Fifth International World Wide Web ConferenceMay 6-10, 1996, Paris, France
Fifth International World Wide Web Conference
ZAXOR Software, a computer system integrator and software developer
in the Washington DC metropolitan area, conducted a research project to
determine the impact of varying website development techniques on the actual
purchasing behavior of online shoppers. To conduct this study, ZAXOR invited
five groups of fifteen people each, six male and nine female, to use sample
"shopping mall" websites. Each of the groups' sample websites
had one variable development technique changed. The groups purchasing behavior
was recorded under similar conditions, and post-test interviews were conducted
with the participants about their shopping experience.
The ZAXOR study took into account: a) the relationship between purchasing decisions and payment mechanisms, b)the relationship between placement of objects on a website layout, c) the effect of special discount pricing d) the impact of entertaining, non-purchase related activities on a website e) the impact of animation and VRML on purchasing behavior. Results of the study showed that setting up payment accounts in advance of visiting the website had the biggest impact on purchasing behavior of all other techniques. Test subjects spent two and a half times more money in the same amount of time when their payment account logistics were separated from their purchasing decisions.
Special discount pricing had the effect of making items that were not discounted seem undesirable. In effect, special discounts seemed to make online purchasing associated with bargain pricing and lowered overall sales dollar volume.
The impact of animation had the next highest positive impact on dollars spent per person (directly after the impact of payment logistics). Objects which moved drew attention and received the most order volume. Next highest were objects directly to the right and below objects which moved. Objects directly above moving objects were adversely impacted.
Entertaining activities increased the time spent in the shopping mall, but did not have as much impact on dollars spent per person as the introduction of the VRML world, where products could be met along the way while engaged in an entertaining social activity.
The ZAXOR study concluded that the most effective commercial online shopping experience would model the characteristics of a multimedia sensory chat room, rather than a print-media catalog.
1.1 ZAXOR Software, a network document management system integrator, had a number of clients interested in setting up websites for their consumer retail businesses. In newspapers and trade journals, however, these clients had read that "consumers are not buying online." In addition, these clients were bombarded with local newspaper ads offering "home pages" for $45 set up and $5 a month. This made ZAXOR's price quotes appear "too high". ZAXOR needed a way to exhibit to potential customers the commercial benefit to a retail store environment of a well-designed and effective website. It was also important to quantify the impact on consumer purchasing behavior of types of websites. We decided to conduct a study in the retail store environment to observe typical consumers' reactions to online shopping techniques. Our purpose was to quantify the characteristics of a website that would trigger consumers to purchase online.
2.1 The study was conducted from November 12, 1995, to January 15, 1996, in Rockville, Maryland, at the Presents with Presence gift shop. One hundred and two respondents completed the initial pre-test survey, from which seventy-five people then participated in an interactive test of multiple retail website types in ZAXOR's offices. Participants were offered their choice of a free personal home page on our server for one year or a free copy of The ZORB personal document management software, as an incentive to participate. Participants were chosen based on their computer literacy. We wished to have a sample of the population which would normally visit that store: thus, the choice to survey by questionnaire within the store. Within that population, we then chose to ask for further participation from those who had personal computers in their homes and were currently using a mouse. Our choice here was meant to eliminate "mouse aversion" from the test results. We did not require participants to have previous experience using the Internet.
2.2 Our inhouse server was the Netscape Beta Commerce Server running on an Intel 486/100 with 32 MB RAM in Windows NT3.5. We tested with a Netscape 2.0 beta browser and a VRScout VRML plugin. For the test, we set up five workstations with local cached files of the type we wanted to test. The workstations ran on a Novell network to the local server. No aspect of the test directly accessed a remote site -- we wished to eliminate any source of aggravation resulting from factors outside our control. We wanted each workstation to have the same set of experiences during their testing time. The actual effect of running the test locally also reduced the time delay for loading new pages, which we felt would be a significant objection for consumers with their typical 14.4k modems. The impact of running everything locally was similar to the effect online services such as America Online achieve by downloading the files in a batch before the consumer begins to work with that group.
2.3 Each of the five workstations was loaded with a shopping mall website representing a fictitious gift shop similar to Presents with Presence, a fictitious clothing store selling casual men's and women's apparel, a fictitious electronics store selling upscale electronic "gadgets" like massagers, vibrating chairs, talking alarm clocks, designer power tools, etc., a furniture store, and a gourmet coffee shop. Each store had thirty items for sale. None of the stores had familiar names. (We were looking for impacts due to the site characteristics, not the store's reputation.)
2.4 We wished to test for an impact on buying behavior due to the following
sets of factors:
2.5 Our theory was that these factors, commonly expected to have an
impact on sales, would be effective cumulatively, rather than individually.
That is, the question being asked is not "Does aesthetic layout increase
sales more than specials and promotion s increase sales?"; rather
the questions are cumulative: Given an aesthetic layout, does a particular
payment method increase sales? And; given an aesthetic layout and an effective
payment method, do "specials and promotions" increase sales?
And, given an aesthetic layout, effective payment method, and "specials
and promotions", does an interactive entertainment activity increase
sales? On this theory, we needed only to establish the baseline by measuring
the purchasing incidence at an aesthetically pleasing site, and then add
each of the next characteristics cumulatively, measuring the new purchasing
incidence each time to measure the impact of that additional characteristic.
We chose to illustrate five different types of stores in our test in order to provide a diverse enough offering so that the participants could be fairly sure of finding something they wanted and their decisions would be oriented to "do I feel the urge to buy SOMETHING" rather than "do I specifically want THAT item." We wished to focus our testing on triggers to purchase online, rather than on triggers to purchase a specific item. We also did not attempt to test what makes people visit a particular website in the first place. Our study assumed the traffic volume, but did not consider how to generate it. Our study question limited itself to: "After consumers arrive, what triggers them to purchase ANYTHING online?"
2.6 Our study methodology was simple. We built a basic website consisting of our shopping mall. Using the Frames methodology in the Netscape Beta 2.0, we made the site clear and easy to navigate. We made the store templates and structures identical. We made moving from store to store, browsing through the product offerings, and blowing up the pictures of items to full-screen jpegs, simple, one-click operations. We paid attention to the picture quality, aesthetic background textures and colors, convenient linking options, and sales-oriented word descriptions on each page. We made the basic baseline website a high-quality product we could be proud to use as a reference site for our potential retail store customers. On this basic website, each store offered potential buyers a choice of typing in their credit card numbers to a secure Netscape server or of calling an 800 number with their credit card for ordering.
3.1 Each tester was a shopper in the Presents with Presence store. Each had a personal computer in the home and was able to use a mouse. Of the 75 testers, 30 were men and 45 were women. They all lived in the Washington DC metropolitan area, no more than 15 miles from the store. Only 15 of the testers had a personal Internet account at home, although 22 testers had used an Internet account at their office. Their average age was 42, and their average household income was between $60,000 and $75,000. None of the testers had ever purchased online before. The testers were divided into five groups. Each group was balanced to have the same number of men and women, and approximately the same mix of ages. Income levels across the group were fairly homogeneous
We then called in our first group of testers. Each tester's workstation was enclosed in a cubicle with a telephone. They were not able to see other testers, so they would not be influenced by other's behavior. Each person was asked to remove their watch and place it in a locked drawer in their cubicle. There was no clock visible from the cubicles. We trained each tester to use the system, then told them we wanted them to shop in our online shopping mall. We asked that they simply do what they would do if this machine were in their own homes, and make buying decisions the same way they would if they were home right now using their own money, and within their own budgets. We told them to go ahead and use the telephone and call the 800 number if they wanted to buy, but that of course, the 800 numbers all simply went to our own offices, and they would not really be placing an order. We gave them "fake" credit card numbers so they would not feel uneasy about their real credit card numbers. We asked that they do their best to act the same way they would act if this were their own home and they were shopping with their own money. We played upbeat jazz in the background and burned cinnamon- scented incense to simulate the "feel" of a mall shopping experience. The testers were told to take as much time as they wanted to shop.
We recorded their purchasing decisions and the amount of time they spent shopping to use as our baseline. This group of testers was then released and not used again.
Next, we added to our website, a new payment method. In this payment method, basic information about you and your payment preference was recorded shortly after you entered the shopping mall. The option to pay by check or bank draft was added. An account number for you was set up so that you would not have to re-enter this information the next time you returned to shop. Most importantly, however, you could not shop without setting up an account. Once you set up an account, you would not have to enter information separately at each store, and your purchases could be completed by pushing one button.
The second group of testers was invited in, and instructed identically to the first group.
Statistics on how much time each participant took, whether they used the centralized account, how they chose to pay, and their purchasing decisions were collected and the second group was released.
Next we added an area of special offers and promotional discounts to the front of the shopping mall. The third group of testers was invited in and instructed as before. This time, however, they were told an account had already been opened in their name. There would be no need to enter any personal information, as this had already been done when their account was opened. They only needed to enter their account number and password to enter the mall, after which they could add purchases to their shopping cart throughout the mall with no further personal identification.
The third group was monitored for time spent and purchases made, and released.
Now we added to everything else already in place, an area at the front of the mall, directly inside, which offered an interactive entertainment area. Throughout the entertainment area were placed advertisements to various products and stores. Clicking on the advertisements took the shopper to the store, where only the products which had been advertised were displayed as 3D animations, rotating 360 degrees. Blinking underneath the animation was a colorful message saying "Buy me".
The fourth group was given no time limit. They were told their accounts were already in place, just as the third group was. The fourth group's time and purchasing decisions were logged and they were released.
For the final version of the website, the interactive entertainment area was replaced by a VRML world. The VRML world led through the various stores, transporting back to the 2-dimensional store at various points, and back again into the VRML world, in a winding, somewhat disorienting Alice-in-Wonderland version of a shopping mall. Along the path, the shopper would encounter billboards representing the advertised products. Clicking on them, he or she is transported to the 2-dimensional store, with the animations as before. Once in the 2-dimensional world, the consumer has the option NOT to return to the VRML world. He or she can, however, transport back and forth between the worlds at will. Because the entire system was running locally, there was little problem with time delays for loading files when transporting.
The fifth group was told their accounts were already in place, as before. They were given no time limit. They were, however, offered donuts and sodas in the conference room and told they could take a break anytime and help themselves. The concept that they could shop as long as they wanted was reinforced.
3.7 After each set of tests, the testers were asked to complete an "exit interview" about their experience. Results of the exit interviews are summarized in Charts 4 through 7. The charts summarize answers by each of the five groups to the questions: a) Advantages to Online Shopping , Disadvantages to Online Shopping, Shoppers Estimate of Time in the Mall, and Shoppers Satisfaction with Online Shopping
.
4.1 The chart here Average Dollars Per Person by Group illustrates the average number of purchasing dollars spent per person by each of the five groups. The next chart shows the Average Time Spent in the Mall Per Person by Group Number. The third chart shows the Purchases Made by Product Category for each of the five groups.
4.2 The following Exit Interview was conducted with each of the participants after they had completed their shopping test.
4.3 The basic website, Group One, averaged $25 per person. In the exit interview, many respondents cited "convenience" as a negative reason why they would not be likely to spend more time using online shopping. They also expressed concern about typing in their credit card number in real life, even though they were assured the line would be secure. They were less concerned about giving their credit card number over the telephone, although observation of the behavior showed that few shoppers in Group One purchased more than one item during their session. Of the seventy-five participants, eleven made no purchases at all. Forty-six of seventy-five participants in Group One said they "doubted whether they would shop online." Security and convenience were the top reasons cited as negatives to online shopping. "Lack of selection" was the next most frequent negative reason, followed by "concern about quality of merchandise." Participants also said online shopping was "not interesting."
4.4 Group Two respondents, who were introduced to the account payment method, averaging $62 per person, however, had only nominal objections to "convenience" and "security." They, too, had concern over "lack of selection" and "quality of merchandise." No one in the second group bought nothing during their shopping test, and sixty of the seventy-five participants bought more than one item. Group Two respondents were less negative about the probability of shopping online from their own homes -- responding that they "probably would consider it at some time."
4.5 Group Three respondents, who had the option of selecting "specials and promotions," dropped their spending levels per person from the Group Two level to $48 per person. Seven of the respondents made no purchases. Overall their responses were similar to the Group Two responses on the exit interview, with the exception that they felt more positive about the possibility they would shop online, responding that "maybe" they would do so. The biggest advantage they saw to shopping online was "better prices."
4.6 Group Four respondents, who had an interactive experience combined with animations and sound, showed a tremendous increase in purchasing dollars per person, to $87. They also more than doubled the time spent in the mall, from about 45 minutes for the previous groups to slightly under two hours. They showed less concern over lack of selection and quality of merchandise, and were not particularly bothered by security and convenience issues. They answered "probably" to the question of whether they would shop online. Most interestingly, they said the biggest advantage they could see to shopping online would be "to save time"-- an interesting answer for someone who just spent two hours shopping online.
4.7 Group Five respondents, who entered and re-entered a VRML world, purchased over $103 in merchandise and stayed in the mall nearly two and a half hours. They answered "yes" or "probably" to shopping online at home. Their only significantly "negative" responses were "not enough stores" and "can't socialize with other people." Quality and selection seemed to be of no concern in their minds, and one third of the respondents listed "quality" as a reason to buy online! Sixty-two of the seventy-five participants requested to be placed on a notification list when the ZAXOR mall opened. Most dramatic, however, was the fact that the participants consistently underestimated the amount of time they had spent shopping by more than 50%. Group Five respondents also thought "to save time" was a good reason to shop online, but they marked "more relaxing than driving to a store" and "don't have to get dressed up" as roughly equivalent advantages.
4.8 The chart here illustrates the amount of time spent shopping by each of the five groups.
4.9 The chart here illustrates the products purchased by each of the five groups.
The reasons most frequently given for shopping online, "to save time" and "convenience," in themselves do not appear to be strong enough to overcome the negatives of online shopping, most frequently cited as "lack of selection," "concern about quality," and "concern about security." In the ZAXOR study, however, these negative concerns appeared to be in reality simply perceptions arising from the two-dimensional nature and isolation of the mall presentation. When the three-dimensional shopping worlds were introduced to a similarly constituted audience, they had a major impact on the perceptions of quality, selection, and security. This impact made "convenience" and "time-saving" to be considered a more important characteristic of the cybermall -- even though the shoppers spent considerably more time in the mall! This time spent in the mall was not actually perceived to be time spent shopping. Instead, it was considered a leisure activity: thus, the perception that online shopping was "saving time" and allowing more time for relaxation. The VRML world experience was enhancing the shopping experience, offering some replacement for the entertainment and relaxation perceived to be in a real life shopping mall. A multi-user VRML world could potentially offer the diversity and change, as well as the social interaction, of a real life shopping mall.
A key breakthrough in our study was the impact of an account setup in advance, rather than an individual account information collection at the time of purchase. Shoppers need only identify themselves by password when they enter the mall. Purchase decisions are then dissociated from physical payment mechanisms, in the same way credit cards offer this dissociation in real life.
The key factors we found that enhance the perception of satisfaction for a shopper in a cybermall were:
The use of the World Wide Web to showcase consumer products for purchase online offers many advantages for a retailer in today's global economy. Retailers who find their own geographic area in an economic downturn may be able to find customers who are enjoying a more prosperous economy in their geographic area. Potential selling time expands to 24 hours in a day and 7 days in a week. Floor space has no physical limit. Storefront rents go down. Fewer employees are required to handle the same sales volume. Using sophisticated chat vehicles, a customer service department can be setup that rivals the efficiency of the real thing. In-store inventory levels can be low and shoplifting problems nonexistent.
For the consumer, the depth and breadth of product offerings available is no longer limited to a fifteen mile radius of his home. He can choose products from the comfort of his den, at a time that suits his schedule. While his option to choose the social environment of a real-life shopping mall remains, he adds the option to purchase without leaving home. Ideally, his cybermall experience offers an alternative social shopping experience in the form of a multi-user VRML world, with products intertwined seductively, for his convenience and pleasure. We see the successful cybermall more closely emulating the successful 3D multimedia chat room than the LLBean catalog.
Our observations lead us to conclude that online consumer purchasing is a multi-sensory prospect, dramatically different in response than a print catalog. There is much more to observe. We need to understand the impact of a time constraint on buyer behavior online. We need to observe responses to varying product placements and combinations within the mall or the stores. We need to question whether the use of stores and malls is the right analog -- perhaps products in cyberspace aren't required to exist inside buildings. Perhaps online shoppers would be more enticed if they were to meet products independently along the road. Perhaps the products should show themselves as additions to the shoppers own "house," or sit themselves inside his "refrigerator," or appear as clothing on his avatar.
We need to study the impact of repetition of a product, compared to diversity of product line. We need to consider whether the traditional techniques of market segmentation have new categories and dimensions in a cyberworld. Do MUD afficionados respond differently to a particular product presentation than AOL subscribers do? Do these differences transcend the traditional segments of sex, age, profession, and geography? Should we market to technical literacy and politics rather than gender and income level?
Many questions lie ahead to be answered. To answer those questions, retailers and the service providers who build their cyberworlds will enter a dimension previously unexplored by retailers -- that is, the dimension of emerging technologies, where market research requires direct feedback and hands-on studies. In an emerging technology, we can't use a telemarketing company to ask consumers what they want. Without direct hands-on experience, consumers can't answer the question. Confused, they say they are dissatisfied with online shopping because of "lack of selection", "concern over quality," and "worry about security." Our study showed these concerns to dissolve when multiple sensory input familiarized the environment. All of these responses are simple misinterpretations of the feelings of fear generated for many millions of people by new technology. Until today's computer literate children become adult consumers, this feeling will dominate consumer online behavior.
As leaders in the implementation of this evolving technology of cyberspace, we share a vision of the tremendous potential to enhance the global economy and enrich the human condition that is embodied in the concept of this vast communications matrix. Delivery of goods and services to a worldwide population is part of that vision. We must take care in building our structures, remembering that the characteristics of human behavior will ultimately define success or failure of our earnest work.
Alice Richmond , Executive Vice President of ZAXOR Software,
has been in computing since 1968. A graduate of the University of Pennsylvania's
Moore School of Electrical Engineering, she began her career as an assembly
language programmer at Leeds Northrup R&D, and free-lanced her programming
skills to pay her way through engineering school. Chosen by the Bell System
as one of its first management trainees, she headed the Advanced Product
Development Group in Bell of Pennsylvania, before leaving to attend the
Harvard Business School. After graduation from HBS, Ms. Richmond held executive
positions at Computer Sciences Corporation and Unisys before founding ZAXOR
Software in 1992.
