Text vs. Graphical Links in an E-Commerce Web Site

Introduction

For many years following the birth of the Internet, textual links were used for traversal of a website. Users would click a word or a phrase to take them to another page on the site's map. Soon web page designers got cute and began using graphics like pictures, drawings, or word art to represent these links. Some designers opted to stick with the textual links, while some switched their sites over to the graphical links. Others used a combination of graphics and text links for navigation of their site map.

Recently the web has become a booming business venture as an increasing number of businesses now offer their products online. E-commerce, as this advent of online business is termed today, represents a large portion of cyberspace and new e-commerce sites are popping up daily.

The links of a website are crucial to the usability of the site. Websites must provide links that are not misleading and that will create a comfortable atmosphere for navigation through their site. An e-commerce site must provide not only a direct, understandable path to their products but they also must provide a quick way for users to find the product they are looking for.

Currently sites such as www.sears.com, www.dealtime.com, and www.circuitcity.com have similar styles of implementing an online store. For example sears.com has selected to only have a text menu. However if the user hovers his mouse above a selection, a picture of the merchandise is shown. When you select a specific type of merchandise, the user is given a list of different types of styles. When a style is chosen the user is then given a list of more criteria that he has to fill out, such as price range and size, before finally selecting get list, which returns a list of products that fit the specified qualifications. This list contains a description of the products, no picture is provided unless the user selects a specific item. This method of displaying the merchandise does a good job of showing the user the specific item that he or she is looking for, however this approach is very poorly designed for someone just browsing for merchandise.

Like www.sear.com, www.circuitcity.com has chosen to have a text menu. If a user selects a type of merchandise, he is given the stores current special of that type of merchandise and then given the option to fill out a form, like www.sears.com's, to get more products. The list of merchandise returned includes a very small picture of the product along with a description. This method of displaying the merchandise the store has to offer again is well designed. However the addition of a picture with the list of merchandise helps the user just browsing for merchandise.

In addition to www.sears.com and www.circuitcity.com, www.dealtime.com again gives the user only a text menu. The user has to navigate through the text menus until, he or she gets a form, very similar to both circuitcity.com's and sears.com's forms. Once a user provides his or her personal specifications, a list is provided of all available products fitting these specifications. This list does not contain any pictures of the merchandise. Once the user selects a specific item, then a photo and a description are provided.

These three sites give a good example of a typical online store. Most sites provide textual links up to a form, which allows users to specify their personal preferences. Once the user fills out the form, a list of available products is returned. On current sites the use of graphics is not that common, only for the final page in which the item is displayed by itself along with a description of its features.

There are several reasons why these sites do not use graphical links or icons throughout most of their site. It is very difficult in some cases to produce a meaningful visual representation of an idea or concept that is to lead the user down the correct path on their site. Although some say a picture is worth a thousand words, sometimes a simple word or phrase is less misleading as a link in the traversal of a website. Several studies examine the use of icons or graphics instead of text. It is no easy task to create a graphical link that is meaningful and not misleading.

In 1989, Stephanie Houde and Gitta Salomon of the ATG Human Interface Group at Apple Computer, Inc. conducted a study aimed at identifying the visual cues that help people recognize real world items. During the study, if a subject was in search of an item, the subject recorded the thought process, with a tape recorder, that occurred while searching for the item. The five subjects described what they recalled about item, how item was searched for and what notified them when they had found the item. Houde and Solomon discovered that the subjects often failed to accurately recall the characteristics of items they sought. Once a desired item was found, subjects recognized such details as specific words, colors, typography or images. Also, subjects often used physical landmarks to mark the relative location of an item.

With these results in hand, Houde and Salomon examined applications for browsing file systems. File listings, until recent years, were text-based lists. The authors suggest that "richer file representations, that facilitate recognition based on varying criteria, may aid in this browsing task." These results suggest that navigators of an e-commerce site may possibly be able to find the product they are looking for through visual cues that remind them of the desired product rather than just text-based links. Online shoppers could possibly be searching based on the physical traits of a product, i.e. size, color, shape, etc. In addition to the aid of visual cues, the authors found some possible drawbacks to a single representation of an item. The authors found that "users representational needs changed with their current context and personal idiosyncrasies." This issue may hinder graphical links on a website, as there may be varying mental representations among online shoppers for certain products. For instance, shoppers may have different mental images of a microwave or a vacuum cleaner. Our experiment will hopefully provide insight into whether the author's results on the process of finding an object in the physical world can be extended to finding a product in an e-commerce website.

Lodding (1983) work on icons may also provide insight into the issue of textual versus graphical links in a website. Lodding finds icons to be practical based on the powerful image memory and processing capabilities of the human mind coupled with the fact that people find images "natural." Lodding found that icons can be easily learned and recognized as well as possess "more universality that text." As a result of these characteristics, Lodding believes that iconic interfaces can "reduce the learning curve in both time and effort, and facilitate user performance while reducing errors." As the World Wide Web reaches every corner of the world, companies may be forced to internationalize the company's website. Using graphics as links on a website eliminates the need to translate text based links.

Manes (1985), on the other hand, asserts that icons may be confusing, wasteful of space, and totally ineffective in dealing with large numbers of similar commands, files, or concepts. In the case of textual links versus graphical links, graphical links have the potential to consume more screen space than textual links. Also, ambiguity between the images of the same type of product manufactured by two different companies could possibly cause confusion.

Gittens (1986) sees icons as problematic, as it can be difficult to find "obvious pictographic equivalents" of computer system concepts. In the case of e-commerce pictographic equivalents of products consumers can buy are easy to find. We could simply use images of the products themselves and even Gittens would agree these pictures are "obvious pictographic equivalents" of the products. The question remains though: Will the graphical links, even if they are pictures of the products themselves be more effective than textual links or would a combination be most effective?

In order to answer this question we have to first discover the present state of this issue and how the work of people such as Houde and Salamon, Lodding, Manes, and Gittens has affected websites today. After examine existing sites like www.sears.com, www.circuity.com and www.dealtime.com that we discussed you see some examples of what sites are doing, but we also must know the effectiveness of these sites and others on present day Internet shoppers.

A study reported in the E-commerce Times evaluated ten leading online stores and showed that 39% of test shoppers failed in their attempt to purchase an item because the sites were too difficult to navigate. These findings are not surprising considering the issues we have been discussing. The study revealed that quick, simple, and clear sites would enjoy the most business.

Too many times developers will rush towards developing the most attractive site with the latest technology, while overlooking some of the most important aspects of the development. In developing such a site the developer must concentrate most on the content of the site and how the visitors will be able to navigate through it all. There are many things that one must consider in using graphics to represent a link. First an icon can complement text to clearly define a label. Graphics also attract attention from the user and make the site more attractive. They can also enhance borders and better visualize "clickable" zones.

On the other hand, too many graphics can overwhelm a user, and cause navigation problems. Graphics without texts are usually cryptic, and lose their function. Graphics without function are generally irritating and distract the user from the actual content.

These general guidelines for web site development may be found easily on the Internet, and have become acceptable design principles. In our study, we hope to show that the addition of graphics to text will add to the clarity of labels, as well as facilitate the navigation of a web site.

In 1998, Gerald L. Lohse studied the effect of interface design features on on-line store traffic and dollar sales. Lohse examined such aspects as the number of links into the store, image sizes, number of products and store navigation features. From his results, Lohse suggests that browsing product lists can increase the sales of items in the website. Lohse found that "the number of levels between home page and end product pages has no significant effect on visits and sales." Also, Lohse found that storefront variables such as image size, background patterns or the number of buttons on the storefront screen had no effect on store traffic and dollar sales. Lohse does note that "additional product list information such as price, a thumbnail image, and a longer decriptive product name had the largest impact on sales."

The results from Lohse's study raise interesting points and it shall be interesting to see if our results coincide or deviate with his. Lohses's findings on the number of levels between home page and end product pages should help validate our study, since we will only be designing sites with 4-5 levels between the store home page and the products pages. Typical E-Commerce web sites can have many more than 4-5 levels. Results of our study on a relatively "small" website may possibly be extendible to larger E-Commerce sites.

Our study will set up three copies of a mock e-commerce site, each with a different type of links used. One will use only textual links, another will use purely graphical links and the last will user a combination of text and graphics for the links. We will have typical consumers search for products using each type of link. Through the data collected from this experiment we hope to show that a combination of text and graphics for a link would be beneficial in certain situations. In addition we hope that this study will provide a basis for future studies, expanding this issue to discover an ultimate solution to this problem.

Department of Computer Science: Direct questions and comments to the student editorial team