Ensuring Insurance

(1)

IN

DEGREE PROJECT COMPUTER SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020,

Ensuring Insurance

A study of user expectations on trustworthiness

of websites for insurances

DAN STRANDBERG

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE

(2)

English title

Ensuring Insurance

A study of user expectations on trustworthiness of websites for insurances

Swedish title

Försäkra Försäkring

En studie av användares förväntningar på pålitlighet hos hemsidor för

försäkringar

Author

Dan Strandberg

danstra@kth.se

Submitted for the completion of the KTH program;

Industrial Engineering and Management, Master of Science in Computer Science

Supervisor: Adrian Benigno Latupeirissa, KTH, School of Electrical Engineering and

Computer Science, Division of Media Technology and Interaction Design

Examiner: Dr Sandra Pauletto, KTH, School of Electrical Engineering and Computer

Science, Division of Media Technology and Interaction Design

Date of submission: 2020-03-13

(3)

ABSTRACT

The rapidly advancing technology and Internet has opened up for many industries to offer e-commerce

solutions. While an increasing amount of industries are offering solutions over the web, the quality of these

solutions are increasing both visually and functionally. Simultaneously, users are becoming increasingly

accustomed to these increasing standards and it is therefore crucial to keep up with these standards in order

to seem trustworthy in the eyes of the customers. The insurance industry is starting to implement e-

commerce solutions in this increasingly digital world. However, since the expectations on an insurance

company differ from other e-commerce industries, this thesis presents a study on what the people,

specifically Digital Natives, expect from an insurance company in terms of trust and reliability. Based on the

research findings, the goal of this study was to identify and understand the cause of the users preconceived

idea of what trust on an insurance website look like. This was done by reading existing research articles and

reports, sending out a survey, conducting user testing with an existing insurance websites followed by an

evaluation of designed prototype. The results from the survey and the first user testing identified several

factors and characteristics that the users' based the conception of trust towards an insurance company on.

These factors and characteristics were further tested and verified in a designed prototype and the findings

were discussed. Lastly, specific factors of trust for insurance companies were highlighted and a conclusion

was made.

SAMMANFATTNING

Den snabbt avancerande tekniken och Internet har öppnat upp för många branscher att erbjuda e-

handelslösningar. Under tiden som fler industrier erbjuder lösningar på nätet ökar kvaliteten på dessa

lösningar, både visuellt och funktionellt. Samtidigt blir användarna mer och mer vana vid dessa ökande

standarder och det är därför viktigt att hänga med dessa ökande standarder för att uppfattas som pålitliga i

kundernas ögon. Försäkringsbranschen börjar implementera e-handelslösningar i denna alltmer digitala

värld. Eftersom förväntningarna på ett försäkringsbolag skiljer sig från andra e-handelsbranscher presenterar

denna rapport en studie av vad människor, framförallt Digital Natives, förväntar sig av ett försäkringsbolag

när det gäller förtroende och tillförlitlighet. Baserat på resultatet av forskningen var målet med denna studie

att identifiera och förstå orsaken till användarnas förutfattade mening om hur försärkingshemsidor inger

förtroende. Detta gjordes genom att läsa befintliga forskningsartiklar och rapporter, insamling av svar från en

enkät, användartester med befintliga försäkringsbolags hemsidor följt av användartester med en designad

prototyp. Resultaten från undersökningen och den första iterationen av användartester identifierade flera

faktorer och egenskaper som användarnas baserade förtroendet för ett försäkringsbolag på. Dessa faktorer

och egenskaper testades och verifierades vidare i den designade prototypen varvid dessa fynd diskuterades.

Slutligen lyftes specifika förtroendefaktorer för försäkringsbolag fram och en slutsats gjordes.

(4)

Ensuring Insurance

A study of user expectations on trustworthiness of websites for insurances

Dan Strandberg

KTH Royal Institute of Technology

Stockholm, Sweden

danstra@kth.se

ABSTRACT

The rapidly advancing technology and Internet has opened up for many industries to offer e-commerce solutions. While an increasing amount of industries are offering solutions over the web, the quality of these solutions are increasing both visually and functionally. Simultaneously, users are becoming increasingly accustomed to these increasing standards and it is therefore crucial to keep up with these standards in order to seem trustworthy in the eyes of the customers. The insurance industry is starting to implement e-commerce solutions in this increasingly digital world. However, since the expectations on an insurance company differ from other e-commerce industries, this thesis presents a study on what the people, specifically Digital Natives, expect from an insurance company in terms of trust and reliability. Based on the research findings, the goal of this study was to identify and understand the cause of the users preconceived idea of what trust on an insurance website look like. This was done by reading existing research articles and reports, sending out a survey, conducting user testing with an existing insurance websites followed by an evaluation of designed prototype. The results from the survey and the first user testing identified several factors and characteristics that the users’ based the conception of trust towards an insurance company on. These factors and characteristics were further tested and verified in a designed prototype and the findings were discussed. Lastly, specific factors of trust for insurance companies were highlighted and a conclusion was made.

Keywords

User Experience; User Interface; Trustworthiness; Evaluation INTRODUCTION

With fast advancing technology, design concepts such as user experience (UX) and user interface (UI) have become increasingly common over the past few decades [1]. Not only are our technological solutions able to solve more varied and compli- cated problems — they are also becoming more interactive and fashionable. Due to the improving interactivity and visual appeal, the website users are getting increasingly accustomed to the improving standards and are thus starting to expect similar, or improved, designs from all types of internet services [2].

In order to excel online, website owners need to acknowledge and meet these increasing customer expectations, such as good appearance, simplicity and vividness [3, 4].

While an increasing amount of websites are improving their design with an UX and UI awareness, not all of them look exactly the same. Neither should they, because according to a study by Ping Zhang and Gisela M. von Dran, users expect different elements from websites depending on their domains [5]. However, even though many websites are within the same domain, such as e-commerce, with similar purposes, they may yet look significantly different. Using e-commerce as an example, a website for online retailing (e.g. Webhallen¹) may be similar structure-wise compared to another retail website (e.g. Elgiganten²), which is illustrated in figure 1.

Figure 1. Comparison between the design of Webhallen and Elgiganten.

These websites, however, look significantly different compared to a website from another area within e-commerce, such as insurance services (e.g. IF³) which on the other hand is visibly similar to another insurance website (e.g. Folksam⁴).

Because of these similarities, a question that arises is whether

1https://www.webhallen.com 2https://www.elgiganten.com 3https://www.if.se/privat 4https://www.folksam.se

(5)

there is a preconceived idea of what users expect of a website based on its branch of industry.

The purpose of this thesis is to investigate what the average user’s preconceived idea of what they expect an insurance website should include and if the existing design of insurance websites is in accordance with this preconception. Moreover, since an insurance company in the end wants to sell a security, it may be crucial that the company itself can convey a sense of assurance and trustworthiness in order for the customer to feel comfortable entrusting the company with any kind of insurance service [6]. Therefore, to narrow the thesis scope down, the thesis aims to answer the following research question.

How should a website of an insurance company be designed in order to convey trustworthiness for the average insurance customer?

BACKGROUND

The rapid advancement of the Internet and the e-commerce environment has paved the way for detailed studies of the relationship between business and the customer. The following section consists of summaries of research deemed relevant in order to answer the research question.

Visual Perception of Trust

Visual perception has long been acknowledged as an important aspect in human-computer interaction [7]. Research imply that users make reliable judgements of visual appeal, perceived usability and trustworthiness within rapid duration of as little as 50-500 milliseconds [8, 9]. Since trustworthiness is subject to halo effects, the phenomenon of cognitive bias based on good or bad qualities, in this short time window the first step to convey a sense of trustworthiness towards the user is therefore proper visual appeal. There are multiple visual aspects that can affect the perception of trustworthiness, such as the color scheme, typography and content alignment.

The fastest and likely easiest cue to manipulate is color [7].

The psychological effect of color has been studied within multiple of disciples throughout the years [10]. Color is an aesthetic part of the aspectual stimulus and has an influence on people’s affect, cognition and behaviour and, consequently, people’s perception of trust [7]. This was demonstrated in an experimental study by [11], where it was shown that a website’s color scheme had an effect on a user’s perception of trust towards that site. Furthermore, the study also showed, while small in significance, that a blue color scheme was perceived as more trustworthy and contrariwise for a black color scheme. Additionally, in another study exploring the color-trustworthiness relationship it was identified that higher color saturation has a negative impact on the user evaluation of trust [9]. Thus, in addition to some color schemes being more desirable than others, a lower and contemporary level of saturation should also be considered in order to increase the visual trustworthiness. However, it is important to keep in mind even visual preferences may differentiate between cultures which in turn can affect the perception of trust through colors [11].

Trust in E-Commerce

[12] argues that without good experience, there can be no trust as trust forms the base in the hierarchy of customer experience. The hierarchy of customer experience is an adaption of Maslow’s hierarchy of needs and it implies that the lower- order needs of a customer must come before those of a higher order can be fulfilled. What this means is that without trust, not even the most basic transaction can take place. According to [12], trust is initially established through a visual evaluation followed by a behavioural evaluation. Namely, the user first determines if the website looks trustworthy and then evaluates if the websites behaves trustworthy. Furthermore, there must also be a seamless interplay between visuals and functionality.

Thus, a customer’s perception of trust relies on the thorough quality of the website [13]. The level of trust, however, depends on the value of the exchange [12]. For example, there is a lower level of trust needed for a customer buying candy from a grocery store than buying a car from a car dealer, or in this case buying an insurance from an insurance company.

Trustbuilders and Trustbusters

Communicating trust towards customers is not an easy task as many customers likely have different levels of e-commerce experiences [14]. For example, reputation and familiarity are more important elements of trusts for inexperienced e-commerce customers. This means that these customers are more likely to be drawn towards established companies through a mouth-to-mouth trust transfer. Meanwhile, experienced customers, who likely have navigated a multitude of e-commerce sites and will thus be able to compare them to each other, may draw more help from interface and usability qualities. However, since experienced customers value interface and usability qualities and inexperienced rely on reputation and familiarity, any negative experience with either poor interface design or lack of usability on a company’s website will consequently affect both customers directly and indirectly.

According to [14], there are interface factors that has been shown to either build or destroy trust. These are called Trust- builders and Trustbusters. Trustbuilders are factors such as status indicators, order tracking, recourse and trial runs while Trustbusters are factors such as poor usability, inconsistent design and long system response time. While most Trustbuilders and Trustbusters are tied to the interface design, the strongest factors are outside the interface. These factors are for example brand and reputation. However, [14] adds that trust perception strongly depends on personal and cultural factors and that the identified Trustbuilders and Trustbusters may change over time.

The Model of Trust for E-Commerce

The model of Trust for E-Commerce, a model developed and refined by Florian N. Egger, provides a framework making explicit factors likely to affect customer trust in a business- to-consumer e-commerce environment [15]. The model is comprised of four different dimensions — pre-interactional filters (e.g. industry preconception, marketing), interface prop- erties (visuals, branding), informational content (sense of com- petence, risk mitigation) and relationship management (pre-

(6)

and post-purchase experience). Moreover, the different dimensions of the model can be reflected into three types of design principles [16]. These principles imply that trustworthiness can be communicated before the website has been accessed, during and after the website has been interacted with. This means that in order for proper trustworthiness to be conveyed, an e-commerce company must consider the branding of the company, the UI and UX design of the website as well as customer service and fulfilment [16]. All these factors together is a necessity to enhance purchase intentions and customer retention [12].

Digital Natives and Digital Immigrant

People of different generations with different technological experience are often termed as either Digital Natives and Digital Immigrants in several research articles [17]. Digital Natives, defined as people born or brought up while surrounded by digital technology, are more accustomed to receiving information at a fast pace. They have also shown a liking in parallel processing and multi-tasking as well as also preferring their graphics before their texts rather than the opposite [17]. Due to this, they can be seen as native speakers of the digital language and will therefore think and process information completely differently from their ancestors.

Digital Immigrants, conversely defined as people born after the emergence of widespread use of digital technology, can be seen as adaptors to the digital environment. While they learn to adapt to the new technologies, they will always retain artefacts of the past [18]. Due to this, Digital Immigrants are believed to adapt to new technology at a slower pace than Digital Natives [18]. Since the Digital Natives can be seen as native speakers of the digital language, a comparison can be made with learning a language at an older age. Research has previously shown that if a language is learnt at a later time in life, it will be stored in a different part of the brain [17].

Beyond differences in experience specifically with e- commerce, experience with technology in general also have an effect on user perception of trust [19]. According to [19], Digital Immigrants tend to look at risks and benefits and they usually pay relatively little attention to brands or recommendations, presumably because they are less familiar with online services. Digital Natives, on the other hand, rely nearly exclusively on large, established, and familiar brands with a large customer base, preferably specialised online providers because an offline presence negatively affects the trusting beliefs of this group.

METHOD

In order to be able to answer the research question and possi- bly identify certain important characteristics of an insurance website, a combination of both a qualitative and quantitative approach was used. The qualitative data was mainly gathered through user testings while the quantitative data was gathered through a survey that was distributed online.

Survey

The survey was distributed early in the research process to maximise the spread and to contribute to the design of the user

testings. Content-wise, the survey was divided into two parts.

The first part focused on general questions about insurance e.g.

why the respondents chose their current insurance company as well as how they obtained it. In the second part, the survey respondents were to look at the front page of 9 different Swedish insurance companies’ websites and grade, on a scale of 1 to 7, each website’s trustworthiness based on different characteristics such as color and structure. They were also able to further comment on their grading choices.

Furthermore, each of the websites had had the company name and logo censored throughout the image in order to minimise biased answers due to brand familiarity. The purpose of the survey was to quantitatively complement the qualitative data gathered from the think aloud user testing to see if a general opinion could be identified.

User Testing

The qualitative data was gathered through participatory user testing that implemented the think aloud method. The think aloud method was chosen as it reveals hidden thoughts and gives an insight in the users mental model [20]. This was especially interesting in the study as peoples’ perception of trustworthiness is highly subjective. Thus, the think aloud model provides more elaborate insight to why the respondents answered the way they did.

There were two iterations of user testings, with the first iteration aiming to evaluate the design of existing insurance websites and the following iteration for evaluating and reevaluating a prototype design. During the first iteration, the users navigated and executed different tasks on three different Swedish insurance websites while sharing their thoughts of the process.

Each user test lasted, at longest, one hour with roughly 15-20 minutes per insurance company. The second iterations had similar tasks but on a prototype website, which was designed based on the results of the previous iteration along with the answers from the survey.

The user tests were voice recorded with the computer on which the test was conducted and the results were later fully tran- scribed, coded, and analysed (through content analysis). The following figure illustrates the process and each of the research iterations.

Figure 2. Illustration of the research process and iterations.

Target Group

Due to digitisation, the way different generations use and manage their insurance services is different depending on their experience and expertise with technology. Therefore,

(7)

the study focused solely on users who are more accustomed to digital technologies, the so called Digital Natives. In this thesis, people within the ages 20 to 30 are deemed as Digital Natives due to the fact that most within these ages grew up with and during fast changing and developing technologies and are therefore likely more accustomed to technology and online services.

Additionally, while gender may or may not have an effect on participants perception of trust, the purpose of this thesis was not to explore the differences in genders’ perception of trust and thus gender was not accounted for.

RESULTS

In this section the most interesting and relevant results from the surveys and the user testings are presented.

SURVEY

A total of 100 people, aged 22 to 28 with various occupations answered the survey. While both genders were represented in the survey, there were no intention to identify any gender specific behaviour towards trustworthiness. All respondents also lived in Stockholm. The questions asked in the survey mainly concerned visual cues of trustworthiness, but also questions regarding general insurance usage. At the end of the survey, the respondents were to vote on which website they preferred the most.

General Insurance Usage

A total of 92 respondents had or have had some kind of insurance and were thus able to answer the follow-up questions about their general relation to insurance services. Those who had none simply proceeded to the next part.

Among the follow-up answers, where the respondents were students, which were 32 of them, price was the most decisive factor when choosing their current insurance service. The students either chose the alternative with the lowest monthly fee or chose to have no insurance at all. However, for those whose occupations were other than students, the most common reason was that they had chosen the insurance company that their parents were using or had been using, which more often than not were one of the bigger insurance companies in Sweden (e.g. IF and Folksam).

The follow-up respondents were also asked how often they visited their respective company and 76 of the them said that they navigated their website once or twice per year. The most common reason that they visited the website was either to take out an insurance, take advantage of the insurance (such as reporting a damage) or if they had to take part in any important or urgent information. Furthermore, 71 of the follow-up respondents said that they did all their insurance errands digitally.

Additionally, 24 out the 71 who did their insurance errands digitally had to, at one point, call an agent of the company in order to gain more information.

Visual Cues

After the first part of the survey, the respondents were to look at an image of the front page of an insurance company’s website and judge different characteristics with a grading scale from 1

to 7. While some characteristics had little to no effect on the the respondents perception of trustworthiness on the website, there were four characteristics that had the most apparent effects — color, structure, imagery and brand familiarity. At the end of the survey, the respondents were to choose three of the websites which they preferred the most.

Color

One of the strongest decisive factors on whether the survey respondents deemed an insurance website to be trustworthy was the website’s choice of color. The answers showed that the insurance websites that used either blue or green colors as the base color were deemed even more trustworthy over the ones that used other colors. Additionally, beyond some colors performing better than others, so did different saturation of the same color. An insurance company that, for example, used a blue color with a lower saturation on their website scored higher in trustworthiness than one that used a higher saturation.

Structure

Another aspect that had a distinguishable effect on respondents judgement on trustworthiness was the structure of the website. The most common structure among the insurance companies was a symmetrical, centre aligned orientation with evened white spaces and mostly same sized boxes and icons.

There were, however, a selection of websites that did not have its content completely symmetrically aligned. which consequently showed a negative impact on the grading. In these cases, respondents commented that the websites looked messy and unprofessional. A mutual choice among most of the respondents when it came to trustworthiness in structure was websites that used symmetrical and clean structure.

Imagery

While the preference of the type of imagery differed between respondents, there were some characteristics that were more preferred than the others. A common characteristic on the websites where the imagery had had a negative effect on trustworthiness was that the imagery, especially icons, had a playful character. In such case, respondents commented that the they felt that the imagery was too childishly and frivolous. A vast majority of the respondents’ image of an insurance company was indifferent to that of a banking service. Due to this, they expressed that the use playful imagery therefore offset the seriousness of such services. An example of playful as well as realistic imagery that occurred in the survey can be seen in figure 3.

Figure 3. An example of playful (left) and realistic imagery (right).

(8)

Additionally, the insurance company that received the most positive feedback used real life pictures (e.g. pictures of real people, cars etc) instead of graphical imagery (like in the ones shown in figure 3). However, when they motivated their choice they merely said that it was due to the calm and cold characteristics of the site and not explicitly the imagery.

Brand Familiarity

Despite the fact that all website images had their respective company names and logos censored, some respondents were still able to recognise and identify some, mostly the bigger ones, of the insurance companies either due to their signature brand designs or the fact that they were a customer of one. In the cases of brand recognition, an insurance company received a higher trustworthiness rating from a user that recognised the company compared to from one that did not.

“The company has been around for a while, so I know that it is a solid insurance company.”

- Survey 74 However, the positive effect of brand familiarity was most apparent in the answers where the participants recognised their own insurance company. The average difference in trustworthiness rating due to brand familiarity was roughly 19% in the cases where the participants recognised their own insurance company and 8% when they recognised a company that was not one of their insurance provider.

USER TESTING

The following section presents the results from the two user testings, divided into two iterations. A total of 20 users participated in the first iteration and five participated in the latter.

All of the test participants had already answered the survey.

In the testings, the participants were to conduct five different tasks on three different insurance websites. These five tasks were finding the range of services offered by the company, start the process to purchase an insurance, signing in on personal pages, reporting a damage and finally reaching out to customer service. In order to respect each participants pri- vacy, they did not at any time have to provide any of their own personal information and was instead provided substitute information when necessary during the testing. The three insurance websites that were used in the user testings were the same for every participant. They were specifically selected based on the results from the survey in order to further identify the most distinct visual cues of trustworthiness as well as exploring further trust dependent aspects, such as log-in authentications and purchase processes.

First Iteration of User Testing

The user tasks along with the questions asked in the first iteration mainly focused on trust dependent functions. However, the users were also able to freely comment on the visual cues throughout the websites as well, such as the use of color, imagery and structure choices. The three insurance companies that were tested in the first iteration were selected based on how familiar their brands were and on their differences in visual characteristics (e.g. choice of imagery, structural alignment, color and saturation).

Visual Cues

While most of the comments regarding the visual cues that were already mentioned in the survey, the participants had the opportunity of elaborate and explain their reasoning behind their opinions.

Colors

When the participants were to comment on the choice of colors, only three of them explicitly mentioned anything regarding the choice of hue, e.g. blue or green. While only three out of 20 specifically mentioned favourable colors, most of them expressed that the colors felt calm and relaxing. Additionally, the majority of the participants said that they liked that the websites were using a low saturation as it conveyed a sense of professionalism and seriousness. They added that they found cold, grey-toned colors more suitable for insurance companies as they expect an insurance company to be and look professional and formal. Some of the participants even compared their expectations of an insurance company of that to a bank or principal authority.

“Basically, I look at an insurance company similarly to how I look at a bank or even a principal authority.”

- Test participant 11 Imagery

In the user testing, the playful characteristics of the imagery was again a recurring issue. 13 out of 20 participants further emphasised that the playful imagery felt too childishly and therefore made the insurance company look less serious.

Some, however, did express somewhat of an understanding towards why the insurance company may have chosen a playful design. They expressed a belief that the insurance company may have intended that the playful design was supposed to minimise stressful situations and hopefully calm down customers in vulnerable situations. However, they also expressed that other companies achieved the same intended results without a childishly approach, such as using real life pictures and low saturated colors.

Structure

Similarly to color and imagery, most of the participants expressed similar concerns about the structure and alignment of the website’s content in the testing as in the survey. However, in the user tests they were able to give more detailed opinions on the matter. 15 out of 20 test participants expressed that a symmetrical design conveyed professionalism and therefore trust. Conversely, the single website in the interview session that had an asymmetrical design was thought as unprofessional and messy. Many of the test participants expressed that it looked as if the company had no clear plan and structure of the website and/or did not prioritise the development of the website service.

“I don’t feel that the website is very thought-out with multiple and uneven white spaces. It looks like the company just threw

in stuff on the go.”

- Test participant 13

(9)

Trust Dependent Functions

The most apparent findings from the trust dependent functions are presented below. There were mainly two trust dependent functions that were tested out as these were deemed easiest and most reliable to evaluate.

Purchase Process

The first trust dependent functions that were tested, and perhaps the most vital function of an insurance company, was the process of taking on an insurance. The users were to find the company’s range of services and from there take on a home insurance. Before the users went through the process of purchase, a common opinion among the participants was that they wanted each step of the process to be as clear as possible, since they were in fact buying a long term service. They wanted and expected that the price and content of each service, as well as any additional services, was clearly visible and presented with sufficient information. Some also added that while they wanted information about each choice and alternative, they did not want to be overwhelmed by too much information and would rather have elaborated information only on demand (e.g.

a drop down box or similarly).

Another common opinion was that they wanted to feel in control of the process. This was expressed during the testing of one of the insurance companies where the position of the website view jumped to any selected text input box. In this case, many of the participants expressed that they did not like the jumping and some added that they felt out of control and therefore slightly stressed out. When they later conducted the same task on another website which did not have the

“view jumps”, they expressed a relief and complimented the smoother flow of the process.

However, during the purchase process, none of the participants expressed any concerns about the visual representation of the purchase process. This was especially interesting during one of the insurance companies where the visual design of the purchase process differed significantly to the overall design of the website. The company were using an older iteration of the the website design (verified through the Internet Archive service⁵) for it’s purchase process, as visualised in figure 4.

Figure 4. An example of input boxes with the same purpose with two different user interfaces from the same website.

When the participants were asked to share their thoughts on the inconsistency of the visual design, they expressed that while it

5http://web.archive.org

necessarily did not look visually appealing, they were kind of forgiving on the matter. This was because they knew that the company had been around for a while and understand that the purchase process likely are not easily changed due to technical complications. Some also added that it felt somewhat like a contract form which were somewhat expected since they were signing a contract.

“I guess it is a challenge to redesign this part since it is probably built on some old database architecture.”

- Test participant 9 Authentication

Apart from taking out an insurance, the test participants were also later tasked to find and sign onto "My Pages" of each insurance website using a preselected personal registration number.

The perhaps most interesting realisation from this task was the participants’ attitude towards the authentications when signing in on the websites. All of the test’s insurance companies’

log-in authentications relied on the Swedish BankID service

6. Design-wise, while two of the three insurance websites had a well designed authentication procedure, the third one had nothing else except the BankID symbol visible.

However, despite this visual difference, when the log-in authentication was tested, 18 out of 20 participants did not express any particular concern with how the log-in authentication looked as long as the BankID logo was visible. When the participants were asked to share their thoughts on the design of the authentication site, they merely expressed that it might have been more of a fashion improvement rather than a trust improvement as all of the participants already relied on the credibility of the BankID service. They added to that they understood that most services that rely on BankID probably already outsource their authentication process to companies that has experience and knowledge with security procedures.

There were, however, seven participants that expressed that they would have preferred that the company at least had put its logo and company name on the authentication site and web address. Otherwise it felt as if they were completely handed over to another company, instead of the insurance company they intended to sign in to. In spite of this, they and all of the other participants said that since the authentication was done through the BankID service, they believe that their personal information likely is secure and protected anyway.

Usability

In addition to evaluating the trust dependent functions and visual cues of trustworthiness from the websites, the user testing were also able to identify a couple of usability issues. Most of the usability issues had little to no impact on the users’

trust towards the company. There were however one particular usability issues that directly affected the users’ perception of trust and that was the process of getting in contact with the companies’ customer service. In the cases where the participants faced difficulties to find customer service or when the process included too many clicks, the users said that it felt as if the company intentionally tried to mislead them in order to

6https://www.bankid.com/en/

(10)

avoid unsatisfied customers. This in turn made them mistrust the company’s customer service and consequently made them less likely to consider becoming a potential customer of that particular insurance company.

Second Iteration of User Testing

The purpose of the second iteration was to further test the most apparent factors on trustworthiness that had been identified. A clickable prototype website was built based on the results from the first user testing in combination with the survey data. The purpose of the prototype and the second iteration was mainly to test and attempt to verify the most apparent and interesting findings. Additionally, there were also multiple version of the prototype that had minor alterations to structure, saturation and authentication with the intention to test and potentially verify the distinct factors of trust building.

The five people that participated in the second iteration had also participated in the first iteration and the clickable prototype’s design was based on the insurance company that got the most criticism and was least recognised in the first user test iteration. This was done to see if the alterations would provide any improvements as well as minimising the bias of brand familiarity.

Saturation

While no specific preferable color was explicitly mentioned, color saturation was shown from both the survey and the first iteration to have had a decisive impact on the insurance website’s trustworthiness. One of the altered characteristics were therefore slight differentiation in the saturation of the prototype’s colors. There were two different saturation version, working basically like a filter which could instantly be switched on or off throughout the user testing. The intensity of the saturation were similarly to the ones illustrated as number 2 and 4 in figure 5.

Figure 5. An example of a saturation scale for purple.

The purpose of this was not to identify one specific grade of saturation but rather to verify the preference of a lower saturation. During the user testing, the users were to decide which saturation palette they preferred and elaborate on that decision. All of the five participant agreed that the preferred the lower saturation as the cold feeling felt more suitable for an insurance company. However, three of the participants expressed that it might have been a bit too lowly saturated and

would probably have preferred a slightly higher saturation as they would like to see a higher contrast with the background.

Structure

The website that the prototype was based on used more of an asymmetrical structure with several uneven whitespaces, as illustrated in figure 6. Similarly to the the alternation between saturations, the prototype were also able to alternate between different structural alignments on demand.

Figure 6. Illustration of the same content with two different structural approaches.

All of the five participants mutually agreed that the symmetrical structure looked much more trustworthy as it conveyed that the company had thoroughly thought through the content of the website and thus seemed more reliable in that case. Four of the participants also added further that the asymmetrical structure made the website look messy and outdated.

Purchase Process

Originally, the company that the design was based on had the same user interface design on purchase process as the rest of the website. However, the prototype intended to explore the differences between the use of an older interface and a newer.

During the prototype testing, the users were able to alternate between the an “older design” and a “new design (similarly to what is visualised in figure 4). All of the users explicitly mentioned that they naturally preferred the new and modern design, but that it really did not matter when they tried out the older one. The older design made it look like that the insurance company was an older and established company and therefore likely reliable and capital heavy.

Authentication

For the authentication process, the prototype had two different alterations. The aim was to see how big of an impact the BankID brand name had on the perception of trustworthiness as well testing out the importance of a consistent design. There were two versions of the authentication process, one with a lacking design similarly to the original and one with a more consistent user interface. Additionally, both of the versions used a fictional authentication service in order to remove the comfort of the BankID brand.

This showed to have a significant effect on the users trust on the authentication process. In the case the users became

(11)

more wary when navigating the version that had a lacking user interface as they neither knew of the insurance company nor the authentication process they were supposed to use. The users expressed that while they probably should have read up on the authentication service before potentially using that service, the version with a well designed visual representation looked significantly more trustworthy.

Usability

While the prototype included multiple usability and quality of life improvements, the only improvement that showed a considerable effect was the process of contacting customer service. However, three of the test participants said that while a good user experience is much more preferred, it did not really matter in the end as long as the main goal was reachable in one way or another.

DISCUSSION

The purpose of this thesis was to explore users’ expectation on user experience of insurance websites through an evaluation of existing Swedish insurance websites and followed by conducting tests on a designed prototype. With consideration to the research question, the findings imply that people, specifically Digital Natives within the age 22-28, do in fact have a rather similar view on insurance companies. Additionally, their mutual view were seemingly affected by several common factors.

Industry Impact

The impact of the industry showed an interesting effect in both the survey and the user testing. A common justification to why participants thought the way they did was due to the nature of the company. Whether it was visuals or functionality, participants often expressed that they expected the insurance companies to feel like an insurance company — professional and serious. Furthermore, the professional and serious impression was preferably conveyed through low saturated colors, verifying the findings by [7], a well structured and symmetrical content alignment and proper choice of imagery. The imagery, however, was perhaps one of the more interesting aspects in this regard. A possible reason for that could be the ages of the participants. Considering that all the participants were within the age of 22-28 and thus young adults, the dislike against childishly imagery could be because they want to be treated like adults and not coddled with. This, however, is merely a subjective speculation.

Moreover, not only did the industry affect their judgement of the visuals characteristics of the website, it also entailed a forgiving attitude among many participants during the user testing. While [14] argue that an inconsistent design is a proven so called Trustbuster, very few of the participants lost any sense of trust towards the companies that showed a design inconsistency. This, however, goes well along with what [12] argued, that the level of trust depends on the value of the exchange. Still, one could question whether the level of trust should be higher or lower for an insurance company. A possible reason for this could be the fact that all of the Swedish insurance companies are obliged to follow certain Swedish laws and regulations. Therefore, the users may automatically

have gained an increased perception of trust towards this company as they knew that they were likely protected by these mandatory guidelines. As a result of this, the users could have been less critical in their judgement towards the different aspects of the websites, such as the authentication process and purchase process. A possible different approach to mitigate the impact of industry preconception could have been including websites from other industries (e.g. online retailing).

Brand Familiarity

[14] stated that brand familiarity is one of the strongest builders of trust which also was evident from both the survey and, especially, the user testing results. While the visual appeal of the insurance websites had an apparent effect on the participants perception of trustworthiness, most of the participants made it clear that the brand’s reputation undermined this effect. During the testing, the company that was criticised the most in terms of visuals was in fact the company which none of them were familiar with. This can likely be related to the halo effect phenomenon, where the participants more or less overlooked the bad qualities of well know insurance companies due to the their reputation [21].

Considering that all of the participants, regardless of its visual representation, had no issues with each of the insurance companies’ authentication processes because of the presence of the BankID logo, one could argue that this also is an effect of brand familiarity. This insight was most apparent and interesting during the testing of the insurance company which basically had nothing else, not even the company logo and name, other than the BankID logo visible when the users were choosing a method to sign in. However, it could and should also be considered that this, rather naive, trust on the BankID could have rooted in the fact that the users’ did not have to use their own personal information. Consequently, when the users were signing in they were and would not have been exposed to any personal risks if the authentication would have been a fraud or compromised and thus could have had a more relaxed attitude towards the process as it had no direct consequences.

Thus, while all of the participants showed a distinct trust towards the BankID service, the extent of it may be limited. On top of that, the attitude may not exclusively be correlated to the familiarity of BankID but also, as previously discussed, to the fact that they are signing on to an insurance service.

While all of the participants were Digital Natives and the results may only be applicable to them, one can question whether this attitude is exclusive for Digital Natives or not [19]. implies that the Digital Natives, as also evident from the results, are likely to rely on brand familiarity and established organisa- tions when judging a website trustworthiness. However, [19]

also mentions that Digital Immigrants’ tendency to focus on risks and benefits depends on unfamiliarity of brand. In this case, since a majority of the insurance companies are likely familiar to both Natives and Immigrants as well as the BankID service, Digital Immigrants may as well show a similar attitude towards the authentication. This, however, likely differentiates between Digital Immigrants depending on their experience with using online services instead of traditional ones.

(12)

Regardless, since the BankID service is exclusively a Swedish authentication solution and the fact that it is implemented to a wider extent by multiple services, one should consider that these results may only be limited to Swedish users or users residing in countries that provide a similar, widespread solution. Still, the familiarity of a familiar authentication service, in this case BankID, was unarguably an evident factor of trust and should be considered to an extent regardless of industry.

Usability

Surprisingly, poor usability was mostly not shown to be that much of a ’Trustbuster’ as told by [14]. However, beyond the effect of brand familiarity, the cause of this could also likely be due to the necessity of visiting the website. According to the survey answers, once an insurance is taken on by a customer, they rarely revisit the insurance website unless they have to check something important or make use of their insurance.

Thus, it may be that any usability flaw is overlooked since the users are aware that do not have to experience the website again for a while are therefore more patient with the bad experience. The one usability aspect that did in fact prove to have an effect on users trust was the convenience of reaching customer service. This is likely because the customer service portal is one of the services that a customers may want to turn to if any problems or questions should arise.

Critical Discussion of the Method

The method used in this study is a well established methodology within Human-Computer Interaction [20]. As previously mentioned, since the perception of trustworthiness is subjective and differs between individuals, it was therefore necessary to gather elaborated and thoughtful answers from participants. Thus, the think aloud methodology was a rather obvious choice. Yet, there are still a few problems that accom- panies the think aloud methodology. First of all, while the interviews were conducted in a quiet, closed of environment – the setting of the interviews can be deemed as rather unnatural.

Since the participants were supposed to verbally say what they were doing and thinking throughout their usage of the platform, their comments and opinions may be seen as forced and artificial.

Secondly, since the participants in the users were fictionally testing out the trustworthiness of the different insurances, their perception of trust was arguably never really challenged completely. Additionally, as previously discussed, the fact that they did not have to use their own personal information and therefore were never personally exposed. Thus, a perhaps more suitable approach would have been to conduct contextual interviews in a natural environment, where the users’ would have been interviewed midst the process of actually buying an insurance as well as using their own personal information. That way, their thoughts and reasoning would likely have felt less forced and would have come out more naturally and honest.

Furthermore, the creation of user based personas and scenarios along with workshops could have vastly improved the design and the evaluation of the prototype.

Future Research

This study can be seen as a contribution to the continuous study of trustworthiness in e-commerce and has highlighted that there exists distinguishable differences in perception of trust depending on the industry. Additionally, this study may also provide an understanding of what people may expect from an website for insurance company as well as providing design recommendations for it. However, based on the results there are a few newly arisen questions that could be investigated further. One of the perhaps most interesting findings was the participants’ easy attitude towards the BankID authentication.

Thus, a possible future study could be to investigate this even further in order identify how much of an impact a familiar authentication method has on trustworthiness and to what extent. Furthermore, it could also be interesting to see whether this is industry and country specific or not.

CONCLUSION

Despite the potential complications that have been discussed in the previous section, the research question can be deemed as adequately answered to a certain extent. In this thesis, it has been identified that users, Digital Natives, within the age of 22 to 28 tends to expect an insurance company to convey a sense of professionalism and seriousness similarly to that of a bank or principal authority.

Furthermore, the trustworthiness of an insurance company depends strongly on the brand reputation of the company as well as the fact that it is an insurance company and thus tied to certain laws and regulation. There are however a selection of visual design choices that the company can implement in order to provide visual cues of trustworthiness towards the users.

The majority of the test participants and survey respondents agreed that lowly saturated colors and a symmetrical content alignment proved to be more trustworthy as it conveyed a sense of professionalism and seriousness.

Moreover, the study also shows that there is a tendency among younger generation, namely Swedish users around the age of 22 to 28, to have a tolerant attitude towards personal information authentications as long as the Swedish BankID service is present. The study also showed that these users showed a forgiveness of the lack of the design consistency and usability flaws towards well known and established insurance companies with a longer track record.

The findings in this study can provide useful design recommendations for companies that wants to further improve their overall customer experience on their websites as well as ap- pearing more professional and therefore more likely more trustworthy.

ACKNOWLEDGMENTS

First of all, I would like to extend my thanks to my supervisor at the Royal Institute of Technology, Adrian Benigno Latu- peirissa for his continuous guidance, support and feedback throughout the process.

Further, I would like to express my gratitude to Karin Knobel for her time, support and supervising at the principal. I also wish to show my gratitude to Lina Winqvist for giving me the

(13)

opportunity to write this master thesis as well as the whole team for making my stay there really enjoyable.

Lastly, I would like to thank all the test participants and survey respondents who took the time to participate in the study. I also wish to express my deepest gratitude to my brother for his invaluable support and feedback throughout the thesis and my whole education.

REFERENCES

[1] Marc Hassenzahl and Noam Tractinsky. User

experience - a research agenda. Behaviour Information Technology, 25(2):91–97, 2006.

[2] Josh Gothelf, Jeff; Seiden. Lean UX : Applying Lean Principles to Improve User Experience. O’Reilly Media, Sebastopol, 2013.

[3] Carlos Flavian, Raquel Gurrea, and Carlos Orús. Web design: a key factor for the website success. Journal of Systems and Information Technology, 11(2):168–184, 2009.

[4] Pairin Katerattanakul. Framework of effective web site design for business-to-consumer internet commerce.

INFOR: Information Systems and Operational Research, 40(1):57–70, 2002.

[5] User expectations and rankings of quality factors in different web site domains. International Journal of Electronic Commerce, 6(2):9–33, 2001.

[6] Masood H Siddiqui and Tripti Ghosh Sharma. Analyzing customer satisfaction with service quality in life insurance services. Journal of Targeting, Measurement and Analysis for Marketing, 18(3):221–238, Sep 2010.

[7] Nathalie Bonnardel, Annie Piolat, and Ludovic Le Bigot.

The impact of colour on website appeal and users’

cognitive processes. Displays, 32(2):69–80, 2011.

[8] Gitte Lindgaard, Gary Fernandes, Cathy Dudek, and J Brown. Attention web designers: You have 50 milliseconds to make a good first impression! Behaviour Information Technology, 25(2):115–126, 2006.

[9] Alexander Skulmowski, Yannik Augustin, Simon Pradel, Steve Nebel, Sascha Schneider, and Günter Daniel Rey.

The negative impact of saturation on website trustworthiness and appeal: A temporal model of aesthetic website perception. Computers in Human Behavior, 61:386–393, 2016.

[10] Andrew J. Elliot and Markus A. Maier. Color psychology: Effects of perceiving color on

psychological functioning in humans. Annual Review of Psychology, 65(1):95–120, 2014.

[11] Wouter A. Alberts and Thea van der Geest. Color matters: color as trustworthiness cue in web sites.

Technical communication, 58(2):149–160, 2011.

[12] Karl Long. Customer loyalty and experience design in e-business. Design Management Review, 15(2):60–67, 2004.

[13] Winnie wong poh ming. The effects of website quality on customer e-loyalty: The mediating effect of trustworthiness. International Journal of Academic Research in Business and Social Sciences, 4, 03 2014.

[14] J Riegelsberger and Ma Sasse. Trustbuilders and trustbusters: The role of trust cues in interfaces to e-commerce applications. In In: Schmid. Kluwer Academic Publishers, 2001.

[15] Florian N. Egger. “trust me, i’m an online vendor”:

Towards a model of trust for e-commerce system design.

In CHI ’00 Extended Abstracts on Human Factors in Computing Systems, CHI EA ’00, page 101–102, New York, NY, USA, 2000. Association for Computing Machinery.

[16] Florian N Egger et al. Affective design of e-commerce user interfaces: How to maximise perceived

trustworthiness. In Proc. Intl. Conf. Affective Human Factors Design, pages 317–324. Citeseer, 2001.

[17] Marc Prensky. Digital natives, digital immigrants part 1.

On the Horizon, 9(5):1–6, 2001.

[18] Marc Prensky. Digital natives, digital immigrants part 2:

Do they really think differently? On the Horizon, 9(6):1–6, 2001.

[19] Christian Pieter Hoffmann, Christoph Lutz, and Miriam Meckel. Digital natives or digital immigrants? the impact of user characteristics on online trust. Journal of Management Information Systems, 31(3):138–171, 2014.

[20] Jonathan Lazar. Research methods in human computer interaction. 2nd edition, 2017.

[21] Richard E. Nisbett and Timothy D. Wilson. The halo effect: Evidence for unconscious alteration of judgments. Journal of Personality and Social Psychology, 35(4):250–256, 1977.

(14)

www.kth.se

TRITA-EECS-EX-2020:63