RESEARCH

AbstractThe Attraction of Internet to Numerous companies are taking advantage of interactive technology to personalize Web Users their interactions with users. However, in contrast to the proliferation of personalized web services worldwide, Information SHUK YING HO Systems (IS) research on users’ attitudes towards personalized service is minimal. It remains an empirical question as to whether online firms can use personalization as a marketing strategy to attract new users. Our research examines how users’ web experi- ence and their perceptions towards perso- nalization influence their attitudes towards switching to a personalized website. A survey study with 238 subjects was con- ducted. The findings provide empirical evidence showing that if users have involved interactions with their current site, INTRODUCTION recommendations to each individual then personalized services are not attractive and to generate up-sell and cross-sell enough to motivate them to migrate from What is Internet personalization? opportunities. The main objective is one site to another. Users want useful to maximize online merchants’ rev- personalized services, but at the same time, they are concerned with the manner in In the past decade, many online enue. One example is , which firms use their data for personaliza- firms have emerged in almost all which greets the users by name and tion. Practical implications of the findings industries. As the competition inten- lists some book recommendations are discussed. sifies, customer relationship manage- on their individualized pages. The ment (CRM) becomes crucial. recommendations are generated Keywords: personalization, involvement, Online firms are investing large based on the users’ previous pur- information accessibility, perceived amounts of money in providing chases and the preferences of like- usefulness, perceived ease of use, privacy better services for their users. To minded people, and no extra work is remain competitive, Internet busi- imposed on the users. Amazon con- nesses have been adopting differen- tinues to establish its personalization tiating strategies to attract and retain system, and more filtering mechan- users (Bakos 1991). Figure 1 pro- isms are being added to make the vides a taxonomy of CRM systems book recommendations more rele-

Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 based on the level of automation and vant and useful. As stated by Jeff individualization. In this study, we Bezos, the CEO of Amazon, the focus on Internet personalization mission of Amazon is to be the most which is at the highest level in user-centric company in the world Figure 1. Personalization aims to (Anonymous 2002). Other persona- tailor content to individual needs, lization agents are used to (dynami- and to have content arrive at the cally) structure the index of product users at just the right moment. Its pages based on click-stream analysis Authorsimmediate objectives are to under- to minimize the users’ search efforts. Shuk Ying Ho stand and to deliver highly focused, One example is My Yahoo. My

2006 Electronic Markets ([email protected]) is Lecturer for relevant content matched to users’ Yahoo provides a personalized

ß Accounting and Business Information needs and contexts (Albert et al. ‘space’ for each user. It automatically Systems at the University of Melbourne. 2004). The long-term objective is to generates personalized content Her research interests include generate more business opportu- based on the user’s profile. For e-commerce and personalization Volume 16 (1): 41–50. www.electronicmarkets.org DOI: 10.1080/10196780500491162 Copyright nities. instance, it provides information on technology in IS research. Her papers are Companies employ personaliza- the horoscope for the correct star forthcoming in Information Systems tion agents in different ways to sign matched with a person’s date of Research, European Journal of Operational Research, and Journal of generate business opportunities. birth. Apart from automatic perso- Organizational Computing and Some companies use the agents as nalization, it also presents users with Electronic Commerce. advice-giving systems to provide an array of choices and allows the 42 Shuk Ying Ho & Attraction of Internet Personalization

merchants use their data. These studies address issues centred around this dilemma, which is not just a social science issue but also a concern for IS design as restrictions on data collection affect the web mining process. Further, this stream of work proposes technol- ogy frameworks to minimize privacy problems related to personalization. One of the well-known standards is the Platform for Privacy Preferences (P3P), which enables web users to specify their expectations concerning the use of their personal data, and websites to specify their disclosure practices of user data (Cingil et al. 2000; Kobsa 2002). The third category focuses on technologies for mining the enormous amount of user transactions and deriving efficient rules to generate personalized content (Eirinaki and Vazirgiannis 2003; Perkowitz and Etzioni 2000; Ramakrishnan 2000). Works in this area concentrate on computational procedures to analyze transactions and Figure 1. Taxonomy of customer relationship management systems for e-commerce personal profiles. While these studies investigate various aspects of personalization applications, little attention has been given to the basis for understanding the users to select what is of interest to them. The users can relationship between personalization and user responses. personalize not only the content (e.g., weather, finance, sport), but also the layout (e.g., color, background). Recently, the My Yahoo service has been enhanced with Examples of commercial personalization software personalized searching. That is, the users can save pages packages/systems of search results to a ‘personal web’ and block URLs from appearing on the result list. My Yahoo is Many commercial software packages for personalization considered to be one of the forerunners among the are available for small-sized and medium-sized compa- growing number of personalized websites that have been nies. Examples include One-to-One (Broadvision), emerging on the Internet over the past few years Dynamo Relationship Commerce Suite (Art (Manber et al. 2000). Technology Group), Personalization Manager (Net Perceptions) and ADAPTe (ResponseLogic), all of which supply the full range of e-commerce personaliza- Existing information systems (IS) research on Internet tion applications. These packages use artificial intelli- personalization gence and rule-based systems to specialize applications

Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 such as personalization of and product The proliferation of personalized web services worldwide recommendations at online sites. They provide one-to- is drawing increased attention from IS researchers. one marketing and sales solutions and customer-driven Previous work related to personalization falls into three websites with targeting engines. The e-businesses can main categories. The first category includes applications ultimately operate more efficiently by aligning supply of personalization technology. Personalization agents chains with data concerning the composition of the are found to be useful in different domains. These market and purchase patterns to build user profiles, include information dissemination (Light and which in turn determine personalized recommendations Maybury 2002; Loeb 1992), entertainment recommen- (see Figure 2). dations (Smyth and Cotter 2000), search engines The concept of personalization is not limited to online (Manber et al. 2000), and medicine (Bental and stores, and it can be applied to search engines. In 2004, Cawsey 2002). two personalized search engines, A9.com by an Amazon The second category focuses on philosophical issues, subsidiary and MyJeeves by Ask Jeeves, were launched to such as privacy regulations and privacy ethics related to let users store individual search results and then provide data collection and processing (Cingil et al. 2000; Kobsa personalized web searches. analysts expect 2002; Volokh 2000). In practice, website-tracking is other major search engines, such as and Yahoo, performed in an unobtrusive manner without the to soon follow suit with their own personalized services. awareness of the users. Although users demand more Global investment in personalization technologies will customized services, they are increasingly concerned grow from USD500 million in 2001 to USD2.1 billion about the threats to their privacy and how online in 2006 (Anonymous 2001). Electronic Markets Vol. 16 No 1 43

Figure 2. Generation of personalized recommendations

Motivations and research questions Is Internet personalization an effective marketing strategy to attract new users? What are the reasons that There is much publicity about delivering personalized users switch to a personalized website? Our work services over the web and the stakes are high for vendors conducts a survey study among web users to address selling related products. In spite of that, our under- these questions. standing of the impacts of Internet personalization is far The remainder of the paper is organized as follows: from conclusive. Advocates of Internet personalization the following section provides the theoretical back- claim that personalization agents have changed the web ground and states the research hypotheses. The study is into a personal communication medium. By providing then introduced, followed by a consideration of the individualized content, offerings and services, Internet implications of the study. A conclusion and discussion of personalization helps to control aimless surfing activity some limitations of the study are presented in the final (Light and Maybury 2002; Pitkow et al. 2002; Shahabi section. and Banaei-Kashani 2003). Also, personalizing web

Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 content empowers online merchants to deliver user value and to achieve profitable growth (Greer and BACKGROUND AND HYPOTHESES Murtaza 2003; Peppers and Rogers 1997). It has been reported that e-commerce websites using personaliza- Website switching occurs when a user or group of users tion technology have seen annual revenue increases of up switches their allegiance from one website providing a to 52% (Parkes 2001). certain type of services/products to another. This But there remains scepticism on the prospects of website switching may be temporary or permanent. Internet personalization. Nielsen (1998) claimed that For instance, if the Amazon website is not available, a Internet personalization was highly over-rated. user may browse or even purchase from Barnes and Personalization technology is unnecessary if the web Noble. The switch may be limited to just this one architecture is well-designed and content is well- occasion, or it may be longer lasting. organized. Festa (2003) remarked that online merchants In this research we develop a model of website- seeking to personalize their websites in the hope of switching behaviour to study the situation in which users boosting online sales are not getting the expected currently use site A that has no personalization, and to payback. It costs about four times as much to see whether they will switch to site B that does have personalize a website than to run a comparable adaptive personalization. We believe the features of both the site. A recent report by Jupiter Research (2003) current website and the available website are determin- indicated that only 14% of users think that personalized ing factors influencing users’ switching behaviour. On offers or recommendations on shopping websites lead one hand, users rely on experience and their often long- them to purchase more frequently. held attitudes about the current website. Personal 44 Shuk Ying Ho & Attraction of Internet Personalization

experience can be an important factor influencing users’ be a crucial factor affecting their web experience attitudes. On the other hand, incorporation of persona- (Wolfinbarger and Gilly 2001). lization technology in a website acts as advertising. An important website design element is the organiza- Personalization can play an important role in attracting tion of information (Garrett 2002). The tremendous new users. That is, if the users are unsatisfied with their amount of information released on the web creates current site and there exists an alternative that uses the problem of information overloading. This greatly personalization and the users perceive it to be valuable, reduces the navigation capabilities of online users, which then website-switching may take place. in turn negatively affects their satisfaction towards the In the following, we divide the factors that motivate website. Thus, IS researchers proposed and confirmed or inhibit a user’s switching to a personalized website that information relevance (McKinney et al. 2002) and into two categories: (1) experience with current accessibility (Yang et al. 2005) are important web quality websites, and (2) perceptions of personalized websites. factors affecting user satisfaction. As mentioned earlier, personalization is a newly developed technology which aims to solve the above problem. That is, personalization Experience with current websites tailors content to individual needs, reduces the amount of delivered content and increases information accessi- This category of variables focuses on the features of the bility. This aligns with the claims by Jeff Bezo, current websites. Marketing literature on brand switch- personalization can dramatically improve their users’ ing considers two groups of variables, satisfaction chances of finding what they want (even if they didn’t derived out from the website and involvement level know it was what they wanted) from a 1,000 to 1 chance (Shukla 2004). Thus, our study explores how users’ in a regular bookstore, to a 50 to 1 chance at pleasant navigation experiences and their involvement in Amazon.com (Anonymous 1998). Thus, we anticipate the current website influence their switching decision. that the low accessibility of relevant content of the We anticipate that unpleasant experiences caused by current website is a factor motivating one to move to a difficulty in locating relevant content at the current personalized site. website and low involvement will lead to a higher tendency toward website switching. Hypothesis 2: If a user cannot access relevant content in the current website, it is more likely that he/she will switch to a Level of involvement (I). User involvement is of interest personalized website. to researchers because of its potential impact on willingness to revisit the website. Prior research shows that users’ level of involvement has a direct, positive and significant impact on their willingness to purchase Perceptions of the personalized website (Gammack and Hodkinson 2003), their motivation to process the information (MacInnis and Jaworski 1989), This category of variables covers how web users perceive the way they process information (Johnson and the new personalized websites. Compared with a Eagly 1989, 1990; MacInnis and Jaworski 1989), and physical channel, e-commerce moves the information Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 their switching behaviours from one brand/product technology to the foreground and web users interact to another (Keaveney and Parthasarathy 2001). with the technology directly (Koufaris 2002). Thus, the Involvement is a motivational factor. If users are two variables of the Technology Acceptance Model involved in a website, then they recognize the good (Davis 1989), perceived usefulness and perceived ease of features of the site easily, and prefer and insist upon use, are included in our study. In addition, since using it. Ultimately, they repeatedly visit it with little personalization arouses a list of social ethical issues, we thought, and this leads to higher reluctance to switch to also include a construct called ‘privacy concerns of the other websites. users’ and study its effects on users’ switching decision.

Hypothesis 1: If a user is highly involved in a website, then it is Perceived usefulness (U) and perceived ease of use (E) less likely that he/she will switch to a personalized website. of Internet personalization. In IS literature, innovation adoption research provides a theoretical framework with Accessibility of relevant information (R). Internet users which to identify the innovation-related factors in order lack physical contact with the stores or products. Thus, to assess their impacts on adoption decisions. One of the users’ experience is based purely on website information. established models is the Technology Acceptance Model This causes users to rely heavily on technology and (TAM), which has been applied extensively to identify information quality to keep them interested and serviced factors that facilitate or inhibit the adoption of an as they explore the site with ease and pleasure innovation (Al-Gahtani and King 1999; Bajaj and (McKinney et al. 2002). The design elements of Nidumolu 1998; Chin and Gopal 1995; Venkatesh the site, such as ease of navigation, can, therefore, et al. 2003). In this study, two of its constructs, Electronic Markets Vol. 16 No 1 45

perceived usefulness and perceived ease of use of communications generally have higher computer knowl- personalization services, are studied. Perceived useful- edge, the results would obviously be biased towards ness is defined as the degree to which a person believes intensive Internet usage and active web behaviours. To that using a particular system would enhance his/her job reduce the bias, we distributed a paper survey on the performance (Davis 1989; Venkatesh et al. 2003). In the university campus. We had a random sample of 200 context of a personalized website, perceived usefulness is people, and collected 33 (516.5%) responses. Overall, regarded as the level of matching between personalized we collected 238 completed responses. offers and one’s preferences. Perceived ease of use is Of the respondees, 160 were male and 78 female. A defined to be ‘the degree to which a person believes that Kruskal-Wallis test revealed no significant differences in using a particular system would be free of effort’ (Davis the gender ratio (approximately 66% males and 34% 1989; Venkatesh et al. 2003). The ease of use of user- females) of subjects across the two methods of data centric software is very important (Agarwal and collection. Their average age was 26.39 years old. All Karahanna 2000). In our context, it refers to the ease subjects were computer users. On average, they spent of operating a personalization agent on the client side. 4.32 hours on computer work or web browsing every The following hypotheses are proposed: day and they had 4.17 years of Internet browsing. At a 5% level of significance, F-tests showed that age, Internet Hypothesis 3: When a user perceives personalized services to be experience, and Internet usage of the subjects did not useful, he/she will switch to a personalized website. differ significantly across the two groups. Nearly all of them had general knowledge of online Hypothesis 4: When a user perceives personalized services to be shopping, online payment and e-banking, and 70.59% easy to use, he/she will switch to a personalized website. (5168/238) reported that they had shopped online, such as movie tickets purchase and grocery shopping Privacy concerns (C). Apart from the technological from online supermarkets. More than 60% had experi- factors, much IS research also focuses on ethical issues ence with or had heard about personalized websites, related to personalization, and related technical solu- such as Amazon and My Yahoo. tions (Caulfield et al. 2000; Cingil et al. 2000; Kobsa 2002; Kramer et al. 2000; Stewart and Segars 2002; Volokh 2000). Users face a dilemma: although they Measurements of variables demand more customized services, they are increasingly concerned about privacy infringements and how their The statistical analysis was performed with SPSS version information is being used by online merchants. There 9.0. To assess convergent and discriminant validity, we are concerns that their purchase histories and navigation factor-analysed the items for the instruments. Table 1 behaviours could be analysed and abused (Nash 2000; shows results of the factor analysis and reveals that the Pitkow et al. 2002). Thus, we hypothesize that these item loadings were consistent with five distinct theore- concerns will affect the probability of switching. tical constructs. A five-factor solution was obtained with all component eigenvalues greater than one. The factors Hypothesis 5: Privacy concerns discourage a user from switching were level of involvement (I), accessibility of relevant Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 to a personalized website. information (R), privacy concerns (C), perceived useful- ness (U) and perceived ease of use (E) of Internet personalization. These factors explained 68.23% of the total variance in the survey. Items loaded highly (.0.70) THE STUDY METHODOLOGY on their associated factors. The Cronbach alpha values for the constructs are shown in Table 2. Consequently Samples we concluded the instruments were statistically valid and reliable. A self-administrated survey was conducted. The ques- tionnaire consisted of 45 questions and all questions were measured on a 5-point Likert scale. The time Results required to complete the whole questionnaire was 20 minutes. In order to have a larger coverage, we A multiple regression was conducted, with the likelihood disseminated the survey in two ways: electronic survey of switching to a personalized website as the dependent and paper survey. variable. The R-square of the model was 0.316. Table 3 First, we sent emails to 350 full-time staff and students shows the regression model. in a public university. The subjects were invited to take Involvement with the current website was a significant part in the study and to complete the questionnaires factor that influenced whether users switched to a online. From these 350, 205 (558.57%) completed website or not. Users who were highly involved in the responses were returned. Since users who use emails for website showed higher resistance against switching to a 46 Shuk Ying Ho & Attraction of Internet Personalization

Table 1. Rotated matrix component from factor analysis

UERCI

I1 0.303 0.113 20.022 0.089 0.888 I2 0.073 0.124 20.042 0.245 0.907 R1 0.076 20.032 0.843 20.053 0.020 R2 0.039 0.003 0.727 20.019 20.050 R3 20.119 20.084 0.736 20.070 20.058 R4 20.013 20.008 0.826 0.026 0.032 U1 0.964 0.031 20.009 0.115 0.094 U2 0.920 0.040 20.075 0.062 0.044 U3 0.886 0.093 0.033 0.161 0.137 U4 0.949 0.030 0.036 0.098 0.099 U5 0.914 0.079 20.003 0.145 0.086 E1 0.080 0.903 20.075 20.021 0.041 E2 0.048 0.743 20.078 0.019 0.084 E3 0.046 0.880 20.040 0.079 20.025 E4 0.003 0.910 0.054 0.014 0.021 E5 0.027 0.770 20.064 0.027 0.094 E6 0.068 0.866 0.056 20.059 0.075 C1 0.248 0.015 20.085 0.935 0.150 C2 0.222 0.016 20.041 0.936 0.184

Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Rotation converged in 5 iterations. Level of Involvement (I) Accessibility of Relevant Contents (R) Perceived Usefulness of Internet Personalization (U) Perceived Ease of Use of Internet Personalization (E) Privacy Concerns (C)

new website. The result was consistent with Hypothesis such as search engines and taxonomy hierarchies, to 1(t522.95, p,0.05). facilitate their user searching, so that the accessibility Inconsistent with Hypothesis 2, accessibility of of relevant information would be greatly increased. relevant information is not a significant factor affecting Personalization is not necessary. Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 users’ decision (t520.695, p.0.1). That is, users would Participants who perceived personalization to be not switch to a personalized site from a site whose useful were willing to switch to a personalized website content was difficult-to-access or badly organized. There (t53.53, p,0.01), supporting Hypothesis 3. Generally, can be two reasons. First, users do not consider the score of the usefulness of a personalization agent was personalization agents to be a tool to solve the not high (mean52.66), and the standard deviation was information-overload problem. Second, in addition to large (SD51.59). This illustrates that the subjects had personalization agents, there are many other alternatives, various opinions on the usefulness of the agents. To probe further into the topic, we asked some subjects to specify their expectations of the functionality of a Table 2. Reliability personalization agent, and compared the expectations from subjects who perceived the agent to be useful and Constructs Items Reliability those who perceived it to be useless. Those perceiving Level of involvement (I) 2 0.872 the agent to be useless had high expectations, including Accessibility of relevant contents (R) 4 0.791 recommendations matching contexts and tasks, and Perceived usefulness of Internet 5 0.964 layout customization. Those perceiving the agent to be personalization (U) useful had general expectations, including preference- Perceived ease of use of Internet 6 0.922 matching recommendations, personal purchase history personalization (E) tracking, and delivery-tracking. A large difference in Privacy concerns (C) 2 0.945 expectation of functionality leads to a large difference in usefulness perception. Electronic Markets Vol. 16 No 1 47

Table 3. Multiple regression model

Unstandardized Coefficients Std Coeff B Std. Error Beta t Sig.

(Constant) 3.880 .529 7.342 .000 Level of Involvement (I) 20.270 .091 2.184 22.952 .003 Accessibility of Relevant 20.047 .067 2.043 2.695 .488 Contents (R) Perceived Usefulness (U) 0.262 .074 .220 3.525 .001 Perceived Ease of Use (E) 20.005 .082 2.004 2.064 .949 Privacy Concerns (C) 20.143 .067 2.132 22.123 .035

Contradictory to Hypothesis 4, ease of use was not a personalization. The current work is among only a few significant factor (t520.64, p.0.1). Usefulness domi- pioneering efforts (Moe 2003) that empirically assess the nated ease of use, and this aligns with some IS studies effects of Internet personalization. We believe that the (e.g., Davis 1993; Elbeltagi et al. 2005). From a findings of the current work are applicable to a wide practical point of view, it may imply that users use a range of web-based services that target the attraction of technology mainly because of the usefulness of the users. There are five practical implications: technology itself. From a research point of view, the 1. Our work indicates that personalization can be a difference between the findings of prior studies (sig- successful marketing strategy in some types of nificant influence of ease of use on behavioural inten- website only. The nature of the interactions tion) and this study (insignificant influence) may be due occurring among web users and their sites was an to the nature of the sample. Web users today are important factor. If the interactions are frequent generally more computer literate than their counterparts and involved, web users are more reluctant to five to ten years ago. This was supported by the switch to a new site, even though the new site descriptive statistics. The subjects perceived personaliza- provides personalized services. Those e-businesses tion agents as not difficult to use (mean54.51, cannot use personalization as a marketing strategy SD51.44). Hence, ease of use may have been less of to attract new users from a high-involvement site to an issue for this sample than it would have been for the a personalized one. Examples of high-involvement samples used in prior studies. Owing to this general websites included network games and online chat improvement of computer literacy among web users, the rooms. Online stores selling high-involvement relationships found to be valid in prior work may need to products, such as cosmetics, are also included. be re-examined. 2. Content overloading is not a reason to incorporate

Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 Privacy infringement from personalization impacted a personalization agent into a site. Examples of negatively on users’ decisions to switch to a personalized content-overloading sites include knowledge-based 52 , website, supporting Hypothesis 5 (t 2.12, p 0.05). websites (e.g., Microsoft.com) and online stores The subjects were concerned with how online merchants with millions of products (e.g., eBay.com). This used their purchase transactions to generate the perso- aligns with the argument by Nielsen (1998). If the nalized recommendations (mean53.41, SD5.76). owners of content-overloading websites plan to Some subjects raised questions about how much data facilitate the navigation of the users, to retain the were tracked by online merchants. Some even ques- users and to increase their satisfaction, they could tioned whether the merchants would sell their profiles choose to restructure the information architecture and transaction details to some marketing survey of the site or to use a search engine, rather than to companies or other online merchants. employ a personalization agent. 3. Our empirical findings show that perceived useful- ness of personalization is a significant factor in PRACTICAL IMPLICATIONS attracting new users. Thus, it is worthwhile for firms to invest in data mining to analyse the transaction In contrast to the widespread adoption of personaliza- patterns among like-minded people. However, web tion software and the strong advocacy by management visitors may not look for a product for purchase. gurus (Albert et al. 2004; Bakos 1991) on the use of They navigate the web for different purposes, such personalization services as a differentiating strategy, little as random browsing to kill time. Thus, simply has been done to assess the effectiveness of Internet analysing the transaction log may not generate the 48 Shuk Ying Ho & Attraction of Internet Personalization

best personalized recommendations. Online firms services, but at the same time, they are concerned with need to understand both users’ past transactions the manner in which firms use their data for personaliza- and their current context in order to generate useful tion. If users have involved interactions with their recommendations. Therefore, it becomes a chal- current site, then personalized services are not attractive lenge to integrate server-side logged data with real- enough to motivate them to migrate from one site to time click stream analysis. This leads to more room another. It provides insights to online merchants who for future research. plan to incorporate personalization tools in their 4. While Internet personalization has been shown to websites. be effective in attracting new users, its use should be However, this study has three limitations. First, balanced by taking a proactive approach towards although we conducted both email and paper surveys, protecting data privacy on the users’ side. the number of respondents for the email survey was According to our findings, privacy concern is found much more than for the paper survey. Hence, there to be a statistically significant factor and it might be a significant bias in favour of IT use and negatively affects subscribers’ decisions to switch expertise. This might also be the reason why our sample to a personalized website. However, the standar- considers personalization agents to be easy to use. dized coefficient of perceived usefulness of perso- Further studies on how different groups of users (e.g., nalization is 1.5 times that of privacy concern. This IT experts vs. IT novices) evaluate personalization agents implies that regarding personalized recommenda- in different contexts (e.g., online shops vs. information tions, our sample has more concerns about its websites) would be interesting. Second, this paper only usefulness, and its likely relevance and accuracy of addressed users’ decisions to switch to a new website. preference-matching, than about privacy issues. Would they rely on the personalized recommendations Since our subjects have higher IT knowledge than after switching to the new websites? Would they an average user, they understand the personaliza- purchase more? Future studies on the difference in tion process more, and thus, have fewer concerns exploration between personalized and general content about privacy. Whether users with lower IT knowl- could be interesting. Third, since our work focuses on edge still consider perceived usefulness to be more whether personalization influences users’ behaviours to important than privacy concerns can be an interest- switching to a personalized website, we do not know ing question for future research. A recent survey how the users behave after switching to a personalized conducted by the Personalization Consortium site. Can personalization services retain them? Would reports that while 76% of the respondents find it they develop a sense of loyalty and trust in the long run? helpful for websites to remember basic personal Or would they switch back to a website providing information such as names and email addresses, general content? A longitudinal field study might only 50% read the privacy statement and only 38% contribute much to this area of research. of the respondents agree that privacy statements are easy to understand. Online firms should make sure that users have equitable access to their personal References

Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 information and should provide a mechanism for Agarwal, R. and Karahanna, E. (2000) ‘Time Flies When users to quickly change and keep track of inaccurate You’re Having Fun: Cognitive Absorption and Beliefs or unauthorized personal information. about Information Technology Usage’, MIS Quarterly 5. Lastly, the implications are not limited to a 24(4): 665–94. business-to-consumer context. It can be applied to Al-Gahtani, S. and King, M. (1999) ‘Attitudes, Satisfaction an Intranet portal of a company. For instance, and Usage: Factors Contributing to Each in the Acceptance personalization can be used to improve the infor- of Information Technology’, Behavior and Information mation flow of in-house services and interactions Technology 18(4): 277–97. between a firm and its employees. A large corporate Albert, T. C., Goes, P. B. and Gupta, A. (2004) ‘GIST: A Intranet can personalize its content to attract the Model for Design and Management of Content and employees to maximize the effectiveness of infor- Interactivity of User-Centric Websites’, MIS Quarterly mation dissemination. 28(2): 161–82. Anonymous (1998) ‘Earth’s Most Customer Centric Company: Differentiating with Technology’ online at CONCLUSIONS AND FUTURE RESEARCH http://mitworld.mit.edu/video/1/, accessed 12 July 2005. To summarize, our work investigates the motivating Anonymous (2001) ‘Personalization Technology Market and inhibiting factors for online users to switch to a Booming’ online at http://www.nua.ie/surveys/ personalized website. Our work provides empirical index.cgi?f5VS&art_id5905357178&rel5true, accessed 1 evidence showing that users want useful personalized October 2005. Electronic Markets Vol. 16 No 1 49

Anonymous (2002) ‘A Bookstore by Any Other Name’ online Keaveney, S. M. and Parthasarathy, M. (2001) ‘Customer at http://www.commonwealthclub.org/archive/98/98- Switching Behavior in Online Services: An Exploratory 07bezos-intro.html, accessed 12 July 2005. Study of the Role of Selected Attitudinal, Behavioral, and Bajaj, A. and Nidumolu, S. R. (1998) ‘A Feedback Model to Demographic Factors’, Academy of Marketing Science Understand Information System Usage’, Information and 29(4): 374–90. Management 33(4): 213–24. Kobsa, A. (2002) ‘The Adaptive Web: Personalized Bakos, J. Y. (1991) ‘A Strategic Analysis of Electronic Hypermedia and International Privacy’, Communications of Marketplaces’, MIS Quarterly 15(3): 295–310. the ACM 45(5): 64–7. Bental, D. and Cawsey, A. (2002) ‘The Adaptive Web: Koufaris, M. (2002) ‘Applying the Technology Acceptance Personalized and Adaptive Systems for Medical Consumer Model and Flow Theory to Online Consumer Behavior’, Applications’, Communications of the ACM 45(5): 62–3. Information Systems Research 13(2): 205–23. Caulfield, B., Carr, D. F. and Murphy, K. (2000) ‘Playing with Kramer, J., Noronha, S. and Vergo, J. (2000) ‘A User- Plutonium: How to Handle User Data without Inviting a Centered Design Approach to Personalization’, Privacy Meltdown’, Internet World 6(14): 52–8. Communications of the ACM 43(8): 44–8. Chin, W. W. and Gopal, A. (1995) ‘Adoption Intention in Light, M. and Maybury, M. T. (2002) ‘The Adaptive Web: GSS: Relative Importance of Beliefs’, The Data Base for Personalized Multimedia Information Access’, Advances in Information Systems 26(2/3): 42–63. Communications of the ACM 45(5): 54–9. Cingil, I., Dogac, A. and Azgin, A. (2000) ‘A Broader Loeb, S. (1992) ‘Architecting Personalized Delivery of Approach to Personalization’, Communications of the ACM Multimedia Information’, Communications of the ACM 43(8): 136–41. 35(12): 39–47. Davis, F. D. (1989) ‘Perceived Usefulness, Perceived Ease of MacInnis, D. J. and Jaworski, B. J. (1989) ‘Information Use and User Acceptance of Information Technology’, MIS Processing from Advertisements: Toward an Integrative Quarterly 13(3): 319–39. Framework’, Journal of Marketing 53(4): 1–24. Davis, F. D. (1993) ‘User Acceptance of Information McKinney, V., Yoon, K. and Zahedi, F. M. (2002) ‘The Technology: Systems Characteristics, User Perceptions and Measurement of Web-Customer Satisfaction: An Behavioral Impacts’, International Journal of Man-Machine Expectation and Disconfirmation Approach’, Information Studies 38(3): 475–87. Systems Research 13(3): 296–315. Eirinaki, M. and Vazirgiannis, M. (2003) ‘Web Mining for Manber, U., Patel, A. and Robison, J. (2000) ‘Experience with Web Personalization’, ACM Transactions on Internet Personalization on Yahoo!’, Communications of the ACM Technology 3(1): 1–27. 43(8): 35–9. Elbeltagi, I., McBride, N. and Hardaker, G. (2005) Moe, W. W. (2003) ‘Buying, Searching, or Browsing: ‘Evaluating the Factors Affecting DSS Usage by Senior Differentiating between Online Shoppers Using In-Store Managers in Local Authorities in Egypt’, Journal of Global Navigational Clickstream’, Journal of Consumer Psychology Information Management 13(2): 42–65. 13(1/2): 29–39. Festa, P. (2003) ‘Report Slams Web Personalization’ online at Nash, K. S. (2000) ‘Personalization: Clash of the Killer Ps’, http://www.news.com.com/2100-1038-5090716.html, accessed 21 October 2004. Computerworld 34(25): 66–7. Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010 Gammack, J. and Hodkinson, C. (2003) ‘Virtual Reality, Nielsen, J. (1998) ‘Personalization is Over-rated’ online at: Involvement and the Customer Interface’, Journal of End http://www.useit.com/alertbox/981004.html, accessed 1 User Computing 15(4): 78. May 2005. Garrett, J. J. (2002) The Elements of User Experience: User- Parkes, C. (2001) ‘The Power of Personalization: Web CenteredDesignfortheWeb,1stedn,NewYork:NewRidersPress. Technologies Get Personalized to Increase Profits’ online Greer, T. H. and Murtaza, M. B. (2003) ‘Web Personalization: at: http://www.esj.com/features/ The Impact of Perceived Innovation Characteristics on the article.asp?EditorialsID535, accessed 1 May 2005. Intention to Use Personalization’, Journal of Computer Peppers, D. and Rogers, M. (1997) Enterprise One to One: Information Systems 43(3): 50–5. Tools for Competing in the Interactive Age, 1st edn, New Johnson, B. T. and Eagly, A. H. (1989) ‘Effects of York: Doubleday. Involvement on Persuasion: A Meta-analysis’, Psychological Perkowitz, T. M. and Etzioni, O. (2000) ‘Adaptive Websites’, Bulletin 106: 290–314. Communications of the ACM 43(8): 152–8. Johnson, B. T. and Eagly, A. H. (1990) ‘Involvement and Pitkow, J., Schu¨tze, H., Cass, T., Cooley, R., Turnbull, D., Persuasion: Types, Traditions, and the Evidence’, Edmonds, A., Adar, E. and Breuel, T. (2002) ‘Personalized Psychological Bulletin 107: 375–84. Search’, Communications of the ACM 45(9): 50–5. ‘Jupiter Research Reports That Web Site ‘‘Personalization’’ Ramakrishnan, N. (2000) ‘Pipe: Web Personalization by Does Not Always Provide Positive Results’ Jupiter Research Partial Evaluation’, IEEE Internet Computing 4(6): 21–31. Press Releases, online at: http://www.jupitermedia.com/ Shahabi, C. and Banaei-Kashani, F. (2003) ‘Efficient and corporate/releases/03.10.14-newjupresearch.html, Anonymous Web-Usage Mining for Web Personalization’, accessed 1 October 2005. Journal of Computing 15(2): 123–47. 50 Shuk Ying Ho & Attraction of Internet Personalization

Shukla, P. (2004) ‘Learning and Information Dissemination in Toward a Unified View’, MIS Quarterly Logistics Alliances’, Asia Pacific Journal of Marketing and 27(3): 425–78. Logistics 16(4): 82–104. Volokh, E. (2000) ‘Personalization and Privacy’, Stewart, K. A. and Segars, A. H. (2002) ‘An Empirical Communications of the ACM 43(8): 84–5. Examination of the Concern for Information Privacy Wolfinbarger, M. and Gilly, M. (2001) ‘Shopping Online for Instrument’, Information Systems Research Freedom, Control and Fun’, California Management 13(1): 36–49. Review 43(2): 34–55. Smyth, B. and Cotter, P. (2000) ‘A Personalized Television Yang, Z., Caib, S., Zhouc, Z. and Zhoua, N. (2005) Listings Service’, Communications of the ACM 43(8): ‘Development and Validation of an Instrument to Measure 107–11. User Perceived Service Quality of Information Presenting Venkatesh, V., Morris, M. G., Davis, G. B. and Davis, F. D. Web Portals’, Information and Management 42(4): (2003) ‘User Acceptance of Information Technology: 575–89. Downloaded By: [Schmelich, Volker] At: 11:26 24 March 2010