The role of adaptation in mobile technology innovation for the water, sanitation and hygiene sector

Information and communication technologies (ICT) are central in helping to meet the needs of the world's poorest (Kalil 2009), and in contributing to economic growth, sustainable livelihoods and greater freedom (Heeks 2010). Indeed, ICT is reported to change the way international development itself takes place, although this is yet to be fully understood (Smith et al. 2011). The importance is reflected in the high proportion of incomes devoted to this service (de Angoitia & Ramirez 2009). Making the benefits of new ICTs available to all was included in the Millennium Development Goals (Goal 8, Target 18), is among the Sustainability Development Goals (Goal 9, Target 9c), and was the focus of the 2016 World Development Report: Internet for Development (World Bank Group 2016).

Mobile-enhanced technologies are frequently cited as important economic development tools (Donner & Escobari 2010; Wheeler 2011). In fact, the ‘explosive growth’ in access to and use of mobile phones at the ‘bottom of the pyramid’ is described as the dominant change in ICT for development in recent years (Spence & Smith 2010). Average ownership of mobile phones has reached 40 to 50% in developing countries (GSMA Intelligence 2015). South Asia and Africa have experienced the greatest rise in ownership, with user growth at nearly 20% a year since 2007 (GSMA Intelligence 2015). This surge in the accessibility and affordability of mobile phones is transforming the way people share information, and consequently how public services are delivered.

While the growing availability of mobile phones in developing countries has commanded the attention of the development community (Heeks 2008; Unwin 2009; Maumbe & Okello 2010), an estimated 844 million people still have no access to clean water and 2.3 billion to adequate sanitation (WHO/UNICEF 2017). This gap remains despite steady strides to increase access to adequate water, sanitation and hygiene (WASH) services (Figure 1). Water providers, governments and non-governmental organizations (NGOs) around the world are facing increased pressure to improve their operational efficiency, customer service, and/or monitoring and evaluation. With access to mobile phones now exceeding access to improved WASH services globally, many in the WASH sector have begun experimenting with ICT application, including mobile devices, to transform existing operations (Shemie et al. 2012).

ICT-enhanced technology increasingly improves water management (ITU-T Technology Watch 2010; Gourbesville 2011; Hutchings et al. 2012). For example, automated sensors can provide near real-time data on water supply assets (International Telecommunications Union 2011; Thomson et al. 2012) and agricultural management (Panchard et al. 2007). Water providers and regulatory agencies use ICT to engage with the public – e.g., by providing data on their websites, using social media, and offering online and mobile payment options. A review of ICT applications in water management is beyond the scope of this paper and can be found elsewhere (ITU-T Technology Watch 2010).

This paper is focused on mobile data-sharing applications in the water, sanitation and hygiene sectors (mWASH) that transfer information using ‘ordinary phone’ technology, such as SMS, or ‘smartphones’ equipped with GPS that allow survey data to be collected and transmitted to a central database. mWASH applications seem likely to change the way information flows between decision-makers with respect to water use and metering, tariff changes and payments, feedback and complaints, service changes and disruptions, operational efficiency and performance, among others. Such flow has historically been ‘vertical’ in the water sector (i.e. between utilities and customers, etc.). Mobile data-sharing applications may both enhance these flows and/or enable ‘horizontal’ information (i.e. between utilities or customers).

The growing availability of mobile phones also offers a new information pathway to and from end-users. Proliferation of mWASH applications could help address the increased demand for information about water by the poorest people (Hutchings et al. 2012), with the potential to transform existing social relations. A recent report noted: ‘[w]ithin governance, mobile technologies can offer new means for empowering citizens and stakeholders by opening and enhancing democratic processes and mechanisms’ (UNDP 2012). Accordingly, mWASH applications could improve transparency, accountability, and participation (Hellström & Jacobson 2014; Mongi et al. 2015). Some caution is needed, however, as not all mWASH applications will succeed. Several initiatives suffer from the ‘pilot-then-die’ syndrome, and fail to survive beyond rapid prototyping. Moreover, ICT use can exacerbate social divides (Clarke et al. 2013).

This case study analyzes the barriers to successful deployment and uptake of nine mWASH technologies using the Framework for Analyzing a Multi-level Innovation System (FAMIS) conceptual model. The FAMIS model is derived primarily from innovation systems theory (Anadon et al. 2014).

This case study draws on various bodies of literature (innovation systems, WASH sector, and ICT for development), interviews with stakeholders, understandings gained from workshops, and field experience. This is a new field, so much of the literature consisted of reports, workshop proceedings, websites, project performance evaluations, and other online resources, as well as recent, peer-reviewed publications.

Nine mWASH applications, predominantly from Africa and India, were evaluated (Table 1). Open-ended, semi-structured, telephone or face-to-face interviews were conducted with representatives from government, NGOs and academia concerning eight applications. In one case (Maji Matone), only secondary sources – videos, blogs, and articles – could be used. Interviews took place between April and September 2013, with some follow-up in June 2015. One author also has first-hand experience with mWASH technology in the field.

FAMIS was used to analyze the origin of demand for mWASH applications, the cause(s) of mixed results in sustained use, and potential solutions for overcoming barriers.

FAMIS is a diagnostic tool for identifying technological innovation barriers so that appropriate interventions can be designed. It is grounded in innovation systems theory, which seeks to understand stakeholder¹/institutional interactions with technology in developing, promoting and using innovations (Edquist 2005; Markard & Truffer 2008; Coenen & Díaz López 2010). Drawing also on the basic ‘stock and flow’ concept from systems dynamics models (Meadows 2008), the innovation system is developed in FAMIS as a set of seven, non-linear flows or stages through which a technology moves – invention, selection, initial adoption, production, widespread use, adaptation, and retirement. These represent mechanism clusters that impede or enhance innovation for that technology as it moves between innovation stocks of knowledge, inventions, feasible technologies, and technologies in limited or widespread production/use. As with any dynamic system, mechanisms may coincide, happen sequentially, and/or be linked via complex feedback loops (Edquist & Johnson 1997; Lundvall 2010). Applying FAMIS to a technology aids in understanding when and why it succeeds or fails, and how stakeholders and institutions can be targeted for intervention. The framework's stocks, flows, and mechanisms are shown in Figure 2.

The rise in mWASH applications has largely been driven by NGOs seeking to improve their operations and expand access to WASH services, following a trend started in other sectors – e.g., health (Murthy et al. 2013). International donors and multilateral government organizations increasingly seek greater accountability, such as through monitoring and evaluation (M&E), which has fostered interest in mWASH. Academic research hubs have incubated and studied mWASH applications, while technology companies have provided expert support. Government and water providers have also begun to incorporate mWASH as part of a push towards ‘e-government.’

The term institutions refers to the informal and formal rules that govern interactions in society; they include laws and also non-binding social norms that shape behavior (North 1990; March & Olsen 2004). The technology revolution since the late 1990s has seen a global shift in institutional culture. In addition, the MDGs, and now the SDGs, have heightened the drive to monitor and evaluate development programs (Winkler et al. 2014), fostering, in turn, more interest in ICT use in the WASH sector. Finally, as nations struggle to achieve water security and realize the right to safe drinking water and sanitation, the WASH sector has received heightened international attention.

Determining whether the insights gained from one type of application could be generalized proved difficult without a framework enabling comparison of mWASH applications. For example, an app developed by an NGO to improve its own monitoring and evaluation is distinct from one designed to help utilities improve their efficiency or to enable individuals to report broken pumps or wells. Not only were the stakeholders different, but the data source(s) and collection methods also varied. As Rhodes demonstrated (2009), distinguishing actors within a network can help unpack ICT implementation effects. Accordingly, to provide a tool for assessing dissimilar mWASH applications effectively, a novel typology was developed to characterize the actors involved and map information flow between them. While it was developed with WASH in mind, the typology may be capable of generalization to other areas. Three definitions are relevant:

• Resources-Allocator means organizations that fund WASH projects, including government agencies, multilateral government organizations, private foundations or companies. Information collected through an mWASH application helps a resource-allocator decide where to invest financial and human resources.

• Service-Provider means water utilities and, for underserved populations, includes NGOs and non-utility water providers – both formal and informal local private providers – commonly known as ‘water vendors’. Service-providers may use mWASH applications to optimize operations and track customer service. Operational efficiency and customer service are key goals for applications intended to aid service-providers.

• End-User means individual water users (including those unserved or informally served) or utility customers, individuals or businesses. End-users use mWASH to access and share water service information. Reliable and affordable access to water and sanitation is the prime purpose of apps designed to help end-users.

In addition to categorizing stakeholders, an effort was also made to understand information flows between them. Figure 3 shows how information can flow vertically between different stakeholder categories and horizontally between users. Understanding these information flow-pathways was crucial to identifying why some mWASH strategies were successful and others were not.

This case study draws on nine different mobile applications used predominantly in Africa and India. The applications are in different stages of development, and have varying goals and target users – see Table 2.

This case study was developed by examining the movement of mWASH applications through the seven FAMIS stages. Use of these stages is intended to provide a common language and structure to discuss mechanisms and barriers faced by different technologies. It does not imply subscription to a linear view of innovation. Based on the collective results from the mWASH applications described in Table 2, the less important stages in mWASH innovation dynamics have dotted outlines in Figure 2. The more relevant stages are colored and critical mechanisms (see below) are highlighted in bold.

The invention stage involves developing the means to meet a need using existing knowledge via mechanisms linking the knowledge and invention stocks, or amplifying the invention stocks from within (Anadon et al. 2014).

WASH organizations frequently hire software development firms to develop applications through a goal-oriented search. Invention barriers may occur because there are limited opportunities to bring together software developers and WASH practitioners. However, a number of fora are being created – e.g., WASH-themed hackathon events, typically lasting several days, in which a large number of people meet to engage in collaborative computer programming – to develop useful prototypes. There are some success stories from hackathons – e.g., Maji Voice and mWater – but most prototypes generated at hackathons are never adopted. Fixation on unique technology and low-cost prototype-invention (relative to adoption) may lead to ‘re-invention’. The study suggests that the invention stage is not a major chokepoint in innovation globally. Rather, a number of open-source applications are now available for the WASH sector, which need effective adaptation.

Selection involves surveying the options and choosing which technologies to promote or invest in, thereby moving a technology from the invention to the feasible technology stock (Anadon et al. 2014).

The selection stage presents no significant barrier to innovation for mWASH applications. The growing emphasis on monitoring and evaluation, as well as the increased availability and accessibility of mobile technology, has driven the interest in mWASH. Traditional development agencies and technology donors increasingly support such endeavors with seed-funding. The most relevant mechanism at this stage is selection by users. If a resource-allocator is driving the adoption of a technology by WASH service-providers or end-users, the mechanism of selection by agents on behalf of users is also applicable.

The production stage allows feasible technologies (tangible and intangible) to move into the stocks of technologies in limited or widespread production or use (Anadon et al. 2014). For mWASH technologies, this stage overlaps with initial adoption and widespread use. Initial adoption – the first use of a technology by a subset of potential users in a particular context – moves a feasible technology into the stock of technologies in limited production and use. The widespread use stage includes the processes by which technologies move from limited use to wide-scale adoption by end-users. Mechanisms that allow the initial-adoption and the widespread-use flows encompass all aspects of technology access, dissemination, and uptake.

Adoption (including production and widespread use) presents many critical barriers to the effective deployment of mWASH technologies. The mechanism of access to information can help or hinder: while it can improve efficiency and transparency, it can also threaten the existing situation and measures of success. For example, WASH service-providers may not want to know that their programming has not been as effective as they had hoped, if they do not have the resources or know-how to address the problem, or if they are afraid that the news will impact their political clout or funding streams. In such situations, the mechanism of benefits to early adopters may also be blocked.

The mechanism of behavior and culture can also impede mWASH application adoption. WASH service-providers and resource-allocators may have policies that directly or inadvertently reward the existing situation, instead of risk-taking on new ideas and projects. End-users may be reluctant to provide information – e.g., via a crowd-sourced platform – because they are unfamiliar with the technology, are illiterate or unable to speak the given language, or do not believe that participation will improve their situation directly.

The relative prices of investing and/or participating in an mWASH system can inhibit adoption. For example, while the cost of a phone call or an SMS message may be small, it could still present a barrier for end-users. Similarly, economic barriers can exist for WASH service-providers, even for open-source tools that can require significant adaptation. Lack of physical infrastructure and effective supply chains are also challenges because mobile devices depend on electricity and network connectivity.

Adaptation involves adjusting a technology for use in contexts that differ from those for which it was initially invented, selected, or adopted (Anadon et al. 2014). Flow through adaptation is characterized by the intentional redesign mechanism, because the initial design almost always needs adaptation to overcome ‘local’ adoption barrier(s) to reach scale.

Intentional redesign poses significant challenges. Open-source applications need to be adapted to fit the specific needs of an organization and/or to collect data in a way that can be integrated with existing systems. For example, cultural and social adaptation may be critical to address language and literacy barriers. Geographic and spatial adaptation is also important, especially when considering mWASH deployment in areas with limited basic infrastructure, like electricity and network connectivity. mWASH technologies often need continuous adaptation as part of a ‘trial and error’ process, which can run against the existing culture and budgets of WASH organizations.

As mWASH is a new area, it is too soon to discuss the retirement stage.²

The most significant barriers to mWASH exist at the adoption and adaptation stages.

An institutional culture of openness and accountability is a major factor in the success of mWASH applications. This relates not simply to willingness to try something new but also to a new way of doing business, because organizations must be prepared for greater transparency in information. For example, wanting a better understanding of the impact of its interventions, Water for People committed to a 10-year monitoring campaign. Looking for more efficient data collection and processing, and working from changes in the health sector, they developed FLOW – a pioneering use of mobile data-sharing applications in the WASH sector. Similarly, the creation of Maji Matone and Maji Voice can be traced in part to reforms in Kenya's water sector that required greater transparency and accountability among service providers (Kenya Water for Health Organization 2009).

mWASH applications are more likely to be adopted by organizations with a demonstrated interest in reform and where a ‘local champion’ exists. The success of the social enterprise company NextDrop in Hubli, in Karnataka, India, where it was first piloted, can be attributed partly to the unique circumstances there. First, Hubli is located in a state that had launched an e-governance system. Second, the municipality had already participated in a World Bank project for round-the-clock water supplies. Third, recent reforms had transferred control of the water utility from local to state level to improve operations (Center for Science and Environment 2012). Finally, NextDrop received enthusiastic support from the water utility's Executive Engineer, who helped facilitate approvals from other government agencies.

Discrepancies in WASH-related statistics can change data collection methods, fostering uptake of new technology. In collaboration with the Indian government, the Water and Sanitation Program of the World Bank developed Outcome Tracker to monitor household sanitation behavior, rather than use proxy inputs, such as money spent – and/or outputs, such as number of toilets built. A much-publicized gap in national open defecation statistics helped overcome adoption barriers, and encourage state and district level agencies to adopt Outcome Tracker.³ As one interviewee said, the statistical discrepancy changed the ‘political economy of monitoring’ in India, which was facilitated by the mobile tool.

Donor and institutional support for ‘trial and error’ is critical to the success of mWASH applications. Part of FLOW's early success can be attributed to the financial and moral support of technology donors who understood that failure is part of innovation. In contrast, multilateral government organizations that traditionally fund development projects generally operate on a slower time-frame and seek a high degree of certainty regarding project success.

Universities also play a pivotal role by incubating new technologies. For example, NextDrop was developed by graduate students at University of California, Berkeley, who created the initial prototypes as part of a class project.

Developing an effective proof of concept is critical for going to scale and achieving public client support. Beta-testing and pilot projects yield critical insights into adoption barriers, and the tools and related practices can be adapted accordingly. The local context has implications for tool choice, e.g., consideration of technical accessibility, education and training, and government-citizen relationships.

By setting expectations appropriately, a distinct ‘pilot project period’ may help to keep the ‘window of adaptation’ open. For example, in late 2011, NextDrop secured permission from the state water utility and municipal corporation to begin a pilot project. The pilot was successful and moved seamlessly to full-scale rollout. However, a problem soon arose as the original Memorandum of Understanding required a particular user rate. The utility became reluctant to raise rates, even though NextDrop was confident that the market could bear them and they would help the social enterprise company become more financially sustainable. Because of the way the initial contract was written, a set of expectations was created at the water utility. As a small, agile organization, NextDrop has been able to develop innovations in its technology, but it has had less success adapting processes that require government approval.

Testing ‘proof of concept’ on a small-scale first can help to identify implementation challenges that could not be predicted. For example, in the Outcome Tracker pilot project evaluating rural sanitation outcomes in India, the Water and Sanitation Program realized that smartphones could neither be charged in rural areas that lacked electricity nor upload data immediately via the Internet. The team therefore adapted the application to deal with these infrastructure challenges. Similarly, mWater was adapted to function in the absence of network service after field-testing, while Transparent Chennai focused on developing a toilet-ranking tool that worked via SMS after recognizing that slum-dwellers did not have access to smartphones. Beta-testing a new product may be obvious to a technology company, but it is not necessarily consistent with how large organizations, like governments, operate.

A successful ‘proof of concept’ can also help to create demand for information. For example, the use of mobile devices to conduct surveys enabled the Water and Sanitation Program in India to collect a greater amount of detailed information from more households. However, for goal transparency to be achieved, a WASH resource-allocator or service-provider must try to ensure that the data collected are accessible to the public.⁴ Akvo FLOW and mWater have made substantial investments in data visualization and analysis functions for their respective public dashboards.

Cultural and economic barriers can discourage individuals from sharing information via mWASH applications. For example, Daraja developed an SMS-based application called Maji Matone to enable villagers in rural Tanzania to report water pump problems. Low expectations of government and the idea of ‘reporting on’ a service provider were major barriers. Moreover, customers had to assume the cost of sending an SMS. After the 6-month pilot, Daraja decided to suspend the program pending fundamental redesign. The next version is expected to abandon the crowd-sourcing approach in favor of a network of recruited, trained volunteers. Crowd-sourcing information may not necessarily be perceived as empowering or beneficial by end-users.

The ‘life-span’ of the information collected via an mWASH app may also influence whether crowd-sourcing from end-users is viable. When the NextDrop and mWater technologies were first conceived, they were envisaged as crowd-sourcing applications. Both organizations continue their direct engagement with citizens but have since shifted towards involving trusted information sources, like utility valve-men (NextDrop) and health workers (mWater). This was in large part due to a decision to build local monitoring capacity within WASH service-providers who already had responsibility for monitoring water systems.

Adapting mWASH applications to the existing communication channels of service-providers is crucially important. For example, Maji Voice's widespread adoption by utility staff can be attributed in part to the web-based task management software that enables them to receive, process and report on consumer-submitted complaints. To encourage broad uptake of Maji Voice, the technology was optimized for low-bandwidth. A recent, anonymous staff survey has shown a positive response to it, with the majority reporting that it ‘made it easier to deal with and follow up on specific complaints’ (93%) and ‘improved the way [the utility] deals with complaints’ (98.1%) (Hirn 2015).

Mobile data-sharing applications should also be understood as a means of enhancing, but not replacing, traditional forms of communication. When Nairobi's water provider allowed customers to make complaints using SMS via Maji Voice, it did not eliminate the ability to file a report in person. The majority of complaints still come from these ‘traditional’ channels.

For mWASH to be successful, an organization must monitor and adapt the technology over the long run. Although open-source applications can reduce some technical and economic barriers,⁵ most mWASH apps still need adaptation to meet specific needs. Many early adopters have unrealistic expectations regarding the costs and maintenance associated with mobile tools. Partnerships may be key. For example, Water for People recognized that FLOW would also be useful elsewhere, but did not have the expertise to support the technology. As a result, it partnered with a technology-oriented organization, Akvo, which had the ability and vision to roll out tools like FLOW to NGOs and governments that were working in water or other sectors worldwide.

Applying FAMIS to a sample of mWASH applications reveals several important factors. The most critical barriers occur at the adoption and adaptation stages, but greater attention and investment within the adaptation stage could help to overcome them. The most common blocking mechanisms include access to information, benefits to early adopters, behavior and culture, relative prices, and intentional redesign. Strategies for overcoming these include:

• An institutional culture that emphasizes openness to change and internal accountability – critical to overcoming mWASH application adoption barriers. An organization, whatever its form, must be willing not only to try a new technology but also to accept the consequences of having additional access to information.

• To meet the needs of resource-allocators, service-providers, and/or end-users appropriately, it is important to develop and learn from ‘proof of concept’ scale before going to full-scale. Attention to existing information flows between stakeholders is critical to the sustained use of any mWASH application. While the barriers to rapid-prototyping of mWASH applications are low, successful adoption and adaptation take time, support, and funding.

• An mWASH application will not be successful unless there is continuous effort to adapt it to local capacity and infrastructure. Embedding software design and technical support within the existing institutional context and/or developing effective partnerships are critical to success.

The theme that emerges from this analysis is that the actual technology may be less important than the way in which it is implemented. While much excitement centers on the technological tool, a critical point emphasized repeatedly in interviews and at workshops is that the technology is not a goal in itself; rather, it is a tool in service of another goal, like improved performance or customer service. Successfully implementing an mWASH project requires long-term commitment and financial resources to ensure continuous monitoring and adaptation (Pearce & Howman 2013). Resource-allocators should also prioritize investments in ‘large’ infrastructure, such as access to electricity and networks, which in turn can enhance uptake of ‘small’ infrastructure like mWASH. Indeed, the popularity gained by some mWASH applications is credited less to technological breakthroughs than strong local buy-in and institutional capacity. These products rarely resemble the prototypes, suggesting that emphasis on adaptation is critical to overcoming adoption barriers and promoting innovation.