The user experience (UX) on female apparel e-commerce websites in Brazil

4.1.1 Exploratory interview

An open interview was conducted to understand users’ perceptions when buying clothes online; it was also used as the basis for the elaboration of an online questionnaire. The interviews were conducted using the WhatsApp application; the questions and immediately answered by the interviewees, there were 19 respondents. The interview was divided into two parts: demographic and qualitative questions.

4.1.2 Online questionnaire

Google Forms was used to elaborate the online questionnaire, it contained open questions to gather more information on how users interact with apparel e-commerce website interfaces. The pilot questionnaire was sent on 03/23/16 to 12 people (5 men and 7 women) and concluded on 04/04/2016, 10 people responded. The final questionnaire was made available from 05/05 to 02/07/2016 and was separated into seven sections.

4.1.3 Interview with specialists

To obtain specialists’ views about the issues addressed in this research, semi-structured interviews were conducted. The interviews took place between April 7th and May 24th, 2016, and 6 specialists were interviewed.

4.1.4 Interview content analysis

The interview transcripts were grouped into 8 context units, which were determined according to the Content Analysis methodology [20]. The frequency of which each unit was then determined, the results of which can be found in Table 2.

4.2.1 The usability tests

The usability test objectives were determined based on the results of the online questionnaire and the issues raised in the literature review and are outlined below.

Objective 1: Evaluate the clarity of the information presented in the size table(s). The results of the exploratory interview showed that the lack of trust in the information presented on websites is the main reason that prevents consumers from buying clothes online. The uncertainty about the fit and size and the perceived difficulty to exchange or return a product are impeding factors. Sizing was intensively addressed by the online questionnaire respondents, 81% stated that they do not buy online because they do not know if the product size is right, and 87% because they are unsure about the fit.

Objectives 2 and 3 were grouped because they deal with searching for product information: (2) To evaluate the findability (ease of finding) of the size table(s) and; (3) To examine how the product description (or its absence) influences the user’s purchasing decision. The search for information is part of the decision-making process [14, 21] when consumers intend to buy something, they analyze the brand and the characteristics of the product [21]. The degree of knowledge that the consumer has about a product directly influences their buying behavior [22, 23] as well as the intangibility the online environment poses and the lack of trust can discourage them. On the other hand, if they are well informed, they are more likely to make a purchase [24].

Objective 4: To understand the influence of price on the purchase decision. As there is no direct contact with the product prior to purchase, consumers have a greater perception of risk when buying online. Kim et al. [25] discuss that the perception of risk is directly related to the financial investment, the higher the investment, the higher is the perception of risk. This assumption was reinforced by the results of the exploratory interview and the online questionnaire.

Objective 5: To understand the influence of the complexity of the product. The type of product (shirt, shirt, pants, etc.) affects the fit. Items that are more adjusted to the body require a larger number of measurements to be constructed, and therefore need to have a larger range of sizes to fit the target audience. On the other hand, looser items can fit the same group of consumers with a smaller amount of size options [26].

The fabric of the garment has also to be taken into consideration. For example, knitted fabrics are more elastic and usually provide good fit and comfort [27]. Therefore, when developing the usability test task to fulfill this goal, users were asked to buy a knitted fabric blouse and a buttoned-up shirt. T-shirts and blouses are usually looser and more easily to fit the body, shirts are tighter and generally are produced using woven fabric and require a wider range of size options to fit a range of consumers.

Objective 6: Evaluate the findability of the exchange and return policy. Characteristics such as fabric texture and fit are poorly transmitted digitally; they are non-digital attributes [9]; clothing requires multi-sensory (tactile and visual) contact to be fully evaluated by consumers. This lack of direct contact before a purchase can lead to a lower level of consumer satisfaction and a greater perception of risk [28]. Therefore, consumers develop mechanisms to mitigate the impact of perceived risk and its negative consequences, one of which is to learn more about the guarantees offered by the brands [14, 21]. A clear and easy return and exchange policy, for example, can help improve the user experience and increase their trust.

The tasks used to carry out the usability tests were developed based on the objectives cited above. Six pilot tests were conducted between 01/20/17 and 03/20/2017, 3 on a desktop/laptop, and 3 on a smartphone.

The tests were all performed in the field; in places familiar to the volunteers such as the university campus, homes, or workplaces. The volunteers were asked about their preferred device to buy online, on a desktop/laptop, or a smartphone. They performed the test using their smartphone, or on a desktop/laptop with which they were familiar. This criterion was adopted so that the volunteers did not have to learn to use the device.

The four tasks were performed on the Dafiti website and were presented in writing. The volunteers were asked to perform one task at a time in the listed order. Once the volunteers completed the tasks, debriefing sessions were conducted to assess their perceptions regarding the usability test, and then they were asked to answer the AttrakDiff2™ questionnaire.

4.2.2 Choosing the website to conduct the usability tests

The website chosen to perform the tasks was dafiti.com.br. According to the Alexa.com and the Conversion rankings (June 2017, editions), dafiti.com.br was the top seller in Brazil. It was also the most cited website by the online questionnaire respondents.

Dafiti is a Brazilian online marketplace, the leader in Latin America with branches in Argentina, Chile, and Colombia. It sells around 110 thousand products which are divided into six categories [29]. Each product has a textual description with its name, technical description (fabric type, item code, etc.), seller identification, and photos complement the description. However, no standard is adopted, and some products had incomplete descriptions.

It markets products from an extensive range of brands; consequently, the user is encouraged to reflect on clothing size across brands. The users are also required to know their measurements when using FIT FINDER, or when referring to the various size tables. FIT FINDER is an embedded application to support users in choosing the appropriate size of clothing and accessories. Users are asked to supply information such as age, height, weight, body shape (waist, hip, bust), and their preferred brands, the application can then suggest a size. Several websites, such as Asos.com, use similar tools. Dafiti does not have physical stores; consequently, users are unable to have direct contact with the products prior to purchase.

4.2.3 AttrakDiff2™ questionnaire

The quality of products can be categorized as pragmatic and hedonic. The first refers to a product’s potential to support the goals of ‘doing’ [30]; as buying a sweater. The second is a product’s or service’s potential to support the pleasure of use and ownership, i.e., it is related to the goals of ‘being’, of how to be admired and stimulated.

Usability is a widely accepted concept [31]. ISO-9241-11:2018 [10] defines it as “the extent to which a system, product, or service can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use”. Where efficiency is defined as the “resources used in relation to the results achieved”, effectiveness is the “accuracy and completeness with which users achieve specified goals”.

The norm has been recently updated and the main changes compared to the previous edition are as follows: The scope has been extended to include systems and services; A wider range of goals is considered, including personal outcomes and organizational outcomes; Efficiency has been defined in relation to the results achieved rather than in relation to accuracy and completeness with which users achieve goals; Satisfaction has been clarified to include a wider range of issues.

Probably the deepest change was the definition of satisfaction, it is now defined as the “extent to which the user’s physical, cognitive and emotional responses that result from the use of a system, product or service meet the user’s needs and expectations” it takes into account some aspects of the user experience, and satisfaction includes the extent to which the user experience that results from actual use meets the user’s needs and expectations. It also considers that the anticipated use can influence satisfaction with the actual use. The previous definition of satisfaction only mentioned the absence of discomfort. At the time this study was completed, usability was concerned mostly with tasks, objectives, and cognitive processing of information, it did not take into account the hedonic aspect of interaction with a product, which is closely related to the UX.

Therefore, AttrakDiff2™ was chosen to measure the UX, it is a questionnaire developed by Hassenzahl et al. [32], to address emotional aspects, caused by the interaction with a product. It is a semantic differential questionnaire composed of 28 items. Seven items are meant to capture the pragmatic quality (PQ), fourteen items the hedonic quality that is divided into identity (HQI) and stimulation (HQS), and 7 the attraction (ATT) (Tables 3 and 4). A 7-point Likert scale is used to evaluate the impressions caused by the user’s interaction with the product.

The questionnaire was developed in German and has an official English version. Attrakdiff2™ was translated from English to Portuguese by the researcher and validated by two experts in Portuguese-English/English-Portuguese translation. Once validated, 3 versions of the questionnaire were generated and were then submitted to a panel of LEUI (Laboratory of Ergonomics and Usability of Interfaces - PUC-Rio) researchers, to define the final version (Tables 3 and 4).

A task analysis is the process of describing tasks and is used to guide projects as it helps to anticipate the user’s performance [33]. This process decomposes tasks in a series of steps to make predictions about how difficult or easy they are and how much effort will likely be required to perform them, resulting in a list of steps needed to accomplish a specific goal. The basic measure of task complexity is the number of steps needed to complete it.

In HCI (human-computer interaction), task analysis can be used to: analyze the current situation of a system, redesign its project, or evaluate the outcomes of an intervention. When analyzing an existing computational system, task analysis can be very realistic, describing the user’s behavior in detail [34].

In summary, task analysis is a critical way to understand how users currently do something, as well as to identify obstacles and opportunities for improvement. The way people perform tasks usually bears little resemblance to how they would like to do them or how they might be most effective [35]. Therefore it is necessary to observe them in action.

Once all the usability tests were completed, all the videos were analyzed, and the tasks performed by each volunteer decomposed into steps. It was then possible to quantify the steps each volunteer took to reach the goals of the usability test tasks (Table 5). The tasks were organized into flowcharts, which served as a basis to evaluate the volunteer’s performance.

5.4.1 Sample profile

60 volunteers were recruited, 30 performed the test on a desktop/laptop (Group A), and 30 on a smartphone (Group B). Their ages ranged from 19 to 68 years old, average 33.3, 77% had an iOS smartphone, and 23% Android. 30% of respondents were graduates, 17% of respondents in group A were undergraduates, and 23% in group B, 23%.

5.4.2 Online purchase frequency

37% of group A reported buying online more than twice a month, 23% twice a month, 17% once a month, 13% once every 2 months, and 10% once every 4 months. In group B: 30% reported buying online more than twice a month, 23% twice a month, 23% once a month, 13% once every 2 months, 7% once every 4 months, and 3% once every 6 months. None of the volunteers said they did not shop online.

In group A, 83% of the volunteers had already purchased clothes online and 67% had already visited dafiti.com.br. In group B, 70% had already purchased clothes online and 63% had already visited dafiti.com.br.

5.4.3 Usability test

The usability tests generated both quantitative and qualitative data. Among the quantitative results were: the completion or non-completion of the tasks, the time to perform the tasks, what type of sizing table was consulted, and the number of steps to perform the task. During the tests, it was noted that the time for completing the tasks was not necessarily an indication of ease or difficulty, the delay in completing the tasks was often due to deeper and more careful navigation when choosing the products in tasks 1 and 2. The average navigation time on clothing and accessory websites is almost 4 times higher than in other categories [38].

The qualitative data derived from the volunteers’ comments during the execution of the test and the answers given in the debriefing session. The volunteers were not asked to make comments during the tests but were encouraged if they felt they needed to do so; the most relevant ones will be posteriorly described. The quantitative results of the descriptive research stage will be presented as follows.

In group A, 27 people completed task 1, 26 task 2, 27 task 3, and 24 task 4. Task 4 had the highest number of dropouts, and one person did not complete the task. The task was considered “not completed” when the volunteer declared the task was finished but did not reach the intended goal.

In group B, as in group A, the majority completed all four tasks. 28 people completed task 1, 27 task 2, 27 task 3, and 26 task 4. Group B had a larger number of people who did not complete the task. In both groups, A and B, there were similar numbers and there was no significant difference between them.

In task 4, “To be sure of your size, consult the size guide,” the volunteers were asked to look for a size guide to check their measurements and see if they had made the right choice in tasks 1 and 2. There were three ways of checking the measurements: FIT FINDER, the Size Guide, and the Complete Size Guide. FIT FINDER was the most used, by 43% of Groups A and B. In some cases, when accessing the Product Page, the Size Guide link was visible, but seconds later, while loading the page, this link was transformed in the link to access the FIT FINDER. Some volunteers did not notice this change and others were a bit confused.

On the mobile and the complete website, the Complete Size Guide could be accessed from the Size Guide page. On the desktop/laptop website, it could also be accessed by clicking on the Fashion Glossary link located in the website footer.

In group A, 13% of the volunteers and 20% of Group B did not find any of the three.

5.4.4 AttrakDiff2™ questionnaire

After the usability tests, the volunteers were asked to answer the AttrakDiff2™ questionnaire (Tables 3 and 4). Each pair of words has the objective of measuring the volunteer’s perception concerning certain qualities of the product, divided into pragmatic quality (PQ), hedonic quality identity (HQI), hedonic quality stimulation (HQS), and attraction (ATT). The responses were grouped according to the quality measured (Table 6), an average score for each word pair was calculated and then an average for each quality (PQ: M = 4.0 SD = 0.45, HQI: M = 4.1, SD = 0.47, HQS: M = 4.2, SD = 0.5, ATT: M = 3.8, SD = 0.42). The results of the two groups were also analyzed separately to verify if there was a difference in the desktop/laptop and smartphone user experience. There was no significant difference between the tests performed by groups A and B regarding UX (Group A: PQ: M = 4.0 SD = 0.57, HQI: M = 4.1, SD = 0.44, HQS: M = 4.1, SD = 0.52, ATT: M = 3.9, SD = 0.39. Group B: PQ: M = 4.1 SD = 0.3, HQI: M = 4.2, SD = 0.5, HQS: M = 4.2, SD = 0.49, ATT: M = 3.8, SD = 0.45). The lowest scores were attributed to the attraction item.

After completing the tasks on the Dafiti website the volunteers were debriefed. There were 3 yes/no questions and the volunteers were asked to justify their answers. Following, the users’ most relevant answers to each question: Was the information in the table/guide clear? Yes □ No □ Why?

In group A, 40% answered that the information presented in the size table was not clear, in group B the percentage dropped to 33%. 25% of the volunteers in group A and 50% of B, did not find the tables. Among other reasons, was the absence of necessary, or expected information in the tables. “It could indicate how to take measurements, perhaps with a drawing of the body, because that part of the measurements always confuses me,” “I think weight and height are factors that can vary a lot, they do not necessarily indicate size. I prefer when there are measurements in cm.”

Other reasons for the negative answers were: the subjective perception of the measurements, “Because they are subjective measurements and what the other sees is different from my image in the mirror”. There was a volunteer who was not able to use the measurement system because of her height, there was no size available for her, “I do not know. I could not access it. Maybe it was because of my height (1.83 m)”. And there were still those who were confused about using the FIT FINDER system.

Depending on the product the volunteers selected; a different measuring method was presented to them, either the FIT FINDER or a link to the Size Guide.

60% of the volunteers in group A and 67% from group B answered positively to the question. For both groups, the appropriate information was the main reason for the positive perception: “Because I had the necessary information”, “Because the informed measurements were enough to know the size of the product”.

Some volunteers thought the information was clear but were not sure about their measurements: “But the result was wrong”, “The data were clear, the measurements, but I felt insecure about the accurateness of the results”. Besides the responses of who did not find the information in the size tables: “But I believe that the measurement option may not be so practical since I need time and a measuring tape.”

Do you trust the fit information? 60% of the volunteers in group A and 83% in group B responded negatively to the question. The main reason for the negative response in group B was the lack of identification with the photos, 40% of the answers, “Because the photos are taken on models, their stereotype is very different from real buyers”, “Not much, because in general, the photos of the models do not resemble my physical type.”

40% of the volunteers in group A, and 17% in group B were satisfied with the information about the fit of the products. Among the reasons was the identification with the photos, “Because the photos are very informative about the material and fit”, “With the image available it was possible to have an idea, but it would be better to have the clothes dressed on a person”.

It is also worth highlighting the responses of those who trusted the information provided by the website, “I know myself, and I feel secure about the information on the website”, “Because the suggested size usually matches the sizes I usually buy”.

Based on the information available, are you sure the chosen size is correct? From group A, 47% of the volunteers and 60% from group B gave negative answers to this question. The volunteers did not have enough information to be sure that the chosen size would fit them, “Because there is no way to know how it will look on your body from the given information,” “Not exactly. Because I have already made other purchases on the website that went wrong, precisely because they did not have enough information about the material/fabric of the clothes etc.”

In addition to not having enough information, other volunteers did not trust what was provided, “As much as I find my measurements in the (size) table, I always have doubts about manufacturing variations”, “Because measurements may be wrong, or information may be missing.”

Other reasons less frequently mentioned were: the perception of the difference in sizes among brands and the necessity of having to try on the product to be sure. “Because size varies a lot from brand to brand,” “I would only be sure after checking it when the product arrives at my home.”

From group A, 53% of volunteers and 40% of group B gave affirmative answers to the question. 63% of group A answered yes to this question because they felt trusted the information provided: “Because I found the size guide detailed enough”, “Because I trusted the information”. Only 17% of group B gave similar reasons.

In group B, the most frequently cited justification for the affirmative response was the use of prior purchase experience when choosing the size, 58% stated that they relied on their purchasing history.

The other debriefing questions are highlighted below.

The information on the website was: SUFFICIENT –INSUFFICIENT. Participants’ responses were based on a 7-point Likert scale ranging from 7 (sufficient information) to 1 (insufficient information). The mean was 5.2 and mode 6 for group A, and the mean 5 and mode 6 for group B. These results demonstrate that the general perception of the volunteers was that the website offered enough information. However, the responses to the other questions show otherwise.

Were the exchange policies clear? In group A, 67% thought that the Exchange and Returns Policies were clear and 87% in group B. It was observed that, during the usability test, some of the volunteers who answered negatively to this question did not find or did not read the policies.

Is price a decisive factor when buying online? 97% of the volunteers in group A and 67% in group B answered yes to this question.

What other factors influence your purchasing decision? Other financial factors, such as the shipping cost and the possibility of installment payments, were the most cited. Another highlight was the convenience that online shopping provides as well as offering online exclusive products.

How did you choose your size? The issue of size and fit was widely discussed by fashion specialists and users. 63% of Group A volunteers and 83% of group B used information obtained from previous purchases to choose their sizes: “I usually know that my size is L, so I selected the products without checking the size table”, “ Based on previous purchases”. 13% of both groups used the resources offered by the website (such as the Size Guide and FIT FINDER): “Based on the estimate offered by the website despite not being sure”, “By the product description”. 13% of group A stated that they had used both their previous purchase history and the resources offered by the website. Other answers ranged from “By intuition” to “Guessing”.