Electronic Service Quality in Dutch Shopping Websites

(1)

Electronic Service Quality in Dutch Shopping Websites

Developing a scale based on objective measures

by

Andries D. Cupido

Faculty of Economics & Business, Department of Business & ICT

(2)

2

ABSTRACT

Electronic service quality is concerned with the extent to which a website facilitates efficient and effective shopping, purchasing and delivery. This study reveals objective measures from extant electronic service quality literature to develop a new scale. The seven dimensions of the E-S-QUAL and E-RecS-QUAL models are used as a foundation, and the personalization dimension is added. Extant literature research leads to the unearthing of 36 objective measures. To test the scale, 120 Dutch shopping websites are assessed, divided into selling physical and non-physical products. The results show that both website types score higher in the E-S-QUAL dimensions than in the other dimensions. The results of the different product groups are very similar. Theoretical and practical implications are given.

(3)

3

ACKNOWLEDGEMENTS

The completion of this Master‟s thesis concludes my Master‟s programme Business Administration with specialization in Business & ICT. I am happy to leave the University of Groningen writing this document about one of my big passions: e-commerce. My time in Groningen has been a long and interesting journey, which has allowed me to both professionally and personally develop myself. During my time in Groningen, I have been lucky enough to have met numerous fantastic friends that have helped me enjoy every single day in this city. With many great people in my thoughts, I would like to thank a number of people in particular.

First of all, I would like to express my gratitude to Dr. Eric Lim for providing me with valuable feedback on this thesis.

I am very thankful for the cooperation with my thesis supervisor Dr. Chee-Wee Tan, who has been very helpful during the entire process. Chee-Wee, meetings with you have always been pleasant. In my perception, your comments and suggestions have constantly been “on the spot” and made me leave your office with renewed thesis energy. I hope we will meet again in another setting!

My good friend and business partner Bart, with whom I have spent numerous nights over the last years making our internet company Centillion grow. One valuable lesson that I have learned from our cooperation is that work and fun may come together. Our mission to centillionize the world continues in Amsterdam.

Also, a person who deserves a line of credit is my girlfriend Lydian who has been very supportive throughout the whole process, and who has provided me with the occasional kick to the library when I needed it ;-). Thank you!

(4)

4

1. Introduction

1.1 Background introduction

More and more retail products are being purchased online. In 2009, total retailers‟ online revenues in The Netherlands increased with seventeen percent to 6.375 billion Euros (Blauw Research, 2010). Comparably, other European countries‟ online shopping revenues have constantly shown annual growth rates since 2001 (Hoofdbedrijfschap Detailhandel, 2010).

On the European level, retail sales grew by 18 percent in 2010 and are expected to grow 13 percent in 2011. Also, the number of online buyers is projected to grow from 157 million to 205 million in the coming five years (Carini, 2011). Overseas, the United States online retail market shows similar figures (Mulpuru, 2011) with an expected ten percent compound annual growth rate from 2010 to 2015.

In both continents, these double-digit annual growth figures are driven by ubiquitous web connectivity among consumers, increasing consumer familiarity with and preference for online shopping and the development that retail is increasingly becoming a multichannel environment (Forrester Research 2011a, 2011b).

1.2 Commercial website as vital channel for Internet marketing

(8)

8

1.3 Satisfying the online customer: electronic service quality

To have a commercial website actually add value to a firm, a website should deliver what its users want. According to Constantinides (2004), firms can influence customers‟ perceptions of a website and their subsequent behavior by creating superior web experience. Similarly, Zeithaml, Parasuraman and Malhotra (2002) mention that website quality plays a major role in shaping a consumer‟s web experience. To create a superior web experience, companies must understand the separate components contributing to a quality web experience. Also, there must be understanding of the role of these components as input for online customers‟ decision making processes (Constantinides, 2004).

In extant literature, a variety of studies points to the advantages of satisfying online customers, with varying angles of incidence. Grönroos (2000) states that a commercial website has the function to link the customer to in-depth information about goods and services, and to promote the customer to make an actual purchase at that website.

Reichheld and Schefter (2000) state that building a loyal customer base is essential, especially in an e-commerce context. Competitors are easy to find and easy to access online and switching to a competitor is relatively easy. Also, acquiring new customers online is costly. Therefore, they emphasize the importance of satisfying customers and therewith creating e-loyalty.

Concluding, satisfying existing customers and prospects is deemed an essential part of any online retailer‟s strategy to keep ahead of competition; regardless whether for instance maximum profit, loyalty or retention are the main goals.

Parasuraman, Zeithaml and Malhotra (2005) argue that a customer‟s assessment of a commercial website includes all phases of a customer‟s interactions with the website. Hence, the quality of a website is important to effectively reach a customer, and let a customer perform the desired action. In other words, in order to make customers perform the desired actions they must be presented with excellent service quality at the website. In this context, service may be defined as “the results that customers want, which can be obtained by the interactive processes between customers and service providers” (Harvey, 1998).

(9)

9

1.4 IT budgets exceeded

In difficult economic times like the current, budgeting seems more important than ever. IT trends are nowadays often involved with minimizing costs (Greenfield, 2011). Emphasizing the relevance of IT budgets to be controlled, a recent report (Input, 2010) states that United States IT budgets are estimated to rise with a compound annual growth rate of 3.1% in the years 2010-2015. This implies that IT budgets will play a bigger role in corporate budgets in the near future.

Related, most models currently dealing with electronic service quality are designed in order to evaluate commercial websites that have already been developed completely (Parasuraman et al. 2005). Although the results of these models can be very useful when evaluating and improving the current website, the phase of initial website development –with its associated costs - has passed. Possibly, improvements have to be made that could have been dealt with during the initial development phase. Therefore, a model with relevant items that can be taken into consideration during the development phase of the website would constitute value.

1.5 Using objective measures during website evaluation

Most previous researchers based their judgments on results of questionnaires that were totally or largely based on measures that have a perceptual nature (i.e. the extent to which a website is useful, aesthetically pleasing, or easily navigable). These include the models WebQual (Barnes and Vidgen, 2000), E-S-QUAL (Parasuraman et al. 2005), eTailQ (Wolfinbarger and Gilly, 2003) and also the models created by Cai and Jun (2003), Janda et al. (2002), Kim and Stoel (2003), Palmer (2002); Ribbink, van Riel, Liljander and Streukens (2004); Rosen and Purinton (2002). These questionnaire-based models with perceptual measures have two major disadvantages.

(10)

10 Second, the measures themselves may not fit all companies due to the varying nature of businesses. For instance, aesthetic beauty of a website may be more relevant to a company operating in the apparel market, than for a company operating in the insurance industry. Consequently, analyses of results may be interpreted inappropriately, leading to doubtful consequences. In this matter, a model with universally appropriate measures would be valuable.

1.6 Research goals

Considering the aforementioned, a number of research goals can be formulated for the current study:  Add to the electronic service quality literature;

 Develop a scale with objectively measurable items that can be used by organizations when evaluating, but also when developing a commercial website;

 Gather insights about and look for improvements in the model by applying it to a selection of online shopping websites, and

 Gather insights about the shopping websites themselves and the market they operate in.

1.7 Application area: Dutch online shopping websites.

Most previous electronic service quality research projects address specific industries, mostly in retailing (i.e. Khalifa and Liu 2003, Loiacono, Watson and Goodhue 2002, Zeithaml et al. 2002). This prevents industry-specific factors from influencing the results. Also, selecting a specific geographical area of research may be advantageous as there are a number of constant context variables, and therefore cannot distort the outcome of the research. These variables may include similar payment options, cultural differences, and legal issues such as delivery conditions and obligations to accept returns.

Also, researching solely shopping websites is a deliberate choice. Wolfinbarger & Gilly (2003) mention that many earlier similar research projects deal with all kinds of websites (including news and entertainment websites, among others) and that consumers tend to have different motivations for interacting with different types of websites. Earlier, Wolfinbarger and Gilly (2001) found that most online shoppers are more goal-oriented than experiential. According to Novak, Hoffman and Yung (2000) this is relevant because task-oriented and experiential behavior may differ. According to Zeithaml et al. (2002), the entertainment criteria identified in studies using general websites are not relevant in the online purchase context.

(11)

11 Netherlands are among the top nations regarding Internet connectivity (Greenfield, 2011) and the number of households with broadband access (Cisco, 2010). Currently, very little research has been done on online shopping websites in The Netherlands.

1.8 Outline of the current research

(12)

12

2. Theoretical framework

In order to gain more insights in how to effectively measure electronic service quality in this research, the next sections will be devoted to providing background on service quality and electronic service quality.

2.1 Service quality

Evidence suggests that service quality promotes customer loyalty and retention (Imrie, Durden and Cadogan 2000) which is relevant to any retailer, operating online or offline.

(13)

13

2.2 Electronic service quality

The aforementioned SERVQUAL and SERVPERF models focus on dimensions that are suited for an offline context. However, Collier and Bienstock (2006) mention that the perception of online service quality can be different from the offline equivalent, since services delivered online have certain characteristics that offline services do not possess. For example, technical website issues such as high load times, database crashes and connectivity issues are unique online. Therefore, online service quality models should take this into account and the older models should be adjusted.

2.3 Defining electronic service quality

In order to decide which dimensions should be used for measuring in the current research, electronic service quality should be defined. Zeithaml et al. (2002) state that various researchers have initiated conceptualizing and measuring electronic service quality, but at the same time domains have not yet been clearly defined. For this reason, and due to the infancy of the research field, there is little agreement about the exact meaning of electronic service quality. Furthermore, Zeithaml et al. (2002) add that some conceptualizations are limited to interactions with the website itself; others include post-transactional website aspects such as fulfillment and returns. They refer to the only formal definition so far that does include service quality during all mentioned online shopping phases. Considering the breadth of the earlier mentioned service problems that began to occur during the 2000s (i.e. technical website quality, contact issues, fulfillment failures), it is deemed appropriate to take all phases into account.

Therefore, similar to Parasuraman et al. (2000), electronic service quality is defined as the extent to which a website facilitates efficient and effective shopping, purchasing and delivery.

2.4 Electronic service quality models

In this section, current electronic service quality literature will be discussed. Using the earlier mentioned definition, the models derived from this definition E-S-QUAL and E-RecS-QUAL will be discussed. All seven dimensions used in these models will be elaborated on, while discussing and comparing other relevant models. Following, a new dimension (Personalization) will be presented.

2.4.1 Extant electronic service quality literature

(14)

14

2.4.2 e-SERVQUAL

Based on their definition of electronic service quality, Zeithaml, Parasuraman and Malhotra (2000, 2002) performed a three-stage process using exploratory focus groups, and two phases of empirical data collection and analysis in order to reveal appropriate measures. Their study revealed seven dimensions of electronic service quality: efficiency, reliability, fulfillment, privacy, responsiveness, compensation and contact. Efficiency refers to the seamless accessing, navigating and completing of a transaction at a website. Reliability is concerned with the proper technical functioning and availability of the website. Fulfillment incorporates physical availability of products, accurate and in-time product delivery and other accuracy of service promises. Privacy deals with the protection of online shoppers‟ personal data, including credit card information and shopping behavior data. Responsiveness is concerned with supplying customers with information in case of problems with products. Compensation involves financial compensation in case of product returns. Finally, the contact dimension refers to the possibility for customers to receive live support through telephone or a live online customer service agent. These seven dimensions were divided into two groups: efficiency, reliability, fulfillment and privacy formed the core e-SERVQUAL scale. The responsiveness, compensation and contact dimensions formed the recovery e-SERVQUAL scale. The scales differ to the extent that the core scale is used when an online customer does not experience problems when visiting the website; the recovery scale is used only when online customers encounter complications – such as problems with their orders or when they have questions.

2.4.3 E-S-QUAL and E-RecS-QUAL

The e-SERVQUAL model (Zeithaml et al. 2000, 2002) was further developed into a new model called E-S-QUAL (Parasuraman et al. 2005). To more effectively measure the service quality provided through a website, the authors elaborated on the thought that various earlier researchers often focused too much on interaction with the website instead of looking at the big picture. Not only should researchers be looking at website interactivity or process quality, but the outcome quality and recovery quality should also be taken into account. As Wolfinbarger and Gilly (2003) mention, online shopping activity involves all activities from information search, product comparison and evaluation, making the payment, product delivery and after-sales service such as returns, refunds and customer service. This is acknowledged by Collier and Bienstock (2006).

(15)

15 Analysis of the results of the questionnaires led to the insight that a large proportion of questions related to certain items were not answered. These items all related to service recovery, and appeared not to be answered because the respondents had not experienced the situations covered by these items (Parasuraman et al. 2005). Further analysis led to a division in dimensions similar to e-SERVQUAL, with four dimensions forming an e-core service quality scale: E-S-QUAL. The dimensions were labeled efficiency, fulfillment, system availability and privacy. It should be noted that system availability was called reliability in e-SERVQUAL. The recovery scale was called E-RecS-QUAL, containing the dimensions responsiveness, compensation and contact.

An overview of the E-S-QUAL and E-RecS-QUAL dimensions, along with their definitions is found in Table 1. Also, a rationale is added with every dimension to clarify its meaning. The complete model along with all 33 measurement items can be found in Appendix B.

Table 1. E-S-QUAL and E-RecS-QUAL dimensions (Parasuraman et al. 2005)

Dimension Definition (Parasuraman et al. (2005) Rationale Efficiency The ease and speed of accessing and using

the site.

Are customers able to find what they are looking for?

Fulfillment The extent to which the site‟s promises about order delivery and item availability are fulfilled.

Are customers able to be informed about the logistic handling of their orders?

System availability The correct technical functioning of the site. Is the website functioning correctly?

Privacy The degree to which the site is safe and protects customer information.

Are customers‟ data and privacy guaranteed in a safe shopping

environment?

Responsiveness Effective handling of problems and returns through the site.

Do customers get the assurance that eventual problems will be solved in a correct manner?

Compensation The degree to which the site compensates customers for problems.

Are customers insured against financial risks in case problems occur?

Contact The availability of assistance through telephone or online representatives.

Are customers able to communicate with human representatives to get their questions answered?

2.4.4 Studying the dimensions

Zooming in at the E-S-QUAL and E-RecS-QUAL dimensions demonstrates a big overlap with various other relevant studies.

(16)

16 rendered correctly will prevent the customer from leaving the website. The importance of easily accessible information is supported by the finding that the reduction in search costs is one of the reasons for shopping online (Alba et al. 1997, Ariely 2000, Bakos 1997, Lynch and Ariely 2000). Various other studies have paid attention to the quality of website content, albeit in many different forms (f.i. Agarwal and Venkatesh 2002, Cai and Jun 2003, Collier and Bienstock 2003, 2006; Douglas, Muir and Meehan, 2003, Kim and Lim 2001). Used dimensions include quality, accessibility or accuracy of information. Also ease of use and website load speed were included in numerous earlier studies (f.i. Agarwal and Venkatesh 2002; Devaraj, Fan and Kohli 2002; Loiacono et al. 2002). However, other studies also include perceptual measurement items, which are not desirable to use in the current study. These perceptual items are often related to aesthetics and web design (f.i. Barnes and Vidgen 2001; Evanschitzky, Iyer, Hesse and Ahlert 2004; Fassnacht and Koese 2006).

The fulfillment dimension is also covered in a number of studies. Wolfinbarger and Gilly (2003) mention that fulfillment aspects are among the most important to value the quality of a website. Yang and Fang (2004) state that keeping fulfillment promises are strong predictors of customer satisfaction and dissatisfaction. However, the emphasis often lies on physical fulfillment issues, instead of the extent to which the website facilitates fulfillment aspects (f.i. Wolfinbarger and Gilly 2003; Kim, Kim and Lennon 2006). Also in E-S-QUAL, the fulfillment dimension is defined as the extent to which the site‟s promises about order delivery and item availability are fulfilled. In the current study, no actual products will be ordered and therefore only fulfillment assessments will be made on website characteristics.

The system availability dimension is defined as the correct technical functioning of the site (Parasuraman et al. 2005). A correctly functioning website – due to missing links, non-functioning buttons, or error messages- leads to user frustration and exiting. Consequently, a valuable opportunity to build customer loyalty is lost (Wachter, 2002). Santos (2003) acknowledges that assuring links function correctly and availability of all pages contribute to the total service quality delivered. Although measurement items vary among studies (i.e. McKinney 2002; Kim, Xu and Koh 2004) researchers agree that structurally failing technical aspects of a website possibly hinder customers‟ shopping intentions.

(17)

17 decade ago. Therefore, objective privacy measures that are not bound to temporal (or geographic) restrictions could be useful.

In the E-RecS-QUAL scale, responsiveness is defined as effective handling of problems and returns through the site. It is concerned with the provision of adequate information through the website in case of problems with orders. Responsiveness issues have also been addressed extensively in earlier studies (f.i. Jiang, Klein and Carr 2002; Ribbink 2004; Semeijn et al. 2005).

The compensation dimension shows similarities with the fulfillment dimension. Defined as the degree to which the site compensates customers for problems, on the measurement item level E-RecS-QUAL deals with financial consequences of actual problematic transactions. As actual transactions will not occur in the current study, the items will have to be replaced by items that solely measure on-site aspects.

Finally, E-RecS-QUAL deals with the contact dimension, defined as the availability of assistance through telephone or online representatives. It deals with the possibility for customers to seek contact with a human being to get an answer to a question. In earlier research (Wolfinbarger and Gilly 2003), customers appeared to react negatively to a website not delivering unprompted or delayed customer inquiry services. The measures by Parasuraman et al. (2005) are objectively measurable, thus can be used in the current research.

2.4.5 E-S-QUAL and E-RecS-QUAL to be used as a foundation

The findings of earlier research projects are very useful to determine which dimensions to include in the current research. Parasuraman et al. (2005) mention that in this young research field, only few models have captured a complete picture of electronic service quality with empirical evidence. As the above review of previous research has shown, a number of recurring shortcomings in other models summarize why the E-S-QUAL and E-RecS-QUAL dimensions are the most appropriate to use as a foundation in the current research.

 First, samples used by many studies do not merely consist of visitors to shopping websites. Visitors to other types of websites may value websites differently because they have other intentions.

 Second, several studies address appropriate dimensions, but offer an incomplete picture of service quality as opposed to the formulated definition. Wolfinbarger and Gilly (2003) developed eTailQ, including website design, reliability/fulfillment, privacy/security and customer service as dimensions in their scale. Although eTailQ is very useful, E-S-QUAL and E-RecS-QUAL have split their dimensions up a bit further.

(18)

18 (2001) developed SITEQUAL, measuring website quality on four dimensions: ease of use, aesthetic design, processing speed and security. Comparably, Barnes and Vidgen (2002) developed WebQual, which measures an organization‟s e-commerce offering on five factors: usability, design, information, trust and empathy. Both models contain many useful aspects for the current research, but do not address important parts of electronic service quality as defined, such as fulfillment issues.

Also, Parasuraman et al. (2005) state that the main purpose of E-S-QUAL and E-RecS-QUAL is to generate information for website designers, rather than to measure service quality as experienced by customers. This is congruent with the current research goal to develop a model that can also be used during the development of a shopping website.

The aforementioned leads to the decision to use the E-S-QUAL and E-RecS-QUAL dimensions as a framework for the current research. Additionally, an extra dimension will be suggested in the next paragraph.

2.5 A new dimension: Personalization

In this section, the addition of a new dimension will be suggested to the seven E-S-QUAL and E-RecS-QUAL dimensions to build the current research on.

According to Zeithaml et al. (2002), personal service (in SERVQUAL captured by the empathy dimension) is not critical in the transactional aspects of online service. They state that although in an offline context various forms of personal attention (such as reassurance, courtesy and understanding) are valued by customers; in electronic service quality research they do not seem to be key issues. Although Parasuraman et al. (2005) have removed the personalization dimension from their E-S-QUAL model, relevant studies have extensively included personalization and customization in their metrics (Wolfinbarger and Gilly 2003, Kim et al 2006, Ribbink 2004, Semeijn et al. 2005; Srinivasan, Anderson and Ponnavolu 2002).

Mulvenna, Anand and Buchner (2000) have mentioned a relationship between personalization with online service quality. As this study addresses electronic service quality, the focus is more on personal service than on personalization.

(19)

19 characteristics, aiming at acquiring users‟ personal information in order to personalize their products and accompanying advertisements.

Another major development in the scenery online has been the rise of social media websites. In November 2010, Facebook accounted for 25 percent for all pageviews generated (Dougherty, 2010). Moreover, in the Dutch market social media website Hyves has accomplished a reach of over seventy percent of the Dutch population with over 11.2 million registered accounts (Hyves, 2011). As a whole, social media websites in The Netherlands are growing fast in terms of reach and time spent at these websites (Multiscope, 2010). Also on a global level, the use of social media websites has increased enormously (Nielsen, 2010). Important aspects of these sites are the ability to share information with, and recommend information to, connected people. Another characteristic of these sites is the offered functionality to customize your own profile page.

Related, with technology that has evolved during the past decennium, websites such as Facebook, but also stores such as Amazon are able to “recognize” returning users by storing little pieces of information on a user‟s computer. When the user decides to visit the website again, the website is able to reproduce previously saved user preferences. Wolfinbarger and Gilly (2003) included several such measures in their EtailQ model. For shopping websites, such technology seems very relevant when applied to abandoned shopping carts (thus unfinished transactions), purchasing recommendations based on earlier transactions, and other website customizations that facilitate online shopping, and therewith electronic service quality.

The aforementioned leads to the insight that over the last years, personalizing and customizing have become interesting topics in website use. The likely relevance to shopping websites leads to the decision to include a personalization dimension in the current model.

2.6 Conclusion

(20)

20

3. Methodology

In this chapter, the selected dimensions will be operationalized. After, the tested sample will be addressed. Then, the measurement procedure will be explained and demonstrated.

3.1 Operationalization

After determining appropriate dimensions to be used in this research, measurement items have to be added to the dimensions. This process is also called operationalization. E-S-QUAL and E-RecS-QUAL largely present measures with a perceptive nature, and therewith these items will have to be replaced to adhere to the current research goals. Of the 32 original items, six items can be retained. To complete the new model with objective measures, a considerable number of relevant earlier studies is evaluated. From these studies and their accompanying models, relevant measures are derived. Apart from being relevant, the measures should be objectively measurable to fulfill the research objective. Therefore, some original items are reformulated so they can be valued with Yes or No. The measures should not create overlap with each other, but altogether cover as many aspects as possible in their respective dimensions. This way, differences between websites or groups of websites can be distinguished on a maximum number of factors.

(21)

21

3.2 The final model

As a result of the foregoing, the final model is presented in Table 2. Following, all measures will be described.

Table 2. The final electronic service quality model

Dimension Measurement items Description Sources

Efficiency (A)

A1 - Graphic quality: Text and images displayed properly.

A2 - Sitemap

A3 - Organized categories

A4 - Site load speed under twelve seconds.

A5 - HTML version available

- Text on the website is perfectly legible and images are rendered correctly.

- There is a sitemap of the website available.

- Products and information can be found in logical places.

- Index and subpages load within twelve seconds. - There is a basic HTML version of the website available.

- Fassnacht & Koese (2006)

- Rosen & Purinton (2004)

- Wolfinbarger & Gilly (2003)

Hoxmeier & DiCesare (2000)

Parasuraman et al. (2005)

Fulfillment (B) B1 - Order status/ tracking

B2 - Product availability B3 - Delivery time B4 - Product pictures available B5 - Product specifications B6 - Search function

B7 - Direct online payment

B8 - Ordering

B9 - Comparison matrix

B10 - Recommendation agents

- Customer can track the status of the order. - Product availability is mentioned on the site.

- Product delivery time is mentioned on the site. - Catalog products have pictures available online. - Product specifications are available.

- There is a search function available that functions without errors.

- Products can be purchased directly via at least one online payment method.

- It is possible to order products via the website.

- Different products can be compared through the website.

- A mechanism on the website makes

recommendations to the customer.

- Kim et al. (2006), Tan (2009)

- Wolfinbarger & Gilly (2003), Janda et al. (2002)

- Wolfinbarger & Gilly (2003), Srinivasan et al. (2003)

- Wolfinbarger & Gilly (2003), Tan (2009), Kim & Stoel (2003) - Cenfetelli, Benbasat and Al-Natour (2008), Wolfinbarger & Gilly (2003)

- Häubl & Trifts (2000)

- Xiao & Benbesat (2007)

System availability (C)

C1- Opening times

C2 - Site loads correctly

- Site is always open for business.

- Site loads without structural errors.

- E-S-QUAL

Privacy (D) D1 - Encryption system

D2 - Mailing list prevention

- There is a certified encryption system available.

- Customer is assured not

- Douglas (2003)

(22)

22

D3 - Mailing list withdrawal

D4 - Personal info protection

D5- Privacy policy

D6- Third party verification

to be added to a mailing list unvoluntarily.

- Customers have the option to opt out of the mailing list after subscribing.

- Customer is assured that personal data are not shared with third parties. - There is a privacy policy available on the site. - Security is verified by an independent third party.

- Janda et al. (2002)

- Kim et al. (2006)

- Pavlou & Gefen (2004)

Responsiveness (E) E1- FAQ available

E2 - Feedback channel

E3 - Instant help library

E4 - Return policy

E5 - Website survey

- Frequently Asked Questions section is available.

- Customer is asked to give feedback through a form or e-mail address.

- Online terminology list is available.

- There is a return policy available on the website - The user is asked to fill in a survey about the website.

- Palmer (2002), Singh (2002) - Palmer (2002), Tan (2009) - Kim et al. (2006) - E-RecS-QUAL, Janda et al (2002) - Palmer (2002)

Compensation (F) F1- Resolution center

F2 - Refund policy

- There is a section on the website devoted to solving problems with ordered items.

- There is a refund policy available on the website.

- E-RecS-QUAL

Contact (G) G1- Telephone number

G2 - Live online customer service

G3 - E-mail helpdesk

- A telephone number is available for customers. - Possibility to interact online with customer representative

- E-mail address or form designated for customer service. - E-RecS-QUAL - E-RecS-QUAL - E-RecS-QUAL Personalization (H) H1- Purchase recommendations H2 - Customized items H3 - Customer recognition

- Shopping suggestions are made by the website. - Products can be

customized for customers.

- Customer is recognized at next visit. - Srinivasan et al. (2002) - Srinivasan et al. (2002), Kim et al. (2006), Semeijn et al. (2005)

(23)

23

3.3 Discussing the measures

In this paragraph, the measures for this research will be discussed. The literary sources are mentioned, the motivation to include these measures and a description of how the actual measurements fit in the current research. Items that are objectively measurable are eligible for inclusion, as well as items that can be transformed into objective measures. Furthermore, there should not be overlap between measures. Multiple relevant studies using the same metric will both be included.

3.3.1 Efficiency measures

In Table 3, an overview of the efficiency measures is presented. As defined earlier, efficiency incorporates the ease and speed of accessing and using the site. A plenitude of studies (mainly measuring website quality) focuses on these aspects.

Table 3. Efficiency measures

Measures Description and motivation

Graphic quality In their final model, Fassnacht and Koese (2006) include graphic quality. In this research, graphic quality implies that text on the website is perfectly legible and that pictures and images are rendered correctly.

Sitemap Similar to Rosen and Purinton (2004), the availability of a sitemap is used as a measure. A sitemap is a page on the website that tells the user about the content of the website with static links, and is a document that enables the user to visit every section of the site.

Organized categories

Whether or not the offered products or services can be found in organized categories, was also measured in the work of Wolfinbarger and Gilly (2003). This implies that product offers and accompanying information can be found either within one click from the home page, or through a

hierarchical category structure that leads to the relevant information.

Website load speed

To create an objective measurement item for website load speed, a different approach should be taken than the majority of researchers, who mostly deal with the perception of users instead of creating a hard measure (f.i. Janda et al. 2002). Hoxmeier and DiCesare (2000) found that an application should load completely within twelve seconds, before a user‟s attention is likely to be drawn away and the application would be closed. This is also relevant for websites and therefore will be used in this research to measure website load speed. To measure this, the websiteoptimization.com (Website Optimization, 2010) tool was used. Comparable connection speeds with the average download speed in The Netherlands (Cisco, 2010) were used with the tool to calculate the website load speed.

Basic HTML version

In order to make a website accessible to all users, content should be presented in formats understandable for all web browsers. Flash

(24)

24

3.3.2 Fulfillment measures

In Table 4, the fulfillment measures are described. Fulfillment was defined as the extent to which the site‟s promises about order delivery and item availability are fulfilled. Since it was not feasible to actually order products, only fulfillment items that were measurable by observing the website are used.

Table 4. Fulfillment measures

Measures Description and motivation Order

status/tracking

Part of the fulfillment process is the ability to be informed about the current status of the order. Kim et al. (2006) included the order status/tracking measure. Tan (2009) measures whether or not a

personalized tracking system is provided to track the processing status in current transactions. Wolfinbarger and Gilly (2003) perform a similar measurement. A tracking number that is sent after completing the transaction, but also the availability of a service telephone number, live online chat service or helpdesk e-mail address are considered as feasible alternatives to obtain this information.

Product availability Mentioning product availability on the website is part of the fulfillment

process, since it could be counter-productive for customers to select a product and later find out that the product is out of stock. Product availability information was measured by Janda et al. (2002), and was also addressed by Wolfinbarger and Gilly (2003).

Delivery time information

Depending on a customer‟s needs and wishes, an announced long delivery time can be a reason to decide not to purchase the product and is part of adequate product information. In this study, actual delivery times will not be measured such as by for instance Wolfinbarger and Gilly (2003), but the availability of delivery time information.

Product pictures The availability of product pictures was also measured by Janda et al. (2002) and Wolfinbarger and Gilly (2003) and is considered part of adequate product information. In this research, at least one product picture must be available for each product to receive a positive score for this item.

Product specifications

Similar to product pictures, Janda et al. (2002) used product specifications as part of their „Information‟ dimension. In this research, at least a title and a description of the product must be supplied by the store to qualify for this item.

Search function Instead of browsing through categories, users may feel the need to use a search function to find the product they are looking for. Wolfinbarger and Gilly (2003), as well as Srinivasan et al. (2002) used the availability of a search function in their study. To determine whether this functionality was available at the site, the most generic search term was used to describe the product. In case of one or more relevant results came up, a positive score was awarded.

Direct online payment

(25)

25 the possibility to complete transactions online as a measure. In this

research, at least one direct online payment method should be present to score positively on this item.

Ordering It seems evident that the lack of an option for customers to actually order products online could lead to a sub-optimal amount of sales. Cenfetelli, Benbasat and Al-Natour (2008) use the facilitation of online order processing as a measurement item; Wolfinbarger and Gilly (2003) performed another research measuring the (correct) processing of orders through the website.

Comparison matrix Häubl and Trifts (2000) have found that the use of comparison matrices

cause shoppers to make more efficient and faster decisions. Therefore, in this study the option to compare a product with one or more other products leads to a positive score for this item. Devaraj et al. (2002) included a similar item.

Recommendation agents

Xiao and Benbasat (2007) state that recommendation agents are software systems that elicit individual users‟ preferences or interests for certain products, and make recommendations accordingly. By presenting customers with alternatives they have interest in, they are more likely to make a purchase. In this research, this item is marked as “Yes” if recommendations are made based on at least one of the three of the following:

 After a product has been selected (f.i. “Customers who bought this also bought…”);

 After a search query has been performed (f.i. “You searched for product X, you may also be interested in product Y”);

 Browsing product categories (f.i. You are currently browsing category X, you may also browse category Y”).

3.3.3 System availability measures

In Table 5, an overview of the system availability measures is presented. System availability was defined earlier as the correct technical functioning of the site.

Table 5. System availability measures

Opening times Parasuraman et al. (2005) use availability of a website in the original E-S-QUAL model in the system availability dimension. In this research, a website is given a positive score if it is available for business during all three iterations, and there are no indications that the website is

periodically out of business; regardless of the reason why access to the website would not be available.

(26)

26

3.3.4 Privacy measures

In Table 6, an overview of the privacy measures is presented. In this regard, privacy was defined as the degree to which the site is safe and protects customer information.

Table 6. Privacy measures

Encryption system Douglas, Muir and Meehan (2003) note that when products are sold online, an encryption system should be present as a security measure. In the present study, the transaction sections of all observed websites have been evaluated and scanned for an encryption system, to be recognized by the symbol.

Mailing list prevention Janda et al. (2002) have used the assurance not to be placed on a

mass-mailing list as a measurement item in their study. In the present study, this information was gathered from privacy policies and other places to submit e-mail addresses (such as newsletter subscriptions).

Mailing list withdrawal

Similar to mailing list prevention, the option for a customer to get his e-mail address removed from a mailing list was checked.

Personal info protection

A popular and often mentioned aspect of e-commerce security is the perceived trust in the company not to share personal data with other parties. Janda et al. (2002), and Wolfinbarger and Gilly (2003) included this item in their investigations. Since it was not feasible to actually check whether data were checked or not in the current

research, the research focused on the availability of explicit statements with guarantees of safe personal data.

Privacy policy Despite being a legal obligation in certain regions, publishing a privacy policy online may facilitate users to easily inform themselves about privacy issues. Kim et al. (2006) included the availability of a privacy policy as an item.

Third party verification

Pavlou and Gefen (2004) found evidence for their proposition that perceived effectiveness of IT-enabled institutional mechanisms stimulate buyer trust in an electronic shopping environment, in their case specifically auction sites. This implies that buyers may feel safer in a shopping environment where their security is guaranteed by an identified party. In this research, a website is given a positive score when at least one such guarantee is given.

3.3.5 Responsiveness measures

(27)

27

Table 7. Responsiveness measures

FAQ available Frequently asked questions (FAQ) can be a source of information for customers, since answers to common issues are supplied. At the same time, this lowers the need for helpdesk labor. Palmer (2002) included the availability of FAQ in his responsiveness measures as part of his website performance study. Singh (2002) mentions FAQ as a source of often-used tools to promote sales.

Feedback channel Besides the availability of FAQ, Palmer (2002) also uses the presence of a feedback channel as a measure in the aforementioned study. Tan (2009) also measured if feedback was collected from e-government websites. In the current study, the presence of a feedback channel is recognized in case there is a form or e-mail address available with the explicit intention to receive feedback from users.

Instant help library Similar to the presence of FAQ, an instant help library can provide customers with information regarding the products by explaining terminology in clear language. Kim et al. (2006) included the availability of an instant help library.

Return policy In E-RecS-QUAL, Parasuraman et al. (2005) have included a measure „This site handles product returns well‟. Janda et al. (2002) measured the availability of clearly stated return policies. As it was not feasible to order actual products (thus testing the actual functioning of the return policy) in this research, a positive score was awarded in case a return policy was mentioned on the site.

Website survey Derived from Wolfinbarger and Gilly (2003), a specific method of collecting feedback is done through the use of website surveys. A pop-up window appears when entering or leaving the site, or after using the site for a certain amount of time. Another form of these surveys is through a special survey link or button. When one of these forms was observed, the website was considered offering a website survey.

3.3.6 Compensation measures

In Table 8, an overview of the compensation measures is presented. In this context, compensation was defined as the degree to which the site is safe and protects customer information.

Table 8. Compensation measures

Resolution center Besides giving customers options to contact the company to ask for assistance, customers may value a section on the website dedicated to problem-solving (derived from E-RecS-QUAL).

(28)

28

3.3.7 Contact measures

In Table 9, the contact measures are described and motivated. As part of electronic service quality, the earlier used definition is the availability of assistance through telephone or online representatives.

Table 9. Contact measures

Telephone number Similar to E-RecS-QUAL, the website was given a positive score when a telephone number was supplied to contact the company. This includes both regular telephone numbers and Skype contacts.

Live online customer service

Similar to E-RecS-QUAL, the availability of live online customer service was measured. In practice, this implied that a website was considered to offer this specific type of service in case an online chat service was available.

E-mail helpdesk Contrary to the earlier mentioned feedback forms or feedback e-mail addresses, contacting a helpdesk through e-e-mail has another goal: receiving assistance with completing a transaction at the website, or asking other information regarding the store. In this research, a positive score was awarded in case an e-mail address or form was specifically offered to collect helpdesk inquiries. A similar measure was included in E-RecS-QUAL.

3.3.8 Personalization measures

In Table 10, an overview of the measures for the newly added personalization dimension is presented.

Table 10. Personalization measures

Measures Description and motivation Purchase

recommendations

In the current study, a website is considered to offer purchase recommendations if one or more related products are displayed when selecting a product, or when particular products are highlighted after a search.

Srinivasan et al. (2002) also included the presence of purchase recommendations.

Customized items The availability of customized items was studied earlier by Srinivasan et al. (2002). In the current research, a website‟s product or service is considered customizable if at least one product variable can be altered.

Customer recognition Wolfinbarger and Gilly (2003) deal with customer recognition in a number of measures. Devaraj et al. (2002) include customer recognition in the Empathy dimension in their research. In the current research, a website is determined to have customer recognition features if:

(29)

29

3.4 Tested sample: Dutch shopping websites

The tested sample consists of the websites of the 100 largest online retail companies in The Netherlands, in terms of revenue over the years 2008 and 2009. The information is derived from the two most recent versions of the Dutch Twinkle 100 magazine (Twinkle 100, 2009; 2010), published in 2009 and 2010. Twinkle 100 specializes in collecting corporate information of organizations exploiting shopping websites. This includes financial data and suppliers. Twinkle‟s data are principally gathered by consulting annual reports, the Dutch Chamber of Commerce, and from other sources that were deemed reliable by Twinkle.

Due to the big overlap of companies that were both listed in 2009 and 2010, the unique number of companies is 121. One of the 121 companies appeared to be a payment service provider, which is a company processing transactions for other companies. As this website cannot be characterized as an online shopping website, it was excluded from the sample. In case companies derive their business from more than one website, only the first-mentioned was evaluated. After this, the final sample consists of 120 websites.

To prevent cultural differences from influencing results, the decision was made to apply the model on a narrow geographic region. Cultural differences could result in different websites structures or layouts, depending on the standards and customs of a particular culture. Also, preferences for making payments or contacting a company may vary among different cultures.

Finally, the selected websites are divided into two groups, in order to observe to what extent similarities or differences are visible. In this study, the tested sample is divided in companies selling physical products, and companies selling non-physical products. Although various other criteria would also be practicable, this distinction is interesting due to the relations of the different product types with service; non-physical products‟ selling companies sell a „service‟, where the service aspect with physical products‟ selling companies is adherent. The former product type will be referred to as services.

(30)

30

3.5 Procedure

To evaluate the websites on the 36 measures of the eight dimensions, every website was observed three times: twice with the Mozilla Firefox web browser and once with Internet Explorer. This method of observation was useful (and necessary) for a number of reasons:

 To detect certain features, the websites should be closed and opened again. The websites should be revisited to detect customer recognition and purchase recommendations.

 To improve reliability of the research and to decrease the risk of making errors, all websites were evaluated an extra time.

 Some measurements could only be done using one of the two web browsers. For instance, checking websites for structural errors by observing the presence of the icon using Internet Explorer seemed rather accurate. Using Firefox, checking for a basic HTML version appeared convenient by switching off the Flash plug-in.

The choice to use Firefox twice and Internet Explorer once - instead of the other way around – has no specific reason.

3.6 Examples of observations

(31)

31

4. Findings

In this section, the findings of the current research will be presented. First, perfect scores by dimension will be presented, followed by scores on the individual measurement item level. The scores are also presented in percentages because the population of 120 websites are unequally divided in physical product oriented (69) and non-physical product oriented (51).Using the percentages, it is more feasible to compare the two groups.

4.1 Results on the dimension level

4.1.1 Perfect scores per dimension

In Table 11, the number of websites with a perfect score is displayed per dimension. A perfect score on a dimension implies that all measures in the specific dimension were rated positively. The results are separated between all websites, physical products selling websites, and non-physical products selling websites. The percentage scores of all three categories have been graphed in Figure 1.

(32)

32

Figure 1. Perfect scores by dimension

Table 11 and Figure 1 show big differences among dimension scores, and predominantly similar scores between physical and non-physical product oriented companies. Over ninety percent of both website categories score perfectly on the system availability dimension. Contrary, no perfect scores are awarded for both categories in the responsiveness dimension. For the contact dimension, ten percent of both categories‟ websites have a perfect score. The dimensions fulfillment, compensation and contact also show rather low scores with 5 to 15 percent. For each of those three categories, physical products selling websites score higher than non-physical products selling websites. This also applies to the personalization and privacy dimensions, where respective scores of approximately 25 and 50 percent were recorded. Non-physical products selling websites score better only on the efficiency dimension, with 35 against 20 percent of the subsets scoring perfectly on all efficiency items.

4.1.2 Average scores per dimension

In Table 12 and Figure 2, the average scores are shown per dimension. This is calculated by summing all positive scores on a dimension‟s items. Columns 1 and 2 represent the total sample scores, columns 3 and 4 show the scores for physical products‟ websites, and columns 5 and 6 display the scores for services oriented websites. Since different dimensions have different amounts of metrics, the percentage columns are added to better compare the dimensions‟ scores.

Table 12. Average scores per dimension

(33)

33 Responsiveness (E) 2.34 46.83% 2.22 44.35% 2.51 50.20% Compensation (F) 1.06 52.92% 1.09 54.35% 1.02 50.98% Contact (G) 1.88 62.78% 1.87 62.32% 1.90 63.40% Personalization (H) 1.92 63.89% 1.90 63.29% 1.90 63.40% (Sample size) N = 120 N= 120 N= 69 N= 69 N= 51 N= 51

Figure 2. Average scores per dimension

Table 12 and Figure 2 show a more nuanced picture of the dimension-level scores. There is a clear distinction between the scores in the former four dimensions versus the latter four dimensions. Dimensions A, B, C and D show twenty to thirty percent higher average scores than dimensions E, F, G and H. Comparing average scores per dimension, differences between the two product types are rather small. Figure Y shows that the biggest difference is in the privacy dimension (D), with 8.6 percent higher scores for the physical products‟ websites.

4.2 Results on the measurement item level

Similar to the results on the dimension level, the results on the measurement item level are presented below in tables and figures, sorted by dimension. In each table, the first two columns deal with the absolute and relative scores for the total sample. The third and fourth column present the scores for physical products selling websites, and the final two columns show the scores for non-physical goods selling websites. All individual item scores can be found in Appendix E (Table 24).

4.2.1 Efficiency items

(34)

34 version of the website (A5) and the full loading of the complete site within twelve seconds (A4), all scores are in the 70 to 80 percent range. Furthermore, the only obvious difference between physical and non-physical product oriented websites is in the sitemap measure (A2): two-third against one-third of the respective samples scored positively.

Table 13. Efficiency scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) Efficiency A1 Graphic quality 120 100,00% 69 100,00% 51 100,00% A2 Sitemap 58 48,33% 23 33,33% 35 68,63% A3 Organized categories 114 95,00% 67 97,10% 47 92,16%

A4 Site load under 12 seconds 93 77,50% 55 79,71% 38 74,51%

A5 HTML version available 89 74,17% 52 75,36% 37 72,55%

Figure 3. Efficiency scores per item

4.2.2 Fulfillment items

(35)

35

Table 14. Fulfillment scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) Fulfillment B1 Order status/tracking 113 94,17% 65 94,20% 48 94,12% B2 Product availability notification 109 90,83% 61 88,41% 48 94,12%

B3 Delivery time notification 100 83,33% 55 79,71% 45 88,24%

B4 Product pictures 104 86,67% 66 95,65% 38 74,51%

B5 Product specifications 119 99,17% 68 98,55% 51 100,00%

B6 Correct search function 90 75,00% 58 84,06% 32 62,75%

B7 Direct online payment 100 83,33% 63 91,30% 37 72,55%

B8 Online ordering possible 112 93,33% 63 91,30% 49 96,08%

B9 Comparison matrix 33 27,50% 17 24,64% 16 31,37%

B10 Recommendation agents 59 49,17% 34 49,28% 25 56,82%

Figure 4. Fulfillment scores per item

4.2.3 System availability items

In Table 15 and Figure 5, the scores for the system availability items are presented. All websites score perfectly on item C1, which implies that all tested websites are always available. Measure C2 shows that approximately five percent of both categories‟ websites have a structural loading error.

Table 15. System availability scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) System availability C1 Always available 120 100,00% 69 100,00% 51 100,00%

(36)

36

Figure 5. System availability scores per item

4.2.4 Privacy items

In Table 16 and Figure 6, the results for the privacy items are presented. For all privacy metrics, physical products oriented websites score better than non-physical products‟ websites. For the availability of an encryption system (D1), option to withdraw from mailing lists (D3), presence of a privacy policy (D5) and third party verification (D6), this yields scores of approximately 90 percent. D1 is the only metric with a clearly lower score for non-physical products‟ websites. Promised personal information protection (D4) scores are just above eighty percent for both categories; D2 is the only metric scoring considerably lower with scores of sixty and 49 percent, respectively.

Table 16. Privacy scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) Privacy D1 Encryption system 104 86,67% 65 94,20% 39 76,47%

D2 Mailing list prevention 67 55,83% 42 60,87% 25 49,02%

D3 Mailing list withdrawal 116 96,67% 68 98,55% 48 94,12%

D4 Personal info protection 99 82,50% 58 84,06% 41 80,39%

D5 Privacy policy 110 91,67% 65 94,20% 45 88,24%

(37)

37

Figure 6. Privacy scores per item

4.2.5 Responsiveness items

In Table 17 and Figure 7, the results for the responsiveness items are presented. The results among the items vary strongly, while the results of the different sites‟ categories are comparable. Both categories score between 80 and 95 percent for FAQ (E1), and return policy (E4). A considerable difference in responsiveness is in the presence of a website survey (E5): it was available at only one of 69 physical products‟ sites, versus twelve out of 51 non-physical products‟ sites. Both categories show very low scores for the availability of an instant help library (E3), and is nearly non-existent for the services websites.

Table 17. Responsiveness scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) Responsiveness E1 FAQ 101 84,17% 55 79,71% 46 90,20% E2 Feedback channel 46 38,33% 25 36,23% 21 41,18%

E3 Instant help library 7 5,83% 6 8,70% 1 1,96%

E4 Return policy 114 95,00% 66 95,65% 48 94,12%

(38)

38

Figure 7. Responsiveness scores per item

4.2.6 Compensation items

In Table 18 and Figure 8, the scores for the compensation metrics are displayed. Both categories score near-perfect on the refund policy item (F2). The presence of a resolution center (F1) could only be detected at nine of 69 physical products‟ websites, and three out of 51 non-physical products‟ websites.

Table 18. Compensation scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (S) Services in % (S%) Compensation F1 Resolution center 12 10,00% 9 13,04% 3 5,88% F2 Refund policy 115 95,83% 66 95,65% 49 96,08%

(39)

39

4.2.7 Contact items

Table 19 and Figure 9 show the results for the compensation items. Again, the results are mixed. The majority of both categories‟ websites serve a telephone number (G1) as a contact method, with the nonphysical products‟ websites scoring better than physical products‟ websites. The latter mentioned group scores better on the availability of an e-mail helpdesk metric (G3). Weak scores for both groups are visible for item G2, the availability of online customer service.

Table 19. Contact scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physical in % (P%) Services (N) Services in % (N%) Contact G1 Telephone number 105 87,50% 57 82,61% 48 94,12%

G2 Online customer service 15 12,50% 8 11,59% 7 13,73%

G3 E-mail helpdesk 106 88,33% 64 92,75% 42 82,35%

Figure 9. Contact scores per item

4.2.8 Personalization items

(40)

40

Table 20. Personalization scores per item

Dimension Measure Total population (T) Total in % (T%) Physical (P) Physica l in % (P%) Services (S) Services in % (S%) Personalization H1 Purchase recommendations 66 55,00% 44 63,77% 22 43,14% H2 Customized products 79 65,83% 32 46,38% 47 92,16% H3 Customer recognition 83 69,17% 55 79,71% 28 54,90%

Figure 10. Personalization scores per item

Electronic Service Quality in Dutch Shopping Websites