Keywords

1 Introduction

The first bank to place an Automated Teller Machine (ATM) was the Barclays in 1967 in London and nowadays there is practically no bank that operates without one [1]. The demand for cash of every financial client is the most important factor that influences in the nature of an ATM [2]. According to Hellmann [3], despite the fact that the preference for digital accounts is growing, those financial clients continue employing the ATMs as a source of cash, a feature that its digital competitors, web and mobile banking, cannot offer.

According to the studies conducted by Abd et al. [4], the clients appreciate an ATM that really fits their needs as much as they appreciate their time. In addition, according to Camilli et al. [5], for the final user, the interface is the product; usability problems or frustrating experiences could have a negative impact on the institutional brand. In that sense, Gumussoy [6] consider critical the design of usable banking software with the objective of overcome the complex financial system.

In the systematic review conducted by Aguirre et al. [7], is proved the lack of information about how to apply the usability knowledge in the ATM interface design; this problem was noticed first by Moquillaza and Paz [8], who proposed a set of steps, based on user-centered design methods, for the design of ATM user interfaces.

In the same systematic review, Aguirre et al. [7] found that security systems, banking software, and the self-service technologies are domains closely related to the ATM domain.

Only one framework was reported by the systematic review of Aguirre et al. [7]; however, several methods and techniques for designing ATM and interactive interfaces were also found in the literature. In this study, we use the ISO 13047, the standard for designing usable interactive interfaces, as a start point and will expand the mentioned systematic review in order to find the most accurate methods that support ATM interfaces design.

The paper has the following organization: Sect. 2, where the analysis of user-centered design methods conducted by Aguirre et al. is extended to the domain of self-service technologies, Sect. 3???, where we analyze the expert judgment gathered from surveys and interviews, and Sect. 4, where the proposed framework is presented and modeled as a process.

2 User-Centered Design Methods and Techniques

In the present section, a systematic review was conducted following the guidelines established by Kitchenham and Charters [9] in order to extend the review conducted by Aguirre et al. [7] and expand its scope to self-service technologies domain. The articles reported were filtered by quality criteria and grouped by the methods or techniques that they presented, according to the phases of the ISO 13047.

2.1 Quality Criteria

According to the studies of Realpe-Muñoz et al. [10], the related articles to the domain of usable, interactive and secure interfaces are easily classified by six quality criteria. Acknowledging the criteria proposed, we slightly modified them and add a new quality criterion that includes if an article takes into consideration the design of self-service technologies. This made the criteria more robust for the domain we are interested in (see Table 1).

Table 1. Quality criteria
Full size table

These criteria allow us to filter the articles reported in the literature. The articles with more quality will be compared to the phases of the standard ISO 13047 and serve as a source to obtain methods and techniques for ATM user interface design.

2.2 Systematic Review

The articles selection was conducted in order to analyze the domains of self-service technologies, interactive user interfaces and secure user interfaces. This was made with the objective of reporting all the frameworks, methods and techniques employed in the mentioned domains that ensure usability or, at least, take it into consideration.

The research questions were based on the PICOC criteria proposed by Petticrew and Roberts [11], which specify Population, Intervention, Comparison, Outcome, and Context. In this case, the objective is not a comparison between frameworks, methods or techniques; hence, we discarded it (see Table 2).

Table 2. PICOC table
Full size table

With these criteria, research questions were formulated and are the ones that follows:

  • Q1. Which usable design processes, methodologies or frameworks are used to design self-service technologies user interfaces?

  • Q2. What user-centered design methods or techniques are reported in the literature for the design of self-service technologies user interfaces?

  • Q3. Which usability methodologies or methods are employed in the design of interactive and secure user interfaces?

After defining inclusion and exclusion criteria, the review was started. The consulted databases were the well-known ACM Digital Library, SpringerLink and ScienceDirect.

2.3 Results of the Review

After conducting the review, the most relevant articles that answer the research questions were in total 9 (see Table 3).

Table 3. Relevant articles
Full size table

The nine selected articles were classified using the quality criteria defined in Sect. 2.1. The articles and the criteria they fit can be seen in Table 4. This classification served as a filter, recognizing the article of Kaptelinin et al. [12] as one with low quality for the purpose of this study.

Table 4. Articles classified by quality criteria
Full size table

In the study of Muhammad et al. [13], a commuter line train is redesigned employing methods from the user-centered design methodology. In Moquillaza and Paz [8], the authors proposed an 8 steps process to design ATMs interfaces using an user-centered approach. Siebendhandl et al. [14] redesigned a self-service ticket vending machine using methods from the fourth and fifth phase of the ISO 13047. Camilli et al. [5] made use of user-centered techniques to evaluate and redesign ATMs. Moquillaza et al. [15], proposed UCD methods for the design of an ATM in Perú. RealpeMuñoz et al. [10], proposed a design process based on MPIu+a and expand it to take into consideration the security for authentication systems. In the study of Zhang et al. [16], the authors proposed a six steps process to design interactive and usable public information systems interfaces. Finally, Wong et al. [17], proposed a methodology for the design of graphic interfaces and intangible products used for 8 years, where they used UX methods and a user-centered design approach.

2.4 Methods and Techniques

The ISO 13047:1999 [18] proposed human-centered design methodology for interactive systems. The term ‘human’ is used because the design must be centered in all humans (notice that ‘human’ is more generic than ‘user’) that interact with the system.

The standard proposed a five phases process and is illustrated as a life cycle in the Fig. 1.

Fig. 1.
figure 1

HCD methodology [18].

Full size image

The proposed phases by the standard are as follow: Planning, Context of use specification, Requirements specification, Design solutions, Evaluate designs. It is an iterative process aiming for the continuous improvement and correction of errors of the designs.

A correspondence analysis between the methods proposed in the ISO 13047 and the methods employed by every author in the studies reported in the previous section showed that the planning phase is not taken into consideration. The studies reported are focused on the requirements specification, the design of solutions and the evaluation. The objective of this comparison is to use this knowledge to propose a user-centered design process for the design of usable ATM interfaces that uses the methods and techniques proposed in the literature, taking as a baseline the phases propose in the standard ISO 13047. In Table 5 is shown the correspondence between the standard phases and the methods used in the literature.

Table 5. Correspondence between the standard ISO 13047 phases and the methods used in the literature.
Full size table

The articles employed methods and techniques for usable design. For a more extended analysis of these methods, they were also classified under their correspondence to the ISO 13047 phases. For this, the analysis that Maguire [19] made about the ISO 13047 and the methods that support it, serves as a guide to better correspondence analysis.

Some methods and techniques proposed by the articles reported, take into consideration aspects related to hardware design; for instance, Anthropomorphic Needs [13], Posture Evaluation [13] and Hardware Requirements [14]. Because hardware design is out of the scope of this study, these methods were automatically excluded from further analysis.

After finalized the analysis, some methods were excluded and the most accurate for the design of ATM interfaces were left for expert validation. The methods were classified under their correspondence to the phases of the ISO 13047 and are listed in Table 6.

Table 6. UCD methods proposed for the design of ATM interfaces
Full size table
Table 7. Methods recommended by the experts
Full size table

The Planning phase is initially not considered because it is not considered in the literature either. In the next sections, the phases and methods proposed will be validated by expert judgment with aid of Usability and ATM Domain experts.

3 Expert Judgment Analysis

In the present section, the objective is to validate the utility of the methods proposed in the previous section. For this, expert judgment is employed and is supported by surveys addressed to Usability experts and ATM experts. In this way, only the methods that fit the needs of the experts were left for the final proposal.

3.1 Elaboration of the Survey

In every project, the developers that decide using and user-centered design approach in a real project would encounter fundamental questions such as, what methods do I apply? when? how?. Even if the methods are selected, how to conduct them, how to manage them and how long would they take in practice are still real questions because there are multiple and heterogeneous definitions [20].

In the study of Maguire [19], he analyzed the ISO 13047 and identified characteristics that make the comparison of the methods easier:

  • How and when should the method be applied? - Simplicity

  • How long should the method take? - Time

  • What benefits that it brings to my project? - Cost vs. Benefit.

A survey was elaborated with closed questions about the participant profile, their experience with ATM usage and UCD design, and a Likert scale questionnaire to allow the participants to qualify the methods proposed in Sect. 2.4.

The survey was addressed to ATM domain experts from a well-known bank in Perú and different Usability experts with a Ph.D. degree from Colombia, México, Chile, and Perú.

3.2 Analysis of the Result

The first phase of the survey shows that the ATM domain experts had, in a majority, master degrees. The Usability domain experts had Ph.D. degrees and are in majority men. The answers about the experience with ATM usage and design showed that the most frequent operation made by the two groups of experts is Cash Withdrawal. The Usability experts use an ATM between 4 and 7 times a month, while the ATM domain experts made use of this self-service 10 or more times a month. In contrast, the Usability experts had more experience with UCD methodologies, 66.7% of the participants know the methodology and applied it before, while the ATM domain experts just the 22% of participants claimed to know the methodology.

The previous information serves as a participant profiler. The second phase is still about the UCD knowledge of the participants; however, is more specific to the methods proposed in Sect. 2.4. From this, is seen that the Usability experts have more experience with evaluation methods due to their academic degree. In the other phases, the ATM Domain experts stated that they know or have applied before most of the techniques.

In that sense, the score given by the ATM experts has double weight than the Usability experts in the first 3 phases proposed (context, requirements, and design) while the opposed happens in the last phase, where the Usability experts have more experience.

The participants were asked to give scores to the methods for answering the questions listed in Sect. 3.1, this way only the more simple methods, that not consume a lot of time and resources will be taken in consideration. Also, the participants were allowed to add other methods they know that are not in the list proposed.

In addition to the questionnaire, an interview with the ATM experts was held in order to contextualize their answers. Some identified considerations to have in mind were:

  • The minimum method to consider in the design process must be a decision of the designer/stakeholders.

  • The decision of taking into practice simple, fast or low-cost methods depends on the context of the projects, not on the framework.

  • Substantial changes are usually analyzed deeply and in a considerable time. Not always is necessary fast methods.

In this sense, four set of methods are proposed after filtering the ones with low scores in the questionnaire and adding the proposed by the participants in the correspondent ISO phase.

In Tables 8, 9, and 10, are listed the simple methods, the fast methods, and low-cost methods. In Table 11, the methods that fit the three characteristics are grouped in an optimal process.

Table 8. Simple methods for the design of usable ATM Interfaces
Full size table
Table 9. Fast methods for the design of usable ATM Interfaces
Full size table
Table 10. Low-cost methods for the design of usable ATM Interfaces
Full size table
Table 11. Optimal methods for the design of usable ATM Interfaces
Full size table

4 The Proposal: Framework for the Design of Usable ATM User Interfaces

Defining a framework involves defining its methods, process, activities, and flow since this adapt the framework to the specific characteristics and context of the ATM interface design and defines the real work to be done [21].

The framework encompasses the process for the design of usable ATM interfaces, which, depending on the context, will apply simple methods, fast methods, low-cost methods, or optimal methods.

The phases proposed by the standard ISO 13047 were used in the framework. The phase of context and requirements correspond directly with the second and third phase of the mentioned standard. The last two phases were group together in the third phase of the framework: design and evaluation of interfaces. The proposal is diagrammed in the Figs. 2 and 3.

Fig. 2.
figure 2

UCD Framework for the design of usable ATM interfaces

Full size image
Fig. 3.
figure 3

Detail of the design process for ATM interfaces

Full size image

The previous diagrams represent the generic view of the framework, which encompasses the 4 proposed processes, the simple, fast, low-cost, and optimal. Each process has the same sub-processes, which correspond to the three proposed phases, context specification, requirements specification and design and evaluation of interfaces.

This process is iterative, the designs could be continually improved through evaluation and redesign. To establish a workflow, the Figs. 4, 5, 6, and 7 detail the workflow that must be followed by the designers and stakeholders.

Fig. 4.
figure 4

Context specification details

Full size image
Fig. 5.
figure 5

Requirements specification

Full size image
Fig. 6.
figure 6

Design and evaluation of interfaces detail (1)

Full size image
Fig. 7.
figure 7

Design and evaluation of interfaces detail (2)

Full size image

4.1 Description of the Methods

According to the study of Maguire [19], the methods that support the standard ISO 13047 have an approach to orient the designers on how to apply the method and also an estimated time that it should take. Table 12 simplify the description of every method taken into consideration in the new framework for the design of ATM interfaces.

Table 12. Methods that support the framework for the design of usable ATM interfaces.
Full size table

5 Conclusions and Future Works

In this paper is described how the methods used to elaborate a framework for the design of ATM interfaces were obtained and analyzed. A systematic review was conducted to expand the study of Aguirre et al. [7]. After that, the method found in the literature were classified under the phases of the ISO 13047 to demonstrate that the whole process is not covered in any proposal. With the intention of completing the process, and expert judgment of the methods was analyzed and new methods were proposed by the experts.

Some important information that was gathered from the expert judgment method was that the context is variable and the framework must not limit the catalog of methods available to the designers. The framework does not limit the methods that the designer can apply in each phase; however, established a set of methods as a baseline in each phase.

After obtaining the methods that fit the needs of the experts, the framework was diagrammed as a process, in order to define the real workflow that should follow a real design team. The definition of the framework is complete with the detailed approach of each method that is encompassed in it.

With this framework, the designer would take more into consideration user-centered approaches to reduce the user interfaces usability problems that the ATMs present, improving the phase of design in early stages of a software development process; and improving the user satisfaction strengthening the institutional brand.

As future works, execution of the process by a real team in a real context is necessary to validate the correctness of the framework, it is that it can be executed from start to end without complications. This would demonstrate that the framework is correct and easy to understand. Furthermore, a quantitative validation is left for future studies. The usability scale of an interface designed with the proposed framework has to be higher than the usability scale of an interface designed without a procedural or UCD approach to consider the significant relevance of the framework and formally validate it.