Improving ICT4D projects with Agile software development Joost Dijkers Sietse Overbeek Sergio España Utrecht University Utrecht University Utrecht University Faculty of Science, Department of Faculty of Science, Department of Faculty of Science, Department of Information and Computing Sciences Information and Computing Sciences Information and Computing Sciences Utrecht, the Netherlands Utrecht, the Netherlands Utrecht, the Netherlands J.J.G.Dijkers@students.uu.nl S.J.Overbeek@uu.nl S.Espana@uu.nl ABSTRACT Agile Software Development (henceforth referred to as Agile) ICT4D seeks to bridge the digital divide in developing countries. is a methodology for developing software and was found to in- Important requirements of ICT4D projects are a demand-driven crease the success rate of ICT projects [5]. Agile is collaborative, approach and participation of the local community. The fact that incremental, and iterative [6]. Collaborative development means user collaboration is a principle of Agile software development that work is performed in teams rather than individually. For Agile, (Agile), triggers our interest on whether Agile practices can improve this also means that users should be included in the work process. ICT4D projects. This paper aims to investigate if and how Agile can Incremental development is a development approach in which the contribute to the success of ICT4D projects. In order to achieve this, system is developed in a series of small steps. Iterative develop- existing literature was consulted and an interview was held. This ment means that the development activities, such as requirements paper provides an overview of the critical success factors for ICT4D engineering and software testing, are performed cyclically rather projects and Agile, as well as of the advantages of Agile. Agile can than sequentially. Furthermore, Agile is adaptive, which means only work successfully when ICT4D projects are demand-driven, that rapid change is supported [7]. Agile practices are summarized and when both a cultural understanding and trust are built. Notable by Highsmith [7] as follows: ‘short iterations, continuous testing, ways in which Agile can improve ICT4D projects are by facilitating selforganizing teams, constant collaboration (. . . ), and frequent re- user collaboration, improving team communication, enhancing planning based on current reality’. organizational learning, and by frequently delivering software. However, agile methods that have harvested success in western countries cannot be directly applied in ICT4D projects [8]. There KEYWORDS exist multiple reasons for this. For example, increased user participa- tion has proven to be essential in order to achieve ICT adoption [9]. Agile, ICT4D, digital divide, user collaboration Additionally, inhabitants of poor communities in developing coun- ACM Reference Format: tries often have no ICT or project management skills [10]. Further- Joost Dijkers, Sietse Overbeek, and Sergio España. 2018. Improving ICT4D more, cultural barriers can limit or even prevent the cooperation projects with Agile software development. In Proceedings of ACM Conference of the local community [11, 12]. Research is thus necessary on to (Conference’17). ACM, New York, NY, USA, 8 pages. https://doi.org/10.1145/ what degree Agile methods are compatible with ICT4D projects. In nnnnnnn.nnnnnnn addition, research on the effects of using Agile methods in ICT4D projects is limited, as it is mainly focused on benefits of user collab- oration. For example, using an Agile method was reported to allow 1 INTRODUCTION developers to change the system in a natural way in response to un- The use of ICT in developing countries is the focus of an academic expressed requirements and changes in business environment [13]. field called information and communication technology for devel- Furthermore, Agile methods make ICT more demand-driven and opment, or ICT4D for short [1]. ICT4D is aimed at how the benefits improve the involvement of users [14, 15]. This increased user par- of ICT can be evenly divided between society to bridge the gap ticipation of Agile methods also allows requirements to be elicited between the rich and poor. For example, ICT can improve creating, and knowledge to be created [14, 16]. To address these two prob- sharing, and enhancing knowledge, make production and transac- lems, the following research question is formulated: tions more efficient and cost-effective, and stimulate networking RQ: To what degree can Agile software development improve ICT4D amongst parties (e.g. firms) [2]. However, high rates of failure exist projects? for ICT4D projects [3, 4]. The research question is answered by first performing a literature study on the critical success factors for ICT4D projects, the critical Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed success factors for Agile methods, and the advantages of Agile for profit or commercial advantage and that copies bear this notice and the full citation methods. Then, the results of the literate study are analyzed. on the first page. Copyrights for components of this work owned by others than ACM The structure of this paper is as follows. In section 2 related must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a work on Agile methods for ICT4D projects is discussed to find out fee. Request permissions from permissions@acm.org. why Agile is used in frameworks for ICT4D projects. In section Conference’17, July 2017, Washington, DC, USA 3 the method for arriving at an answer to the research question © 2018 Association for Computing Machinery. ACM ISBN 978-x-xxxx-xxxx-x/YY/MM. . . $15.00 is given. In section 4 the results from the literature study on the https://doi.org/10.1145/nnnnnnn.nnnnnnn critical success factors for ICT4D projects and Agile methods, and on the advantages of Agile methods are presented. In section 5 The benefits Agile methods can bring specifically to ICT4D these results are analyzed to determine how suitable Agile methods projects are discussed in order to understand how Agile can im- are for ICT4D projects and in what ways they can improve ICT4D prove ICT4D projects. Data regarding the advantages of using Agile projects. In section 6 the analysis is discussed and related to existing is gathered to answer this research sub-question. literature. Finally, in section 7 an answer is given to the research question and suggestions for future research are given. 3.2 Data gathering 2 RELATED WORK These sub questions are answered by using existing literature. Google Scholar was primarily used as search engine, as well as Bon, Akkermans, and Gordijn developed an ICT4D framework that IEEE Computer Society Digital Library and SpringerLink. Some is partially based on Agile [18]. Other inspirations of the framework common search terms that were used are: ‘Agile software develop- are: Living Labs, use case analysis, and requirements engineering. ment’, ‘Agile software development advantages’, ‘Agile software The framework is specifically designed to address several ICT4D development critical success factors’, ‘ICT4D’, ‘ICT4D critical suc- concerns, such as a lack of understanding of the local needs and the cess factor’, and ‘ICT4D Agile’. The forward snowballing technique context. The discussed benefits of Agile are that it fosters creativity, was used as well. personal commitment, and collaboration with the user. CSFs for ICT4D projects were in literature referred to with the Distributed Agile Methodology Addressing Technical Ictd in following terms: critical success factor, lesson learned, step (to Commercial Settings (DRAMATICS) is an Agile method for com- ensure sustainable development), and activity (that led to success). mercial ICT4D projects [19]. The discussed benefit of Agile is the For Agile, the following two terms were found and used to describe collaboration with users. CSFs: critical success factor, lesson learned. Advantages of using Speedplay is a framework for ICT4D projects which takes inspi- Agile were sometimes also called benefits. ration from Agile, Action Research, and Participatory Design [20]. Some of the inspirations from Agile are iterative development, flexi- bility, and collaborative development. The discussed benefit of Agile 3.3 Data analysis is the user collaboration. To answer the first research sub-question, the CSFs for Agile are The Nordic Model is a framework for ICT4D based on Nordic discussed in the context of ICT4D projects and by relating these socio-cultural background and shared values, and is described as to the CSFs for ICT4D projects, resulting in a series of steps that an Agile method [21]. The reasons for using an Agile method were need to be taken in setting up an ICT4D project before Agile can be frequent and immediate feedback from the users and informal com- applied. To answer the second research sub-question, a comparison munication to achieve equality and inclusion of all users. between the advantages of Agile and the CSFs for ICT4D projects These frameworks differ in terms of other inspirations (e.g. Par- was made. For each CSF for ICT4D projects it was determined if ticipatory Design for Speedplay) or application (e.g. business ICT Agile can improve the degree to which that CSF is satisfied. projects for DRAMATICS). However, all these frameworks have been tested in ICT4D projects with success. The success of Agile or Agile inspired frameworks suggests that Agile methods can be 3.4 Interview beneficial to ICT4D projects [22]. The primary reason for using In order to attain a greater insight into how Agile can improve Agile methods for ICT4D seems to be improved collaboration with ICT4D projects, an interview with ICT4D and Agile experts was the user. held. A semi-structured interview is a good choice when the pur- pose of the interview is to elicit a person’s viewpoint regarding a 3 METHOD specific matter [25]. In a semi-structured interview, there are prede- termined questions, but there is flexibility in asking these questions. 3.1 Research sub-questions For example, new questions can be added ad hoc. The results from This paper aims to answer the main research question: ‘To what the interview were used to validate the findings from the literature degree can Agile software development improve ICT4D projects?’. study, and are thus discussed in the Analysis section. It does so by answering the following research sub-questions (SQ): The interviewees are all part of the organization W4RA, of which SQ1: Can Agile methods successfully work in an ICT4D project? the name stands for the Web alliance for Regreening in Africa. On its website, W4RA gives its mission as follows [26]: ‘to support An analysis is performed to determine if Agile can work success- farmer-managed regreening activities specifically by enhancing fully in ICT4D projects. It is essential for Agile to work successfully information, communication, and knowledge sharing for rural de- in order for the ICT4D project to succeed [24]. In order to analyze velopment’. An example of an ICT4D project done by W4RA is this, critical success factors for ICT4D projects and Agile are gath- RadioMarché [27], which is a voice-based market information sys- ered. Critical success factors (CSF) are defined by Alias, Zawawi, tem that allows farmers to advertise their products to communi- Yusof, and Aris [23] as: ‘Inputs to project management practice ties in their local language. The interview was held with: prof. dr. which can lead directly or indirectly to project success’. As such, Akkermans, who is the director; ms. drs. Bon, who is the program CSFs give a good impression of where Agile methods might have manager; and with ms. drs. Tuijp, who is the communication officer. an important effect. When referring to their expertise, all three interviewees will be SQ2: How can ICT4D projects benefit from an Agile approach? collectively referred to as ‘the interviewees’. 2 4 RESULTS A network of local partnerships will allow for the participants to gain access to resources they might otherwise not have had access 4.1 Critical success factors for ICT4D projects to, such as skilled people or financial mechanisms [29, 31]. Monitor and evaluate project progress regularly (ICT4D- The creation of local content must be facilitated (ICT4D- CSF1) [10, 28, 29]. Monitoring and evaluating project progress CSF8) [10, 29, 30, 32, 33]. Local content is content being in local allows for the team members to measure the effects of ICT on language as well as having inspiration from local culture, created development [29]. The goal of evaluation should be to discern by locals [32]. The reason for the importance of local content is that changes in the welfare of the members of the local community, and only a select portion of the population will be able to understand to adapt the project accordingly. Evaluation should be an iterative content from, for example, The United Kingdom, due to language and adaptive process. Monitoring and evaluating also allows for and cultural barriers. An example of what local content can be is problems to be identified earlier, which, if acted upon, can ensure a information for farmers regarding which vegetables can be grown more effective and efficient project [10]. on their fields [30]. An ICT4D project must be demand-driven (ICT4D-CSF2) The political context must be analyzed and considered [28–32]. The ICT4D project must satisfy the present needs while (ICT4D-CSF9) [10, 29, 31–33]. The political situation in a coun- also allowing for the needs of future generations [31]. Furthermore, try can affect an ICT4D project on two levels: micro and macro implementing an ICT4D project in an area where there is not suffi- level [31]. On a micro level issues regarding ownership can arise cient demand will not result in a sustainable ICT4D project [28]. A due to a lack of defined ownership over processes and resources, or critical element in achieving this is making sure that the stakehold- from unsuccessful transfers of ownership. On a macro level issues ers have ownership over the ICT4D project (ICT4D-CSF5) so as to can arise due to increased bureaucracy or because the project is increase their involvement in and acceptance of the ICT4D project. turned into a political statement. Relevant skills must be built and trained (ICT4D-CSF3) [10, An ICT4D project must have a project champion (ICT4D- 28–31, 33]. Project management, implementation, and ICT skills are CSF10) [10, 28, 29, 36]. Renken and Heeks [36] define an ICT4D scarce in developing countries and need to be taught [10]. Illiteracy project champion as follows: ‘Any individual who makes a decisive is also an important problem [34]. Building and training these skills contribution to the ICT4D project by actively and enthusiastically can be expensive however, so mechanisms for knowledge shar- promoting its progress through critical stages in order to mobilise ing to reduce costs are recommended [31]. ICT training also helps resource and/or active support and cooperation from project stake- overcome technophobia [35]. Finally, this training be a continuous holders’. Multiple ICT4D project champions are necessary, to reduce process [28]. the risk of the project falling apart if an ICT4D project champion Efforts must be made to retain staff (ICT4D-CSF4) [10, 31]. leaves the project [10]. The effects of talented staff leaving can be disastrous [31]. Other The right technology must be chosen (ICT4D-CSF11) [10, than traditional intrinsic (e.g. praise) and extrinsic (e.g. salary) re- 29–31, 33]. Ferguson and Ballantyne (2002) argue that the tech- wards, project ownership (ICT4D-CSF5) can be a major motivational nology chosen plays an important role in the long-term success of incentive due to the involvement it brings. ICT4D projects (ICT4D-CSF11). The reliability of ICT infrastructure, Project ownership must be given to local parties (ICT4D- the availability of technology, and the maintenance and upgrading CSF5) [28, 29, 31]. Local ownership is defined as the active partic- of ICT are key factors [31]. Technology also needs to be affordable ipation of the local community in all phases of the development for the people involved with the ICT4D project [10]. process [31]. Successful local ownership will result in the commu- A cultural understanding of the local community must be nity viewing the ICT as an integral part of their daily lives [28]. developed (ICT4D-CSF12) [11, 33]. Cultural understanding can Local ownership is related to two others CSFs: it improves the be necessary to avoid conflicts during the constant interaction alignment of the ICT4D project to the needs of its stakeholders between outsiders and the local community [11]. Cultural under- (ICT4D-CSF2) and it improves the motivation of staff (ICT4D-CSF4). standing can also be necessary to become accepted within the local An ICT4D project must be economically self-sustainable community and gain their trust, as well as to gain access to their (ICT4D-CSF6) [10, 12, 28–32]. Many ICT4D projects rely on donor resources [12]. money for their continued survival, which means these projects Trust between the local community and outside parties risk falling apart as soon as enthusiasm and funding from outside must be built (ICT4D-CSF13) [11, 12]. Trust can be a contribut- partners disappears [32]. Economic self-sustainability is therefore ing factor to the willingness to cooperate with another party, and important to ensure the long-term success of ICT4D projects. How- becomes necessary if that cooperation results in the trustor being ever, donor money is important initially, because due to the experi- put at risk [37]. In an ICT4D project the local community would mental nature of many ICT4D projects it cannot be expected for be the trustor, and the outside party the trustee. There are two fac- these projects to be profitable from the get-go [18, 32]. One par- tors that determine the level of trust [37]: the trustor’s propensity ticular important aspect to ensure economic self-sustainability is to trust and the trustee’s perceived trustworthiness. The propen- marketing, because the inability to inform the community about the sity to trust differs among individuals, but factors that influence benefits of ICT4D projects is one of the main reasons why ICT4D the propensity are history with development, personality, and the projects fail [28] culture. Trustworthiness has ability, benevolence, and integrity as Local partnerships must be built to achieve synergies antecedents. Ability refers to the skills and expertise of a party (ICT4D-CSF7) [10, 28, 29, 31]. Ferguson and Ballantyne (2002) ar- within a domain (e.g. knowledge about ICT). Benevolence refers to gue for the importance of building local partnerships (ICT4D-CSF7). what degree the party desires to help the trustor without regard to 3 extrinsic rewards. Finally, integrity refers the trustee’s adherence which the user is positioned is dynamic [42]. Agile methods are to principles. more robust to change than traditional methods because of two reasons [41]. First, Agile firms typically use a more simple software 4.2 Critical success factors for Agile software architecture, postponing any complex and binding changes as much development as possible. This makes the architecture more robust to change. Sec- ond, Agile firms typically allow for requirements variability in the A survey study to the CSFs of Agile projects on four dimensions contract between the developer and the user. In those situations, to project success revealed six CSFs for Agile methods [24]. These users can specify or adjust requirements at the beginning of each four dimensions are quality (the quality of the delivered product), iteration. scope (to what degree the product meets the user’s requirements), Improved communication with the user (Agile-ADV2). In timeliness (whether the product is delivered on time or not), and Agile methods, face-to-face communication with the user instead cost (whether the real costs and effort put in were as projected). of rigorous documentation is the norm [42]. The iterative nature of The six CSFs are discussed below. Agile allows for more frequent communication with the user. Fur- Team environment (Agile-CSF1). A good team environment thermore, this more frequent face-to-face-communication allows contributes positively to the quality of the product [24]. The entire for an improved elicitation and validation of requirements, which team should be located in a single place, the team should be small, reduces the likelihood of requirements changing later on [42, 43]. and the team should be self-organizing. If a project has multiple Higher quality of software (Agile-ADV3). Software quality teams they should work collaboratively rather than independently. principles that quality professionals have been preaching for are Teams should be small because as a team has more members, coor- included in Agile methods [6]. An example of such a principle is dination becomes more difficult [38, 39]. test driven development, which is an approach that suggests writ- Team capability (Agile-CSF2). Team capability positively con- ing automated tests first, and code afterwards if the tests fail [44]. tributes to the timeliness and cost of a project [24]. A good team Frequent user feedback is also mentioned as a reason for improved member should have high competence, expertise, and motivation. software quality [39]. A good manager should have an adaptive management style and Increased user satisfaction (Agile-ADV4). There are several possess knowledge on Agile. Additionally, relevant technical train- factors that contribute to this [45]: improved communication with ing should be provided to the team members. Highly competent users (Agile-ADV2), increased user involvement (Agile-CSF3), and team members are important to compensate for the smaller team the improved quality of software (Agile-ADV3). size [38]. Finally, developers must possess domain knowledge in Good, internal communication (Agile-ADV5). The required order to be able to communicate with the users [40]. strong focus on internal communication (Agile-CSF4) results in an User involvement (Agile-CSF3). User involvement positively improved understanding of the requirements, tasks, project status, contributes to the scope of the product [24]. To achieve good user and resource allocation among all team members [43]. involvement, a positive user relationship should be built. The user Improved employee job satisfaction (Agile-ADV6). Job sat- should have complete authority regarding the project. Finally, the isfaction is higher for Agile methods for six reasons [46]: employees user should have a strong commitment and presence. experienced less stress, felt more productive, enjoyed the internal Project management (Agile-CSF4). Project management pro- communication (Agile-ADV5), found the job environment more cesses positively contributes to the quality of the product [24]. Re- pleasant and comfortable, were more motivated, and were more will- quirement management processes, project management processes, ing to continue using their software development method. The im- and configuration management processes should all be Agile. A proved software quality (Agile-ADV3) also contributed to a higher working schedule should be put in place and followed. Progress job satisfaction [45]. should be tracked. There should be a strong focus on communica- A higher return on investment (Agile-ADV7). The return tion, for example with daily face-to-face meetings. on investment (ROI) is higher in projects done with Agile methods Agile software engineering techniques (Agile-CSF5). Agile for several reasons [47]: higher software quality (Agile-ADV3), software engineering techniques positively contribute to the qual- increased user satisfaction (Agile-ADV4), lower costs, and higher ity and scope of the product [24]. These techniques are: coding productivity. standards, simple design, refactoring, limited but sufficient docu- Increase in successful projects (Agile-ADV8). The more of mentation, and integration testing. the principles of the Agile approach is applied in the project, the Delivery strategy (Agile-CSF6). Delivery strategy positively higher the project success [5]. Hayes’ study (as cited in [39]) attrib- contributes to the scope of the product, and the timeliness and cost uted this occurrence to the iterative nature of Agile. An iterative of the project [24]. A good delivery strategy prioritizes the impor- cycle instead of a sequential cycle supposedly increases the visibil- tant features of the product first. Furthermore, software should be ity of the project. With this increased visibility the potential success regularly delivered. of the project would then become clearer, which gives insight into whether adjustments can or have to be made, or if the project has 4.3 Advantages of Agile software development to be cancelled entirely. More robust to changing requirements (Agile-ADV1). Require- Improved control over projects (Agile-ADV9). Hayes (as ments are inherently variable because both the developer and user cited by Mahanti [39]) argues that Agile methods improve the acquire more knowledge about the domain of the application [41]. Requirements also change because the business environment in 4 control over projects due to several reasons: ‘Short iterations, multi- example, the concepts of collaboration and iterations. Furthermore, disciplinary teams, knowledge sharing, continuous integration, and according to the interviewees, these principles are shared with feedback’. fields of science relevant for ICT4D, for example social sciences. Improved organizational learning (Agile-ADV10). Agile meth- If that is the case, teaching Agile (or rather, the principles behind ods focuses on teamwork and foster organizational learning within Agile) to team members without an ICT background may not be those teams [48], for example with pair programming [49]. a problem. After the training is done, Agile project management processes and Agile software engineering techniques need to be 5 ANALYSIS applied. One possible difficulty that can arise here is the interfer- ence of donors. According to the interviewees, the commitment of 5.1 Suitability of Agile software development the team is stronger to the donors (who are the customers) than to for ICT4D projects the local community (who are the users). In the experience of the The suitability of Agile for ICT4D projects is analyzed by comparing interviewees, managers of donor companies prefer contracts, clear the CSFs for Agile methods and for ICT4D projects. roadmaps, and traditional software development. A good team environment (Agile-CSF1) is also necessary in or- Team capability (Agile-CSF2) requires two further considera- der for an ICT4D project to be successful. For ICT4D projects it is tions to be made. First, the team members need to possess domain beneficial for all the team members to be located near each other, knowledge, or gain domain knowledge through interaction with because an active presence within the local community can con- the local community. The latter method will require cooperation. tribute to developing a cultural understanding (ICT4D-CSF12), and Second, in satisfying the CSFs demand-driven (ICT4D-CSF2), local because ownership over the project by the local community re- ownership (ICT4D-CSF5), and local partnerships (ICT4D-CSF7), quires their active participation (ICT4D-CSF5). However, this is motivation of the local community for the project is built. Both not always possible for ICT4D projects. For example, according to domain knowledge and motivation are important aspects of team the interviewees, members of W4RA go several times a year for capability. extended periods. The goal is to do as much work as possible in In summary, several CSFs for ICT4D projects have to be consid- those time frames, because it is not financially doable to remain ered before Agile can be used in ICT4D projects. First, the ICT4D there during the entire project. Another aspect of the team environ- project needs to be demand driven (ICT4D-CSF1). Second, Agile ment is that teams should be small, which fits with ICT4D projects practices need to be taught to the local community members in- because of budget restrictions. volved as well as to team members without an ICT background User involvement (Agile-CSF3) requires cooperation with the lo- (ICT4D-CSF3). Third, a cultural understanding must be developed cal community, which is also important for ICT4D projects (ICT4D- (ICT4D-CSF12). According to the interviewees, an ICT4D project CSF2). From the ‘Critical success factors for ICT4D projects’ section, should start out by looking at what the local community has, and important prerequisites for cooperation can be identified: devel- by letting the local community explain what they do. Field research oping a cultural understanding (ICT4D-CSF12) and building trust is important in this initial step, and the goal is to determine in what (ICT4D-CSF13). Multiple factors contribute to developing a cultural ways ICT could be used. In this initial step a cultural understanding understanding. These factors are cultural interpreters, local part- can be developed, thus allowing Agile to be used once develop- nerships, and a strong presence within the local community. For ment initiates. Fourth, trust must be built (ICT4D-CSF13). Similarly building trust the trustor’s propensity to trust and the trustee’s to developing a cultural understanding, trust can be built in that perceived trustworthiness are of importance. initial step. However, trust cannot be fully built, because it is in An incremental Agile delivery strategy delivers software regu- iterative and dynamic process [37], which is also echoed by the larly, and prioritizes the most important features first (Agile-CSF6). interviewees. If a demand-driven approach does not build enough Such a strategy allows for more user involvement [51], which sub- motivation within the local community, local ownership (ICT4D- sequently means that cooperation is an important prerequisite. CSF5) and building local partnerships (ICT4D-CSF7) also becomes The following three CSFs for Agile can be challenging to sat- a prerequisite for using Agile. Building local partnerships may also isfy for similar reasons: team capability (Agile-CSF2), project man- help building a cultural understanding (ICT4D-CSF7). According agement processes (Agile-CSF4), and Agile software engineering to the interviewees, local partnerships may also help building trust. techniques (Agile-CSF5). Project management and ICT skills are Finally, one incompatibility between Agile and ICT4D exists. Agile often lacking in developing countries (ICT4D-CSF3). If people from may be incompatible for ICT4D projects because it is not always the community are actively included in the software development possible to work on location. process, they will need to be educated on Agile. This relates to both Agile processes in project management and Agile software engineering techniques. Furthermore, because ICT4D is a multi- 5.2 How Agile software development can disciplinary field [50], not every team member may have a back- improve ICT4D projects ground in ICT. It is thus possible that people assigned to roles such The effects of Agile on ICT4D projects are determined by analyzing as cultural interpreter or business strategy (to ensure economic which advantages or characteristics of Agile can influence which self-sustainability) are not familiar with Agile either, and will also CSFs for ICT4D projects. have to be educated. However, teaching Agile may not be difficult, Monitor and evaluate project progress regularly (ICT4D- or even necessary. According to the interviewees, it is not Agile that CSF1). The improved control over projects (Agile-ADV9) suggests needs to be taught, but rather the principles behind Agile. So for 5 that Agile can contribute. In particular, the good internal communi- there are two important concepts: the trustor’s propensity to trust cation of teams can be of benefit here (Agile-ADV5). The frequent and the trustee’s perceived trustworthiness [37]. Agile cannot in- meetings can (and should) be used to discuss the current progress of fluence the trustor’s propensity to trust, because there is no reason the ICT4D project. Furthermore, the frequent delivery of project de- to believe Agile can influence factors such as personality or cul- liverables (Agile-CSF6) and the feedback from users (Agile-ADV2) ture. Agile also cannot initially influence the trustee’s perceived can help with regular evaluation. trustworthiness, because there is no reason to believe Agile can in- An ICT4D project must be demand-driver (ICT4D-CSF2). fluence the ability of the trustee (although skills and expertise, such Agile has several advantages that can contribute to ensuring an as domain knowledge, are required to make Agile work successfully, ITC4D project is demand-driven. First, user collaboration is one of Agile-CSF1), the benevolence of the trustee, or the integrity of the the key pillars of Agile, and as a result Agile has good communica- trustee. However, the perceived trustworthiness of the trustee is tion with the user (Agile-ADV2). Elicitation of requirements is done dynamic and affected by the results of trust-taking behaviour of the iteratively and frequently, ensuring that the ICT fits the demand trustor. In the context of ICT4D projects, trust-taking behaviour of the users. Second, should requirements change, which occurs of the trustor can be considered as letting an organization into more than normally for ICT4D projects [17], then Agile has the the local community or allowing them to develop ICT that will advantage of being robust to change (Agile-ADV1). According to impact their lives. The iterative nature of Agile and the frequent the interviewees, demos, prototypes, workshops and movies make delivery of working software (Agile-CSF6) will allow for more fre- the local community familiar with ICT and helps them understand quent outcomes of trust-taking behaviour. And the increased user how ICT could help them in their lives. satisfaction when using Agile methods (Agile-ADV4) suggests that Relevant skills must be built and trained (ICT4D-CSF3). these outcomes will be more frequently positive. One result from section 5.1 is that skills pertaining to Agile have Agile might have some small effects on the following CSFs for to be built and taught as well, thus making this CSF more time ICT4D projects. By contributing to other CSFs for ICT4D projects, consuming to achieve. However, Agile might also help building Agile helps ensure the continued use of and therefore the demand and training Agile related skills and other ICT skills, because Agile for the ICT. Sustainable demand is one of the pillars of an economi- methods foster organizational learning (Agile-ADV10). For example, cally self-sustainable ICT4D project (ICT4D-CSF6). Additionally, by pair programming helps build programming skills. The net effect of focusing on working software and by frequently delivering software Agile upon building and training relevant skills is thus unknown. (Agile-CSF6), local content (ICT4D-CSF8) can be created earlier on Efforts must be made to retain staff (ICT4D-CSF4). Two in the ICT4D project. Finally, because Agile contributes to ensuring benefits of Agile that can help are the increased job satisfaction a demand-driven ICT4D project, an improved understanding of of employees (Agile-ADV6) and increased organizational learning the local community’s needs is gained. This will allow the right (Agile-ADV10). One research found that job satisfaction is nega- technology to be chosen with greater accuracy (ICT4D-CSF11). tively correlated to turnover intention, which is to say that increas- Agile was not believed to have any significant impact on three ing job satisfaction will reduce the intention to leave the firm (or the CSFs for ICT4D. The collaborative nature of Agile (see the Intro- project) [52]. Furthermore, that same research found that an organi- duction) might improve the communication with and satisfaction zational learning culture is strongly, positively correlated with job of partners (ICT4D-CSF7) and project champions (ICT4D-CSF10), satisfaction. Finally, the correlation between learning culture and similar as to how it improves the communication with and satisfac- turnover intention was not significant. However, organizational tion of users (Agile-ADV2, Agile-ADV4). However, it cannot help learning culture is still a good construct to increase employee re- with seeking partners and building partnerships, or with finding tention, because organizational learning culture is indirectly linked project champions. Finally, Agile was not believed to be able to to turnover intention through job satisfaction [52]. analyse and consider the political context (ICT4D-CSF9). However, Project ownership must be given to local parties (ICT4D- the adaptive nature of Agile (see the Introduction) and the subse- CSF5). An important antecedent for user participation is the user’s quent robustness to changing requirements (Agile-ADV1) allows perceived support by the organization [53]. This can be achieved by, an ICT4D project to respond to changes in the political context. for example, listening to the problems of the users and solving these Additionally, in terms of frequency, the most important advan- problems [54]. Agile can indirectly increase participation of the local tages and characteristics of Agile can be identified. Improved com- community due to the positive effect of Agile for ensuring an ICT4D munication with the user (Agile-ADV2) has a positive effect on four project is demand-driven. Furthermore, the good communication CSFs for ICT4D, improved organizational learning (Agile-ADV10) with users (Agile-ADV2) can help with facilitating participation. on three, and good communication within the team on two. Fur- A cultural understanding of the local community must thermore, while not defined as an advantage, the focus on frequent be developed (ICT4D-CSF12). A strong presence within the lo- delivery (Agile-CSF6) has a positive effect on three CSFs. cal community helps develop a cultural understanding [12], which Agile contributes to by focusing on user collaboration and by im- 6 DISCUSSION proving user communication (Agile-ADV2). Furthermore, the good, internal communication (Agile-ADV5) and the improved organi- ICT4D projects have high rates of failure [3, 4] and the Agile method- zational learning (Agile-ADV10) of Agile can help disseminate the ology was found to increase the success rate of ICT projects [5]. cultural understanding throughout the organization. However, Agile methods successful in western countries cannot Trust between the local community and outside parties carelessly be applied in ICT projects for developing countries [8]. must be built (ICT4D-CSF13). To summarize how trust is built, Examples of why are the worse or lack of ICT and management 6 skills [10] and cultural barriers [11, 12]. Furthermore, current re- the parties must also consider that all those three critical success search on why to apply Agile methods is mainly limited to the factors are iterative processes. Therefore, the Agile method must benefit of improved user collaboration [14–16]. also allow for efforts to be made towards satisfying those three The main value of this research lies in the insight it gives into critical success factors in later stages of the project. how Agile can improve ICT4D projects and in what ways. This SQ2: How can ICT4D projects benefit from an Agile approach? is important both for Agile frameworks as (for example [19, 21]) Agile can positively contribute towards satisfying all but four and for frameworks using Agile elements (for example [18, 20]), critical success factors for ICT4D: monitor and evaluate the project because there might be additional aspects of Agile that could further regularly, ensure a demand-driven ICT4D project, make efforts to improve the frameworks discussed in the related works section retain staff, give local owernship to the local community, ensure or form the theoretical basis of new ICT4D frameworks. Existing economic self-sustainability, create local content, choose the right literature mainly focuses on user collaboration as a reason for technology, develop a cultural understanding, and build trust. For why Agile methods can improve ICT4D projects [14–16], whereas three critical success the effect is insignificant: build local part- this paper found other aspects of Agile as well that can improve nerships, understand the political context, and ensure a project ICT4D projects. Most notable are organizational learning, team champion. For the remaining critical success factor, building and communication, and frequent delivery. training skills, there is both a positive and negative effect, thus However, several limitations to this research must be addressed. resulting in an uncertain net effect. The most important advantages First, a limitation regarding the ICT4D literature. In the interview or characteristics of Agile, in terms of frequency, are the improved that was held, the interviewees raised a potential problem in ICT4D communication with the user, improved organizational learning, literature. On one end there are case studies. While interesting, the good communication within the team, and the focus on frequent question is what their findings mean for ICT4D as a whole. On the delivery. other end there is desk research. Such research attempts to create As addressed in the discussion, the results of this research have policy for ICT4D as a whole, but lacks a link to real ICT4D projects. not been validated in practice. Case studies are thus necessary in The interviewees’ opinion thus suggests that the ICT4D literature which Agile methods are used, so that the proposed benefits can used in this paper is inadequate to provide a conclusive answer be assessed. Furthermore, as ICT4D is a multi-disciplinary field, it to the research question. Furthermore, literature for the CSFs for might prove worthwhile to investigate the effects of other software ICT4D and Agile and for the advantages of Agile was not collected engineering techniques or approaches from other sciences, which in a systematic way, thus providing no guarantee that the CSFs can then be used to design an ICT4D framework. of ICT4D and Agile and the advantages of Agile are exhaustive. A final limitation is that the findings of this research have not been 8 ACKNOWLEDGEMENTS validated in practice or through rigorous expert interviews. The authors wish to thank prof. dr. Akkermans, ms. drs. Bon, and ms. drs. Tuijp for sharing their expertise during the interview. 7 CONCLUSION The relationship between Agile and ICT4D was explored largely REFERENCES due to the focus of Agile on user collaboration. Theory postulated [1] Geoff Walsham. 2017. ICT4D research: reflections on history and future that there are additional variables to consider in an ICT4D project, agenda. Information Technology for Development 23, 1 (February 2017), 18–41. DOI:http://dx.doi.org/10.1080/02681102.2016.1246406 and as such an answer to the following research question was [2] Nagy K. Hanna. 2010. E-Transformation: Enabling New Development Strategies, sought: ‘To what degree can Agile software development improve New York, NY: Springer-Verlag New York. ICT4D projects?’. This paper arrived at an answer to the research [3] Richard Heeks. 2002. Information Systems and Developing Countries: Failure, Success, and Local Improvisations. The Information Society 18, 2 (2002), 101–112. question by answering two sub-questions, the answers of which are DOI:http://dx.doi.org/10.1080/01972240290075039 summarized below. Collectively, these answers provide an answer [4] Independent Evaluation Group. 2011. Capturing Technology for Develop- ment: An Evaluation of World Bank Group Activities in Information and to the main research question. Communication Technologies Volume 1. (2011). Retrieved March 6, 2018 from SQ1: Can Agile methods successfully work in an ICT4D project? https://openknowledge.worldbank.org/bitstream/handle/10986/2370/653750 PUB0v10B0BLIC00ict0evaluation.pdf Four critical success factors for ICT4D need to be satisfied before [5] Pedro Serrador and Jeffrey K. Pinto. 2015. Does Agile work? — A quantitative an Agile method can work: the projects needs to be demand-driven, analysis of agile project success. International Journal of Project Management 33, 5 (2015), 1040–1051. DOI:http://dx.doi.org/10.1016/j.ijproman.2015.01.006 skills pertaining to Agile need to be taught to the stakeholders [6] Scott Ambler. 2005. Quality in an agile world. Software Quality Professional 7, 4 actively involved in the development, a cultural understanding (2005), 34-40 [7] Jim Highsmith. 2002. What is Agile Software Development? Crosstalk - The must be developed, and trust must be built. Though not necessarily Journal of Defense Software Engineering 15, 10 (2002), 4-9 prerequisites, local ownership and building local partnerships can [8] Joerg doerflinger and Tom Gross. 2010. Bottom billion architecture. Pro- also play an important role in ensuring that Agile can work cor- ceedings of the 4th ACM/IEEE International Conference on Information and Communication Technologies and Development - ICTD 10 (2010). rectly by increasing the motivation of the local community. Local DOI:http://dx.doi.org/10.1145/2369220.2369228 partnerships can also contribute to developing a cultural under- [9] Joerg Doerflinger and Tom Gross. 2010. Technical ICTD - A User Centered standing and building trust. The advice for parties who seek to set Lifecycle. Communications: Wireless in Developing Countries and Networks of the Future IFIP Advances in Information and Communication Technology up a demand-driven ICT4D project is to select an Agile method for (2010), 72–83. DOI:http://dx.doi.org/10.1007/978-3-642-15476-8_8 ICT4D projects that allows for a substantial pre-development phase, [10] Tina James. 2004. Information and communication technologies for develop- ment in Africa, Ottawa: International Development Research Centre (IDRC). in which relevant skills can be taught and developed, a cultural [11] Kirstin Krauss. 2016. Demonstrating Critically Reflexive ICT4D Project Conduct understanding can be developed, and trust can be built. However, and ICT Training in Rural South Africa. Proceedings Annual Workshop of the 7 AIS Special Interest Group for ICT in Global Development at AIS Electronic [33] Almamy Touray, Airi Salminen, and Anja Mursu. 2013. ICT Barriers and Library, paper 13. Critical Success Factors in Developing Countries. The Electronic Jour- [12] M.S. Sandeep and M.N. Ravishankar. 2015. Impact sourcing ventures and local nal of Information Systems in Developing Countries 56, 1 (2013), 1–17. communities: a frame alignment perspective. Information Systems Journal 26, DOI:http://dx.doi.org/10.1002/j.1681-4835.2013.tb00401.x 2 (2015), 127–155. DOI:http://dx.doi.org/10.1111/isj.12057 [34] Aaron Ciaghi and Adolfo Villafiorita. 2011.Crowdsourcing ICTD best practices. [13] Punita Bhatt, Ali J. Ahmad, and Muhammad Azam Roomi. 2016. So- In International Conference on e-Infrastructure and e-Services for Developing cial innovation with open source software: User engagement and Countries, 167-176 development challenges in India. Technovation 52-53 (2016), 28–39. [35] Arul Chib and A.l.e Komathi. 2009. Extending the Technology- DOI:http://dx.doi.org/10.1016/j.technovation.2016.01.004 Community-Management model to disaster recovery: Assessing vul- [14] Rüdiger Heimgärtner, Alkesh Solanki, and Bernd Hollerit. 2014. Enhanc- nerability in rural Asia. 2009 International Conference on Information ing Usability Engineering in Rural Areas Using Agile Methods. Design, and Communication Technologies and Development (ICTD) (2009). User Experience, and Usability. Theories, Methods, and Tools for Design- DOI:http://dx.doi.org/10.1109/ictd.2009.5426694 ing the User Experience Lecture Notes in Computer Science (2014), 445–452. [36] Jaco Renken and Richard Heeks. 2013. Conceptualising ICT4D project cham- DOI:http://dx.doi.org/10.1007/978-3-319-07668-3_43 pions. Proceedings of the Sixth International Conference on Information and [15] Degif Teka, Yvonne Dittrich, and Mesfin Kifle. 2016. Usability challenges in an Communications Technologies and Development Notes - ICTD 13 - volume 2 Ethiopian software development organization. Proceedings of the 9th Interna- (2013). DOI:http://dx.doi.org/10.1145/2517899.2517928 tional Workshop on Cooperative and Human Aspects of Software Engineering [37] Roger C. Mayer, James H. Davis, and F. David Schoorman. 1995. An integrative - CHASE 16 (2016). DOI:http://dx.doi.org/10.1145/2897586.2897604 model of organizational trust. Academy of management review 20, 3 (1995), [16] Anna Bon and Hans Akkermans. 2014. Rethinking technology, ICTs and devel- 709-734. opment: Why it is time to consider ICT4D 3.0. (2014). Retrieved March 8, 2018 [38] Mikael Lindvall et al. 2002. Empirical Findings in Agile Methods. Extreme from https://w4ra.org/wp-content/uploads/2015/01/ICT4D3.pdf Programming and Agile Methods — XP/Agile Universe 2002 Lecture Notes in [17] Matt Haikin. 2013. Reflections on applying iterative and incremen- Computer Science (2002), 197–207. DOI:http://dx.doi.org/10.1007/3-540-45672- tal software development methodologies (Agile, RAD etc.) to aid and 4_19 development work in developing countries. Retrieved March 8, 2018 [39] Aniket Mahanti. 2006. Challenges in Enterprise Adoption of Agile Methods - A from http://www.hiidunia.com/wp-content/uploads/downloads/2013/07/agile- Survey. Journal of Computing and Information Technology 14, 3 (2006), 197. blarticle-part-11.pdf DOI:http://dx.doi.org/10.2498/cit.2006.03.03 [18] Anna Bon, Hans Akkermans, and Jaap Gordijn. 2016. Developing ICT Services [40] Hilkka Merisalo-Rantanen, Tuure Tuunanen, and Matti Rossi. 2005. Is Extreme in a Low-Resource Development Context. Complex Systems Informatics and Programming Just Old Wine in New Bottles. Journal of Database Management Modeling Quarterly, 9 (2016), 84–109. DOI:http://dx.doi.org/10.7250/csimq.2016- 16, 4 (2005), 41–61. DOI:http://dx.doi.org/10.4018/jdm.2005100103 9.05 [41] Alberto Sillitti, Martina Ceschi, Barbara Russo, an Giancarlo Succi. 2005. Man- [19] Joerg Doerflinger and Andy Dearden. 2013. Evolving a Software Development aging Uncertainty in Requirements: A Survey in Documentation-Driven and Methodology for Commercial ICTD Projects. Information Technologies & In- Agile Companies. 11th IEEE International Software Metrics Symposium (MET- ternational Development 9, 3 (2013), 43-60. RICS05). DOI:http://dx.doi.org/10.1109/metrics.2005.29 [20] Maria Angela Ferrario, Will Simm, Peter Newman, Stephen Forshaw, and Jon [42] Lan Cao and Balasubramaniam Ramesh. 2008. Agile requirements engineering Whittle. 2014. Software engineering for social good: integrating action research, practices: An empirical study. IEEE software 25, 1 (2008), 60-67 participatory design, and agile development. Companion Proceedings of the [43] Minna Pikkarainen, Jukka Haikara, Outi Salo, Pekka Abrahamsson, and 36th International Conference on Software Engineering - ICSE Companion Jari Still. 2008. The impact of agile practices on communication in soft- 2014 (2014). DOI:http://dx.doi.org/10.1145/2591062.2591121 ware development. Empirical Software Engineering 13, 3 (2008), 303–337. [21] Henrik Hansson, Peter Mozelius, Jarkko Suhonen, Erkki Sutinen, Mikko Vesise- DOI:http://dx.doi.org/10.1007/s10664-008-9065-9 naho, and Gunnar Wettegren. 2009. ICT4D with a Nordic flavor - A stepwise [44] David Janzen and Hossein Saiedian. 2005. Test-driven development con- and multithreaded approach. In IST-Africa 2009 conference proceedings. IIMC cepts, taxonomy, and future direction. Computer 38, 9 (2005), 43–50. International Information Management Corporation, 1-9. DOI:http://dx.doi.org/10.1109/mc.2005.314 [22] Aaron Ciaghi, Adolfo Villafiorita, and Lorenzo Dalvit. 2014. Understanding Best [45] Chris Mann and Frank Maurer. A case study on the impact of scrum on Practices for ICTD Projects: Towards a Maturity Model. In ICTs for Inclusive overtime and customer satisfaction. Agile Development Conference (ADC05). Communities in Developing Societies. Proceedings of the 8th International DOI:http://dx.doi.org/10.1109/adc.2005.1 Development Informatics Association Conference, 349-360 [46] Katiuscia Mannaro, Marco Melis, and Michele Marchesi. 2004. Empirical Anal- [23] Zarina Alias, E.m.a. Zawawi, Khalid Yusof, and N.m. Aris. 2014. Determin- ysis on the Satisfaction of IT Employees Comparing XP Practices with Other ing Critical Success Factors of Project Management Practice: A Conceptual Software Development Methodologies. Extreme Programming and Agile Pro- Framework. Procedia - Social and Behavioral Sciences 153 (2014), 61–69. cesses in Software Engineering Lecture Notes in Computer Science (2004), DOI:http://dx.doi.org/10.1016/j.sbspro.2014.10.041 166–174. DOI:http://dx.doi.org/10.1007/978-3-540-24853-8_19 [24] Tsun Chow and Dac-Buu Cao. 2008. A survey study of critical success factors in [47] David F. Rico. 2008. What is the Return on Investment (ROI) of Agile Methods? agile software projects. Journal of Systems and Software 81, 6 (2008), 961–971. . (2008). http://ww.davidfrico.com/rico08a.pdf DOI:http://dx.doi.org/10.1016/j.jss.2007.08.020 [48] Sridhar Nerur, Radhakanta Mahapatra, and George Mangalaraj. 2005. Chal- [25] Edwin Van Teijlingen. 2014. Semi-structured interviews. (2014). lenges of migrating to agile methodologies. Communications of the ACM 48, 5 https://intranetsp.bournemouth.ac.uk/documentsrep/PGR%20Workshop%20- (January 2005), 72–78. DOI:http://dx.doi.org/10.1145/1060710.1060712 %20Interviews%20Dec%202014.pdf [49] Harald Holz and Frank Maurer. 2003. Knowledge Management Support [26] W4RA. About W4RA. Retrieved March 11, 2018 from https://w4ra.org/w4ra/ for Distributed Agile Software Processes. Advances in Learning Soft- [27] W4RA. RadioMarché, Voice-based market information system. Retrieved March ware Organizations Lecture Notes in Computer Science (2003), 60–80. 11, 2018 from https://w4ra.org/radiomarche-voice-based-market-information- DOI:http://dx.doi.org/10.1007/978-3-540-40052-3_7 system/ [50] Devinder Thapa and Mathias Hatakka. 2017. Introduction to ICT4D: ICTs [28] M.E.Kutu Mphahlele and Maisela E. Maepa. 2003. Critical success and Sustainable Development Minitrack. In Proceedings of the 50th Hawaii factors in telecentre sustainability: a case study of six telecentres International Conference on System Sciences, 2579 in the Limpopo Province. Communicatio 29, 1-2 (2003), 218–232. [51] Amy Law and Raylene Charron. 2005. Effects of agile practices on social factors. DOI:http://dx.doi.org/10.1080/02500160308538028 Proceedings of the 2005 workshop on Human and social factors of software [29] United Nations Development Programme (UNDP). 2001. Essentials: engineering - HSSE 05 (2005). DOI:http://dx.doi.org/10.1145/1083106.1083115 Information communication technology for development. (2001). [52] Toby Marshall Egan, Baiyin Yang, and Kenneth R. Bartlett. 2004. The effects of http://web.undp.org/evaluation/documents/essentials_5.pdf organizational learning culture and job satisfaction on motivation to transfer [30] David Pieter Conradie, C. Morris, and S.J. Jacobs. 2003. Using in- learning and turnover intention. Human Resource Development Quarterly 15, formation and communication technologies (ICTs) for deep rural de- 3 (2004), 279–301. DOI:http://dx.doi.org/10.1002/hrdq.1104 velopment in South Africa. Communicatio 29, 1-2 (2003), 199–217. [53] Cedric Hsi-Jui Wu. 2011. A re-examination of the antecedents and impact of DOI:http://dx.doi.org/10.1080/02500160308538027 customer participation in service. The Service Industries Journal 31, 6 (2011), [31] Julie Ferguson, Peter Ballantyne, and Galin Kora. 2002. Sustaining ICT-enabled 863–876. DOI:http://dx.doi.org/10.1080/02642060902960768 development practice makes perfect? In Sustaining ICT-enabled development [54] Albert O. Hirschman. 1970. Exit, voice and loyalty: Responses to decline in practice makes perfect? The Hague: International institute for communication firms, organizations, and states. Cambridge, MA, Harvard University Press and development (IICD). [32] Kenneth Keniston and Deepak Kumar. 2003. The Four Digital Divides, Delhi: SAGE Publishers. 8