Empirical Studies

---

The empirical work is organized as follows. We begin by briefly examining general willingness to purchase items online, to contextualize the relative purchase likelihood of those products of specific interest. This is followed by a survey of Web sites selling surfboards. As typical major reasons for shopping cart abandonment relate to checkout processes (Vividence, 2001), we also present a study of a contemporaneous sample of Web sites to show some norms in this regard. We then summarize a commercial study of a swimwear e-tailer (Gammack and Carter, 2001) which explicitly redesigned the interface to overcome the problem of shopping cart abandonment. We then describe the development of a e-tailing virtual environment, Beachtown, which allows experimental examination of end user interactivity, and finish with a study of surfboard purchase willingness.

Study 1: Australian E-Tailing Mini-Survey ^[1]

The first of our empirical studies concerns a preliminary online survey aimed at benchmarking the current general likelihood of online purchase for a range of products. Twelve questions, including demographic ones, were listed on an Australian surfing oriented Web site page, advertised and completed online in early 2002. Sixty-eight respondents from 10 countries replied, with ages ranging from 12 to 60, averaging just under 29. Gender was roughly balanced, with 53% female and 46% being males. The number of respondents from each country was as follows: Australia (24), Japan (9), UK (2), USA (1), France (11), Germany (8), India (7), Sweden (1), Canada (3) and two unknown. Question 6 asked about products liable to be bought online: results are shown in Figure 1.

Figure 1: Relative Likelihood for 68 Subjects of Products Liable to be Bought Online

These figures indicate that all categories would at least be considered, that almost everyone would buy tickets and t-shirts online, and a small percentage would even buy or consider buying a vehicle ^[2]. Tickets, the most popular category, are naturally suited to online purchase as it is their information content, rather than their quality that is relevant, and similarly with software. Of the apparel categories the comfort fit and low risk associated with t-shirts may be considered to be less critical than the fitting of shoes, with jeans somewhere in between, accounting for the relative likelihoods of those categories. Finally, around 2/3 of respondents would buy sporting equipment online, although it might be argued that selection of these items depends also on customer brand knowledge and experience, and the individual's level of perceived risk of unwittingly purchasing a counterfeit product online.

A number of factors were preselected as likely to influence purchase and subjects were asked to rank these in question 10. The reasons given as to the importance of each factor, in order were: price (cheaper than shops), security of payment, advertised on conventional media, known or unique product, speed of order fulfillment, with the least important two factors being the company having a local office and a detailed product description. This latter conflicts somewhat with the findings of Bellman et al. (1999) who found that looking at product information was the major predictor of online buying, but our minisurvey affirms Bellman et al.'s (1999) finding that security remains a major concern. Although the small numbers, self-selection and subjective reporting of the subjects limit the survey, it suggests the potential for a future detailed use of this method for our research purposes, paying attention to these factors.

Study 2: Surfboards Online

The sporting equipment category was introduced, as there was an intention to specifically examine the possibilities for marketing surfboards online. The substantial number of respondents who have already purchased, or would consider making this purchase suggests that the category is more similar to apparel than to groceries, and so an analysis of the current situation in Australian online surfboard vendors was undertaken. Surfboards can be bought "off the rack" but are generally custom built to specifications. The manufacturing process is also customized craftsmanship by "shapers", as mass production is not a possibility for this product. It is notable that the specifications for a board can be established online, and repeat purchases can be negotiated with shapers to increase customization. This process was to inform subsequent research on purchase of high involvement, hedonic products via the Web. A search was performed on the Google directory for the category "Business > Industries > Manufacturing > Consumer Products > Sporting Goods > Water Sports > Surfing" and also for "Shopping > Sports > Water Sports > Surfing > Surfboards." In mid-2002 this analysis showed 41 entries in the latter category, nine of which (22%) were Australian companies or stores. Some 71% offered custom-built surfboards, with three (7%) offering surf apparel only and a further three offering both custom boards and apparel. Seven sites offered other products comprising skim boards, body boards, board fins and board wax. None of the custom surfboard manufacturers offered Web ordering, but instead requested details of the desired board, customer surfing preferences, details and experience. The board shapers would then respond to the customer by discussing needs in detail and negotiating appropriate product custom features. The objective of this feature negotiation technique is the supply of a surfboard appropriate to the surfer's skill level and surfing needs, despite a lack of board design knowledge. The frequent lack of buyer knowledge about the finer points of surfboard design is evidenced by both the sales technique and the fact the majority of sites carried considerable educational material on the topic.

From the marketing viewpoint the sales approach is predictable due to the high involvement nature of the good, its complex features, and the customer's perceived risk, which includes financial, social and performance dimensions. Furthermore a surfboard is a complex product about which effective satisfaction judgments can only be made after a considerable period of usage experience. Consumers thus have to "live through" the period of cognitive dissonance and, in the case of a surfboard, have to merge their own skill levels and the board's capabilities. Those eventual post consumption satisfaction judgments, combined with the high involvement nature of the good, can produce consumers who are highly motivated to make positive or negative word of mouth recommendations within the close-knit surfing community. The selection of an appropriate surfboard, even if it involves considerable time and expense on the part of the retailer, is an essential part of generating positive consumer ascription and building their particular brand of surfboards.

In terms of site content almost all sites offered high resolution photographs, many offered surfing clips, surf reports and beach cams which offered increased customer involvement with surfing and enhanced the shopping experience. Some 10% of sites also offered music or sound effects. Not one of the 29 Web sites offered surfboards via direct e-tailing facilities, although surf wear and accessories were. Animations were commonly used, leading to some delays in screen loading. However, such problems will decrease in future, as data speeds increase. Advanced site features were rare, as only two sites offered computer generated visualization of the surfboard in response to the range of specific dimensions entered by the potential buyer.

It is clear from this brief survey of Web surf sites that the industry recognizes that novel, pleasant, or involving surfing images and features such as film clips, animations, music and surf travelogues tend to increase attention and situational involvement. This can gain increased attention and absorption, lead to consumer immersion in the surfing culture and experience, and perhaps even move the Web shopper towards the flow state (Csikszentmihalyi, 1990) as posited by Hoffman and Novak (1995). Although the technology exists in games arcades, current technical limitations, including hardware and bandwidth, prevent the "test riding" of the shopper's custom surfboard in Virtual Reality before construction begins. However, there are incremental steps in that direction that can be taken, that can be tested on real life shoppers, simply because such features have real value in such an experiential and high involvement product class. The manipulation of relevant variables and the provision of increasingly "VR" features offer a rich research agenda with commercial application.

Study 3: An Analysis of Purchase Screens for E-Tailing Sites

A minisurvey of Web site marketing and particularly sales closure methods was conducted utilizing a convenience sample of 32 of some of Australia's most popular commercial Web sites identified in a list provided by Hitlist.com, who aggregate relevant Australian information. This sampling method was adopted to ensure the inclusion of a diverse range of businesses. The findings will be discussed in terms of the businesses included in the sample, the approach to selling, including the information and facilities provided to prospective buyers, and the mode of sales closure.

The sample comprised 26 retail businesses-including seven computer retailers, six clothing businesses, two supermarket chains. In addition, two service business-accommodation booking and online gaming, and three sites supporting non retailing business-including a national standards association. The remainder included a vintner, magazine subscriptions services and a religious supplies store. Each business conducted direct online sales, although three stores allowed payment only upon the receipt of goods. Many of the businesses required positive customer identification apart from quoted credit card details, as more than 85% required prospective purchasers to register a membership and establish a user name and password before they would allow a purchase request to be processed. All of the retailing businesses had shopping cart facilities, whereas the non-retailing business had single-order facilities only.

In terms of the facilities provided to shoppers, most businesses provided photographs of products in cases where they were relevant. Of those, only 26% provided the option for the shopper to select a larger image, although in some cases larger images were displayed as part of the multi-screen selling process. Notable was the fact that 33% of the computer stores engaged in list retailing without product images, and apparently competed solely on price. Variations were observed in sales closure methods, for example, 10% of businesses did not inform the customer at all of the delivery method. The purchasing systems generally imposed a high workload on customers with considerable detail filling required for ordering, despite the previous registration of personal details, including address, by prospective customers. Little programming effort was expended to reduce customer workload, for example, states and postcodes were not available in pop-up menus, even in purely Australian sites. However, in some 20% of cases there was a tick-box provision for a default delivery address, derived from previously entered data.

Differences in terminology were evident in the purchase completion screens and the number of steps in completion ranged from one to four. Eleven businesses had single step screens, which required the user to scroll the screen during fill-out due to the obscuring of parts of forms. Eight businesses each had two or three screen sales processes, and only four businesses had four screen sales closures. By contrast, 88% (28 of 32) businesses offered the opportunity to pay directly online via credit card. Of the remaining four businesses, two online supermarkets and one computer store required payment on delivery of the goods, with another computer store, apparently sensitive to the potential for fraud, did not accept credit cards but demanded a direct debit payment before initiating the filling of an order (a later visit to the site revealed that the store had closed its doors).

In summary, there was little consensus among the businesses about how to interact with customers apart from the provision of shopping basket and credit card payment facilities. By contrast, and a negative feature, was the workload imposed upon customers by the retailing process. Only two businesses had a final order confirmation screen, only one of which was foreshadowed, somewhat cryptically, more than one screen in advance. The approach to the completion of order requests, far from attempting to "seduce" window-shoppers, and make light of the task, could be characterized as treating the customer more like a warehouse clerk than a wary quarry apt to take fright easily.

Study 4: Online Swimwear Purchase: A Real Case Study

Gammack and Carter (2001) report in detail on the commercial redevelopment of a leading swimwear reseller's Web site ^[3], using commercial data before and after redevelopment, rather than intention to purchase as the dependent variable. The approach taken was aimed at smoothly drawing customers into the purchase process, using color and design features for allure. Only some relevant points are summarized here.

In the original design of the swimwear site, critical support information explaining delivery timeframes, product guarantees, privacy policies and allaying security concerns was noticeably absent, negatively impacting the customer experience. This was directly addressed, using everyday and reassuring language, providing always-visible links to contact and company information. Additionally, three images were provided for every product including a small, medium and large photograph, analogous to the way in which closer examination in apparel shopping occurs in the physical world. Despite studies (e.g., Vividence, 2001) suggesting long checkouts caused abandonment, a secure six stage checkout was implemented that took a purchaser through the buying process, firstly asking the user to select their delivery location and desired delivery method. The shipping costs were then shown and added to the purchase price and exact details of the delivery address requested. On submitting this page, the billing details were asked for. This page was automatically filled out from the information on the previous page but could be changed if delivery and payment details were different. Then on the fourth step, the customer was asked for their credit card information.

We believe that while Internet credit card payments are a concern to purchasers, or particular demographics of these, that waiting until a few stages into the checkout before these details are requested would create more trust and involvement in the purchasing process and reduce the likelihood of credit card purchasing anxiety. This approach parallels the "assumptive close" process frequently used in the personal selling process for complex high value items such as cars (McColl-Kennedy, 2000, p. 606). Asking for credit card details immediately on entering the checkout may increase buying anxieties so much that the shopping cart is abandoned without any further chance of converting the desire to purchase an item into a sale. The experience of being asked for non threatening information first and building up towards the most threatening information may help anxiety levels remain low and as each new web page loads successfully while moving through the checkout process, consumer psychological investment increases and the chances of the shopping cart being abandoned decrease. A summary page and a confirmation message comprised the last two steps.

The previous swimwear Web site had generated approximately $5,000 in sales over 18-24 months. The first week and half that the new site was available, $1,118 in sales was generated from six orders. No orders had been received for the prior two months. (Gammack and Carter, 2001). Table 1 shows the change in sales since the web redesign.

Table 1: Change in Monthly Sales Figures since Web Site was Redesigned

Date	Number of orders	Products sold	Sales value ($)
October 2000	6	13	1118
November 2000	17	33	2130
December 2000	15	32	2279
January 2001	17	31	2503

Hits have increased, and although hits are not a good measure in themselves, in the first four months since the redesign in October 2000, 54 orders totaling AU$8,342 were received. Although mostly from within Australia (41), other countries including the US (8), Singapore (3), the United Kingdom (1), and Saudi Arabia (1) have placed orders, about four times more than previously. This outperforms the expected increase due to a baseline increase in online shopping: in 2001 revenue from online shopping (in the US) grew 40% compared to 2000 revenue (Grant, 2002). Thus, all else being equal, the redesigned Web site may be considered to have been more effective than average during this period.

Beachtown: A Virtual E-Tailing Site

The case study from a real industry example provides several observations that can be tested experimentally. However, it is difficult to apply manipulations to real sites engaged in business, and dependent variables such as quarterly sales figures are typically commercially sensitive data. We have therefore set up a specialized resource to examine empirically some of the psychological factors involved in consumer use of Web sites ^[4] This has all the appearance and functionality of a commercial Web site, but can be tested in online or offline settings, and using real or simulated products as required. We are interested in hedonic products and services, particularly relevant to tourism destinations, since travel and tourism is the major online industry in terms of sales and growth. Recent figures show 52.2 million North Americans use the Internet for travel planning, and 16.5 million book online (TIA, 2000). Tourism researchers have made the point that tourism is "an industry based almost entirely on information, (and) the decision by the consumer to accept or reject the elusive tourism product will increasingly be subject to a compelling description, representation and successful dissemination of data" (Buhalis et al., 1999). Apart from holiday related services, commodity products related to the tourism industry can also be specifically examined within the context of the pseudo-Web site. In addition to the swimwear products described in study 3, surfboards are a product closely related to beach life, and this is our touchstone hedonic product, and our final study will describe our work to date for these categories of items.

As mentioned, a pseudo Web site allows experimental investigation of involvement and purchasing behaviors. Our initial study entailed designing separate Web sites representing the attractions of "Beachtown" ^[5], at varying levels of complexity, and with parameters that can be manipulated to investigate specific effects. Initially these are being researched using convenience samples of surrogate consumers, but the intention is to provide live Web sites and monitor use by "real people", as well as panel samples of relevant demographics. These levels of complexity correlate to increased potential for interactivity with the site.

Although several sites representing the considerable attractions of "Beach-town" exist already we wished to prepare a site with several levels of usability and facilities within which we could empirically test hypotheses concerning user interaction without infringing on the purposes of these organizations, and readily obtain representative copy and source photographs. The usability levels (Figure 2) included a plain text version, then increasing levels of sophistication using frames, graphics and sound, animation and finally virtual reality (VR) and video. Further stages, yet to be implemented, would involve streaming interactive video and 3-D immersion.

Figure 2: Front Page of Beachtown Experimental Site Showing Immersion Levels

Although VR is unlikely to figure largely in current e-commerce or information based sites, its potential as a surrogate for streaming video in low bandwidth situations, and its potential to allow testing of interactivity meant we included it in our design. The technologies are starting to become mainstream though, with for example real estate sites commonly providing realistic walkthroughs of scenes. Some of these future potentials are likely to become realized as interactive television services become established and broadband removes the long download time issues. Meantime psychological work concerning retention of information, navigation, usage patterns and perception can be tested to provide insights into the customer experience.

The online resource includes an interactive map showing selected stop locations available on a free hop-on hop-off bus. Each numbered item is hyperlinked to an information page on the location using this simple interactive map. This corresponds to the fourth level shown in Figure 2, and embraces the three more basic levels. This structure allows for the varying regional access and bandwidth levels extant at the time of design.

Although the pages linked are currently marked up with specific colors and content, we plan to generate these pages from a database, adapting colors and other information presentation details to user preferences and reflecting the site owners' intentions. For example, Beachtown can be marketed as a great place for a romantic evening, with its late opening cafes, cinemas, pubs, clubs and oceanside walks. Pavement cafes for breakfast and people-watching, game machine and safe beaches for children and long shopping hours, all offer other potentials for targeting particular end users, such as tourists and day trippers. Hostels, backpacker events and music venues offer other selections from its potential, which can be shown using other interactive maps. Beachtown has a European feel, a bohemian and new age culture, art galleries, heritage buildings, parklands and specialist shopping malls which all address other niche markets, which may be presented in particular different ways based on commercial considerations. The shared information such as the free bus service around the city, the climate, the location of markets and street theatre, transport interchanges and major restaurants can be used as the site is produced under different criteria. The ways in which the various particular classes of end user engage the potential in the suite of Web sites can be explored in specific contexts, however, our main interest here lies in the engagement with the greater interactivity available through (initially non-immersive) virtual reality.

Non-immersive virtual reality allows realistic three-dimensional images on conventional PC platforms. These virtual worlds, often found in computer gaming environments, afford an experience of smoothly navigating a physical-like space. Navigation may be ground-based or floating, but more sophisticated interactive navigation devices allowing "movement" in several dimensions, require considerable user learning (Boyd, 1995) and entail increased cognitive load, and so we restrict consideration to the analogue of physical walking, using public domain software and a standard PC configuration. Figure 3 shows a screenshot of Beachtown taken from a VRML world modeled on the boat harbor area, one of the hyperlinked locations on the interactive map.

Figure 3: Screenshots from the VRML Version of the Site

Dynamic sites have numerous ways to retain attention and increase engagement, and as more information is gathered about the customers, in particular by learning which aspects they are willing to spend time exploring, more directly relevant information can be targeted. In the experimental setting, questionnaires addressing information retention, and perception of site can be looked at in conjunction with web log information, prior to any live trials with real products as we describe later. With the VR we can enrich their experience in ways that are likely to become more sophisticated in future. For instance, the screenshots in Figure 3 allows virtual wandering by the oceanside through use of the controls of the VRML browser plug in. The two views are rendered as the user moves and it becomes "natural" and easy to imagine sitting under a parasol with a cool drink overlooking the water. Checking the menu and prices of nearby restaurants is immediately possible in ways that textual sites cannot provide. Booking a table can then become a simple decision as the "enticement" towards a purchase decision has happened. Audio of birdcalls and other effects shown to enhance the experience can be used to produce the desired allure.

To date we have tested this by demonstration, observation and solicited professional opinion, prior to more precise hypothesis testing. Gammack and Carter (2001) note that a number of specific hypotheses arise from professional observations made in producing Web sites for clients, which, although capable of systematic testing in an experimental situation, can been prioritized as being the most relevant applied research to do from a commercial viewpoint⁶. Our final study describes work with a representative panel of consumers, using focus group and questionnaire techniques, as preliminaries to further extensive testing. Focus groups, or group depth interviewing, are a common technique in consumer research, (Chisnall, 1991) and are used to obtain qualitative information on attitudes and feelings relevant to products and services, often as a preliminary to wider surveys or experimentation.

Study 5: The Beachtown Focus Group

In this study the Beachtown resource was presented to six panelists, representing typical consumers, with the following demographics, three male and three female Web users, aged between 22 and 46 years, four of who had made previous real life Web purchases. All panelists had a moderate level of enduring involvement with holidays as measured by the Slama and Taschian (1987) enduring involvement scale. The panelists were given a scenario suggesting that they were seeking information on a beachside holiday destination. In a focus group situation they were shown demonstrations of the hypothetical destination Web site at three levels of complexity, basic html with tables and frames, interactive maps with multimedia and virtual reality, lasting about two minutes each. After each display each individual completed a six-item questionnaire concerning intention to buy, memorability, perceived attentional involvement, information effectiveness and the perceived risk of purchase of the product class via the Web. Following this a plenary session was held to highlight feelings, attitudes and issues concerning the use of the site for a potential holiday. Subjects were then told that they had chosen the holiday and qualified for a discount on goods from "BJ's surf shop", which etailed swimwear and surfboards amongst other apparel and accessories. Subjects were then shown the relevant web pages, shown alternative checkout processes of two and six pages and asked a parallel set of questions for these goods, which are also hedonic, high involvement and conducive to more intensive virtual experience. The last scenario invited the group members to purchase a custom-built surfboard at a heavily discounted rate. Two surfboard specification formats were displayed to the group. The first was a text based form, the second was an interactive design specification process which produced an accurate animated visualization of the surfboard design. A focus group was then held, asking particularly on the issue of increased involvement and intention to purchase issues. As a preliminary to controlled testing on individuals, the demonstration procedure adopted denied the subjects direct interactivity, this approach controlled for the amount of time spent on each page, and the navigation paths of consumers, as well as equating the order effects. Subjects were however, given the chance to surf the Web site afterwards voluntarily, and comments were noted.

The panel clearly preferred the interactive map and multimedia display as a source of holiday information, whereas the virtual reality display produced a higher likelihood of visiting the destination. Discussions revealed that the increased likelihood of visiting the destination was borne out of a desire to compare the virtual reality tour with the actual scenery. Although the likelihood of booking the destination accommodation via the WWW was low, it also improved after the virtual reality session. Although the panelists found the virtual reality session stimulating, they strongly expressed the opinion that, for holiday destination information, virtual reality tours based upon panoramic photographs would have more credibility than the animated scenes which were used in the demonstration. In comparative terms the more complex information displays produced higher levels of involvement and attention than the text and frames version of the information. In addition, the richer formats were more memorable. Viewing the BJ's surf shop site aroused some interest in visiting the store during the planned beachside holiday. However, the two-step Web sales approach was preferred for its speed rather than the six step procedure for all but one panelist. The group member who preferred the six-step process was a regular buyer of PC and technical equipment via the WWW and consequently preferred the purchasing system that afforded detailed checking of item specifications. In the last stage of the focus group the members' likelihood of purchasing the heavily discounted custom-built surfboard was assessed after exposure to the two specification formats. When compared to the Web based text form, the animated visualization method of designing the surfboard produced very significant increases in the likelihood of purchase.

In summary, the results of the focus group study support the proposition that more complex information environments increase attention and involvement. In addition, the group discussion revealed a willingness to use the WWW to research holiday destinations and attractions online, but there was reluctance to book online due to risk perceptions. These findings match those of the state tourism authority (Ward, 2002) in relation to their national and international Web sites. However, online booking may become more popular if consumer risk perceptions can be lowered, perhaps by providing the virtual reality tours based upon panoramic photographs as suggested by the focus group. In relation to Web purchasing procedures, it is possible that consumers prefer more stages and a confirmation screen if the product or service is more complex. Conversely, relatively straightforward products with limited specifications may be more suited to brief purchasing processes, emulating more closely the almost instant gratification of physical shopping. By contrast, sales of highly customizable products, like the surfboard used in this example, may be facilitated by animated visualization systems. The study suggests that the patronage of features like virtual reality displays "...will not be driven by an infatuation with the technology, but rather by the development and provision of information and purchasing services that offer real benefits to the consumer..." (Hodkinson, 2001). These observations suggest that extensive research will be required to determine appropriate product display modes and sales procedures in order to stimulate Web purchase.

^[1]We thank Brendan Neil of http://www.surfyn.com for conducting this study.

^[2]Upon inquiry it emerged that one respondent had actually bought a car worth $40,000 online. This was a special edition vehicle available from a reputable overseas outlet and such high value hedonic purchases are not atypical. Indeed, some Ducati motorbikes are only available via the web.

^[3]See WAEcC (2000). Full report available from the Centre for Electronic Commerce and Internet Studies, Murdoch University, Perth, Australia 6150.

^[4]Credit for much of the code base is due to a Murdoch University student project group, Neutral Multimedia (Alvin Chan, John Darrington, Sean Grosse, Luie Matthee, Brenton McArthur, Luke Thin and Tony Yiu). Vickie Williams of Vickie Williams Computing also provided the original prototype.

^[5]Kotler et al. (2001) have suggested that Brisbane rename itself as "Beachtown", an eventuality considered unlikely to occur. For experimental purposes however this name was adopted due to its generality in an Australian context, and Beachtown is modeled on a composite of actual locations.

---