were unable to find suitable clients. As such, designing a scalable robot solution needs to focus on establishing requirements for the technology that are based on a careful analysis of specified user segments in unpredictable use contexts.
2.
Test technological maturity in real settings rather than in demonstration centers. Initially, the developers of BathroomBot was praised for its user involvement and bottom-up approach, with the idea emerging from practice, early sales, and successful product demonstrations at various events. However, respondents point to issues concerning usability and the lack of user testing in the later stages of product development. Hence, the BathroomBot case is an illustrative reminder of the importance of ongoing user involvement in real-world settings throughout the innovation process (Čaić et al., 2018; Donetto et al., 2015; Grönvall & Klyng, 2013) to ensure technological maturity.

Once a user segment has been targeted, Moore (2014) recommends committing to develop a “whole product” that adequately solves the users’ core problem and provides a compelling reason to buy the product. With the BathroomBot idea originating from a work-practice level and the product being developed in close collaboration with municipal stakeholders, BathroomBot appeared to be aligned with Moore’s recommendation. However, the scope-oriented explanations for BathroomBot’s failure suggest that this approach to ensure problem-solution fit can be deceiving in the specific context of robot innovation for public sector services. From this perspective, we learn three lessons for diffusing robotic solutions in public sector services:

3.
Carefully define and validate the user segment. A key element in Moore’s (2014) approach to crossing the diffusion chasm is carefully targeting a specific user segment to “secure a beachhead” (p. 79) for entry into pragmatic adopter segments. In the BathroomBot case, we saw how this can be a cumbersome task in the context of public sector services. The robot developers and the municipal managers wrongly assumed that a large number of the clients receiving assistance when using the toilet shared a similar need for a robotic solution. This blindsided the development team as the potential user segment(s) was (were) not adequately specified nor validated in real-world settings to help guide the development process. This stresses the importance of carefully defining and validating the user segment in practice as a basis for project scoping.
4.
Establish service-centered project scoping. In the BathroomBot case, we observed how the use contexts made it necessary to take into account the entire process of the service as opposed to just addressing a subtask. Rather than framing the problem in terms of a specific task (e.g., dressing and undressing the client), the scope needs to be centered on the service (e.g., enabling the client to use the bathroom without assistance from care workers) embedded in a sociotechnical system of practical, material, emotional, and social components and processes (Oborn et al., 2011; Van Aerschot & Parviainen, 2020). This will require a system thinking approach (MacLachlan & Scherer, 2018) in which robotic innovations, users, and service providers are considered interdependent parts forming a unified whole (Lusch & Nambisan, 2015). This shift in scoping the innovation project can provide a more realistic understanding of why and where the initiative might fail and support a more cohesive conceptual design process to create a robotic solution that satisfies an actual need.
5.
Critically assess whether the robot can replace existing work practices. According to Moore (2014), developers need to position their product in relation to existing solutions to allow pragmatists to make comparative evaluations and inform their decision to adopt the technology. The BathroomBot case shows the pitfalls of creating a business case based on automating an isolated subtask instead of the service as a whole. If the robot does not make the user independent for the entire service or if the current service is provided in bundles with other services (e.g., eating, bathing, cleaning), then the robot will not result in fewer visits, resulting in no economic savings. By focusing on an isolated task and overestimating a vaguely defined user segment, the initial business case was indeed too good to be true. While this added to the hype (Ramiller, 2006) and early support for BathroomBot as a disruptive technology that improved the clients’ quality of life and was a good investment for the municipalities, it also created unrealistic expectations among consultants, politicians, and investors who neglected to critically assess vital assumptions in the innovation idea.

While Moore (2014) proposes the use of alliances and partnerships to accelerate the formation of a whole product infrastructure, the competing logic-oriented explanations in our analysis suggest the need for new ways to organize long-term collaboration between robotic companies and public sector organizations to overcome lengthy sales and implementation processes, prevent gatekeeper bottlenecks, and better allocate resources for continuous development. This points us to the following lessons for managing the collaboration among different stakeholders in the diffusion process:

6.
Align expectations between private developers and public sector organizations. Did the municipalities buy a complete product or engage in a development project? This question is unclear to many of the stakeholders involved in the BathroomBot project. What the robot company interpreted as making the first of many future sales and being on the way to widespread diffusion was, from the perspective of the municipalities, the beginning of long-term collaboration for further testing and development. By not committing to utilize the inputs of the visionary early adopting municipalities to further develop the robot, the company became unable to cross the chasm to the pragmatic majority. The story of BathroomBot highlights the critical importance of listening carefully to the expectations and concerns of all parties involved before moving forward in the diffusion process (Greenhalgh et al., 2012). Additionally, sustaining long-term development, implementation, and distribution processes requires ongoing mobilization of resources. This may be aided by partnering with an external distributor to sell and support the robot as part of its existing product portfolio (Ward et al., 2017), while the participating municipalities cover the costs of implementation (e.g., reimbursing the company for training). This might help keep the developing company afloat while the diffusion chasm is navigated.
7.
Consider potential conflict of interests in the distribution channel. Crossing the diffusion chasm of robot technologies in the public sector requires both centralized decision making (managerial support, mobilization, and resource allocation) and decentralized efforts for mobilizing users and adapting the solution to specific use contexts. In the BathroomBot case, care workers operated as gatekeepers for identifying and training potential users of the robot. As pointed out in existing literature on robot diffusion and demonstrated in this study, whether the care workers will be champions or showstoppers depends on how they perceive the robot (Frennert et al., 2020). However, as also shown in this study, creating a positive perception of robotics is no easy task as such an approach would need to take into account that different stakeholders (i.e., developers, managers, staff, and clients and their families) can have different views on, expectations of, and interpretations of the same robotic technology (Čaić et al., 2018; Nielsen et al., 2016).

5.2 Boundary conditions

Notwithstanding these lessons learned, our study has two limitations in particular. First, while the interviewees represent the diverse group of stakeholders involved in the project (developers, company investor, external distributor, and municipalities), perspectives from the adopting municipalities are primarily reflected through project managers in charge of testing, implementing, and evaluating the robot. As different stakeholders can have widely varying views, expectations, and interpretations of the same robotic technology (Mettler et al., 2017; Nielsen et al., 2016), including clients’ perspectives could provide even more insights into the issues related to the BathroomBot’s failure.

Second, our findings are based on a single case study. Future research could benefit from multiple and comparative case studies to further explore the challenges involved in crossing the diffusion chasm for robot innovation projects in the public sector. Although the results were derived from a single robot project in Danish eldercare, we believe that they have relevance for other digital innovation projects aiming to transform the delivery of public sector services. However, we may speculate whether the robotic nature of the innovation we have studied has made a difference for the lessons learned in this study, and to what extent our findings would be relevant for other types of public sector digitalization initiatives. Service robots take up a physical and highly visible presence (Wirtz et al., 2018). Diffusing such “digital-physical information systems” (Sprenger & Mettler, 2015, p. 271) in clients’ private residences involves new challenges related to aesthetics (Salvini et al., 2010) and interactions with physical and human surroundings (Sparrow & Sparrow, 2006) – emphasizing the importance of ongoing user involvement in real-world settings throughout the innovation process, as discussed in the lessons learned.

6. Conclusion

In this article, we have examined why a seemingly successful robot innovation failed to be diffused on a larger scale. While cutting-edge robotic innovations can generate hype that can form a “defensive shell” to endure early criticism (Ramiller, 2006, p. 5), they also involve high risks of diffusion failure when confronting the pragmatic early majority (Moore, 2014). Our study demonstrates how the challenges in diffusing robots in public sector services involve context-specific aspects regarding the technology’s fit to the use context, the project scoping, and the alignment of organizational processes between private developers and adopters in the public sector. Based on these three reasons for failure, we have identified key lessons to help navigate the diffusion chasm in robotic innovation projects in the public sector. For future research, we recommend that scholars pay more attention to what can be learned from digitalization failures – not only success cases – because mistakes and failed attempts are inherent parts of the innovation trajectory that needs to be more amply examined.

Footnotes

Authors biographies

Jon Aaen is PhD Fellow at Aalborg University, Denmark. His research focuses on emerging technologies and digital innovation processes in public sector organizations. He has published on these topics in journals such as European Journal of Information Systems and Scandinavian Journal of Information Systems.

Jeppe Agger Nielsen is Professor at Aalborg University, Denmark. His main research interests concern digital innovation and digital transformation from an institutional theory perspective. Jeppe Agger Nielsen’s research is published in leading journals such as MIS Quarterly, Organization Studies, International Journal of Management Reviews, British Journal of Management, and Government Information Quarterly.

Appendix A: directly referenced documentary data*

Id	Name	Year	Source
D1	News item: New welfare technology makes clients independent	2015	Media search (infomedia.dk)
D2	News item: Brave new world – The robot that undresses you	2018	Media search (infomedia.dk)
D3	News item: Municipalities fail to exploit the business potential in welfare technology	2017	Web searches on “BathroomBot”
D4	News item: Free choice for all? Welfare technological challenges in municipal eldercare	2016	Chain referral
D5	News item: [BathroomBot Company] – Bankruptcy	2019	Web searches on “BathroomBot”
D6	News item: Pull your pants down with BathroomBot	2015	Web searches on “BathroomBot”
D7	News item: Take disruption seriously and maintain the quality in your core services	2017	Chain referral
D8	News item: BathroomBot provides more freedom	2015	Media search (infomedia.dk)
D9	News item: BathroomBot pulls your pants down	2015	Media search (infomedia.dk)
D10	News item: The clients are the least skeptical	2015	Media search (infomedia.dk)
D11	News item: Main attractions for culture event – Recommendations from five celebrities	2016	Media search (infomedia.dk)
D12	News item: The robots are coming	2017	Media search (infomedia.dk)
D13	News item: These technologies will take care of us: Nursing homes new staff	2017	Media search (infomedia.dk)
D14	Report: [Municipality#1] Yearly report, 2015	2016	Searches on participating municipalities’ websites
D15	Report: [Municipality #4] Vision and strategy for digitalization and welfare technology in the eldercare department 2016–2020 (Appendix 3 – Description of projects in 2017)	2017	Searches on participating municipalities’ websites
D16	Regional Government Report: Ecosystem analysis of the robot technology market in the region	2018	Web searches on “BathroomBot
D17	News item: Debating business development and EU funds for local companies	2018	Media search (infomedia.dk)
D18	Website: This year’s largest international event in the healthcare industry – [BathroomBot] is coming!	2016	Web archive (developers’ webpage)
D19	News item: Assistive Robots Showcased at RoboBusiness Europe	2016	Web searches on “BathroomBot”
D20	News item: Meet BathroomBot – A robot that puts on and take off your trousers for toilet visits	2016	Web searches on “BathroomBot”
D21	Website: The nominees for the CareWare Entrepreneurship Award and the Idea Award have been found	2016	Web searches on “BathroomBot”
D22	Website: BathroomBot - Nominated For The Careware Entrepreneur Award 2016	2016	Web searches on “BathroomBot”
D23	Report: Product evaluation of BathroomBot] in [Municipality #1]	2017	Chain referral
D24	Website: New assistive technology for home care clients soon to come	2016	Web archive (developers’ webpage)
D25	Website: Patent – Assistive Device For Disabled Persons	2017	Web searches on “BathroomBot”
D27	News item: BathroomBot helps take off the pants	2015	Media search (infomedia.dk)
D28	News item: Robot helps municipal clients on the toilet	2015	Web searches on “BathroomBot”
D29	News item: The robots are coming – I can assist seven clients instead of one	2017	Chain referral
D30	News item: This is how we get public digitalization that works	2017	Media search (infomedia.dk)
D31	Report: [Municipality #3] Project description	2017	Chain referral

*All documents translated from Danish to English to preserve anonymity.

References

Aaen

Nielsen

J. A.

, & Elmholdt

K. T.

(2018). 10 år med velfærdsteknologi: Strategi og praksis [10 years with welfare technology: Strategy and practice]. Samfundslederskab i Skandinavien, 33(3), 236-259.

Beane

(2019). Shadow learning: Building robotic surgical skill when approved means fail. Administrative Science Quarterly, 64(1), 87-123.

Blackman

(2013). Care robots for the supermarket shelf: A product gap in assistive technologies. Ageing and Society, 33(5), 763-781.

Čaić

Odekerken-Schröder

, & Mahr

(2018). Service robots: Value co-creation and co-destruction in elderly care networks. Journal of Service Management, 29(2), 178-205.

Cho

Mathiassen

, & Gallivan

(2009), Crossing the diffusion chasm: From invention to penetration of a telehealth innovation. Information Technology & People , 22(4), 351-366.

Clifton

Díaz Fuentes

, & Llamosas García

(2020). ICT-enabled co-production of public services: Barriers and enablers. A systematic review. Information Polity, 25(1), 25-48.

Compagni

Mele

, & Ravasi

(2015). How early implementations influence later adoptions of innovation: Social positioning and skill reproduction in the diffusion of robotic surgery. Academy of Management Journal, 58(1), 242-278.

Cresswell

Cunningham-Burley

, & Sheikh

(2018). Health care robotics: Qualitative exploration of key challenges and future directions. Journal of Medical Internet Research , 20(7), e10410.

Danziger

J. N.

Dutton

W. H.

Kling

, & Kraemer

K. L.

(1982). Computers and politics: High technology in American local governments. Columbia University Press.

10.

Donetto

Pierri

Tsianakas

, & Robert

(2015). Experience-based co-design and healthcare improvement: Realizing participatory design in the public sector. The Design Journal, 18(2), 227-248.

11.

Dunleavy

Margetts

Bastow

, & Tinkler

(2006). New public management is dead: Long live digital-era governance. Journal of Public Administration Research and Theory, 16(3), 467-494.

12.

Engelberger

(2000). A day in the life of Isaac. Industrial Robot, 27(3), 176-180.

13.

Fichman

R. G.

(2000). The diffusion and assimilation of information technology innovations. In Zmud

(Ed), Framing the domains of IT management: Projecting the future through the past, Pinnaflex Educational Resources inc, pp. 105-128.

14.

Fountain

J. E.

(2004). Building the virtual state: Information technology and institutional change. Brookings Institution Press.

15.

Frennert

(2019). Lost in digitalization? Municipality employment of welfare technologies. Disability and Rehabilitation: Assistive Technology, 14(6), 635-642.

16.

Frennert

Aminoff

, & Östlund

(2020). Technological frames and care robots in eldercare. International Journal of Social Robotics. doi: 10.1007/s12369-020-00641-0.

17.

Gioia

Corley

, & Hamilton

(2012). Seeking qualitative rigor in inductive research: Notes on the Gioia methodology. Organizational Research Methods, 16(1), 15-31.

18.

Glende

Conrad

Krezdorn

Klemcke

, & Krätzel

(2016). Increasing the acceptance of assistive robots for older people through marketing strategies based on stakeholder needs. International Journal of Social Robotics, 8(3), 355-369.

19.

Greenhalgh

Procter

Wherton

Sugarhood

, & Shaw

(2012). The organising vision for telehealth and telecare: Discourse analysis. BMJ Open, 2(4), Article e001574.

20.

Greenhalgh

Wherton

Papoutsi

Lynch

Hughes

A’Court

Hinder

Fahy

Procter

Shaw

(2017). Beyond adoption: A new framework for theorizing and evaluating nonadoption, abandonment, and challenges to the scale-up, spread, and sustainability of health and care technologies. Journal of Medical Internet Research, 19(11), e367. doi: 10.2196/jmir.8775.

21.

Grönvall

, & Kyng

(2013). On participatory design of home-based healthcare. Cognition, Technology & Work, 15(4), 389-401.

22.

Krishnan

Mintz

Foreman

Hodge

J. C.

, & Krishnan

(2019). The acceptance and adoption of transoral robotic surgery in Australia and New Zealand. Journal of Robotic Surgery, 13(2), 301-307.

23.

Local Government Denmark. (2021). Det velfærdsteknologiske landkort [Welfare Technology Database]. https://www.kl.dk/kommunale-opgaver/velfaerdsteknologi/det-velfaerdsteknologiske-landkort.

24.

Lo Iacono

Symonds

, & Brown

D. H. K.

(2016). Skype as a tool for qualitative research interviews. Sociological Research Online, 21(2), 103-117.

25.

Lusch

, & Nambisan

(2015). Service innovation: A service-dominant logic perspective. MIS Quarterly, 39(1), 155-176.

26.

MacLachlan

, & Scherer

(2018). Systems thinking for assistive technology: A commentary on the GREAT summit. Disability and Rehabilitation: Assistive Technology, 13(5), 492-496.

27.

March

J. G.

Sproull

L. S.

, & Tamuz

(1991). Learning from samples of one or fewer. Organization Science, 2(1), 1-13.

28.

Martinez-Martin

, & del Pobil

A. P.

(2018). Personal robot assistants for elderly care: An overview. In Costa

Julian

, & Novais

(Eds.), Personal assistants: Emerging computational technologies. Springer, pp. 77-91.

29.

Meijer

(2015). E-governance innovation: Barriers and strategies. Government Information Quarterly, 32(2), 198-206.

30.

Mergel

Edelmann

, & Haug

(2019). Defining digital transformation: Results from expert interviews. Government Information Quarterly, 36(4), 101385.

31.

Mettler

Sprenger

, & Winter

(2017). Service robots in hospitals: New perspectives on niche evolution and technology affordances. European Journal of Information Systems, 26(5), 451-468.

32.

Miles

M. B.

, & Huberman

A. M.

(1994). Qualitative data analysis: An expanded sourcebook (2nd ed.). Sage.

33.

Moore

G. A.

(2014). Crossing the chasm: Marketing and selling technology products to mainstream customers, (3rd ed.). Harper Business.

34.

Nielsen

J. A.

, & Andersen

J. G.

(2006). Hjemmehjælp mellem myter og virkelighed. [Homecare between myths and reality]. Syddansk Universitetsforlag.

35.

Nielsen

J. A.

Andersen

K. N.

, & Sigh

(2016). Robots conquering local government services: A case study of eldercare in Denmark. Information Polity, 21(2), 139-151.

36.

Nielsen

J. A.

Mathiassen

, & Newell

(2014). Theorization and translation in information technology institutionalization: Evidence from Danish home care. MIS Quarterly, 38(1), 65-86.

37.

Oborn

Barrett

, & Darzi

(2011). Robots and service innovation in health care. Journal of Health Services Research and Policy, 16(1), 46-50.

38.

Papadopoulos

Koulouglioti

Lazzarino

, & Sheila

(2020). Enablers and barriers to the implementation of socially assistive humanoid robots in health and social care: A systematic review. BMJ Open, 10(1), e033096.

39.

Ramiller

N. C.

(2006). HYPE! Toward a theory of exaggeration in information technology innovation. Academy of Management Proceedings 2006, 1, A1-A6. doi: 10.5465/ambpp.2006.27169062.

40.

Robinson

MacDonald

, & Broadbent

(2014). The role of healthcare robots for older people at home: A review. International Journal of Social Robotics, 6(4), 575-591.

41.

Rogers

E. M.

(2003). Diffusion of innovations. Free Press.

42.

Salvini

Laschi

, & Dario

(2010). Design for acceptability: Improving robots’ coexistence in human society. International Journal of Social Robotics, 2(4), 451-460.

43.

Sharkey

, & Sharkey

(2012). Granny and the robots: Ethical issues in robot care for the elderly. Ethics and Information Technology, 14(1), 27-40.

44.

Sparrow

, & Sparrow

(2006). In the hands of machines? The future of aged care. Minds and Machines, 16(2), 141-161.

45.

Sprenger

, & Mettler

(2015). Service robots. Business & Information Systems Engineering, 57(4), 271-274.

46.

Van Aerschot

, & Parviainen

(2020). Robots responding to care needs? A multitasking care robot pursued for 25 years, available products offer simple entertainment and instrumental assistance. Ethics and Information Technology, 22(3), 247-256.

47.

Van de Ven

(2007). Engaged scholarship: A guide for organizational and social research. Oxford University Press.

48.

Walsham

(1995). Interpretive case studies in IS research: Nature and method. European Journal of Information Systems, 4(2), 74-81.

49.

Ward

Fielden

Muir

Holliday

, & Urwin

(2017). Developing the assistive technology consumer market for people aged 50-70. Ageing and Society, 37(5), 1050-1067.

50.

West

D. M.

(2004). E-government and the transformation of service delivery and citizen attitudes. Public Administration Review, 64(1), 15-27.

51.

Wirtz

Patterson

P. G.

Kunz

W. H.

Gruber

V. N.

Paluch

, & Martins

(2018). Brave new world: service robots in the frontline. Journal of Service Management , 29(5), 907-931.

Lost in the diffusion chasm: Lessons learned from a failed robot project in the public sector

Abstract

Keywords

1. Introduction

1 Following Fichman (2000), we define diffusion as “the process by which a digital innovation spreads across a population of organizations” (p. 1). This differs from the concept of adoption, which concentrates on the uptake of a digital innovation by a single organization (or individual).

2.1 Robots in public sector service delivery: Diffusion challenges

2.2 The diffusion chasm

3. Method

3.1 Research setting and case

Table 2 Summary of data sources

Table 3 Data analysis

4.1 Setting the scene: A daydream turning into a nightmare

4.2 Technology-oriented explanations for failure

4.3 Scope-oriented explanations for failure

4.4 Competing logic-oriented explanations for failure

5. Discussion

5.1 Lessons learned for crossing the diffusion chasm

6. Conclusion

Footnotes

Authors biographies

Appendix A: directly referenced documentary data*

References

¹
Following Fichman (2000), we define diffusion as “the process by which a digital innovation spreads across a population of organizations” (p. 1). This differs from the concept of adoption, which concentrates on the uptake of a digital innovation by a single organization (or individual).

Table 2
Summary of data sources

Table 3
Data analysis