e-Commerce and Screen Design:

What aspects of Web design truly make a difference in an online store when trying to either increase the number of visitors or increase sales?

Lohse and Spiller (1998) conducted a study to determine the relationship between Web site screen design and e-commerce. More precisely, they tried to predict store traffic (visits per month) and sales (dollar sales per month) as a function of interface design features. Their goal was to facilitate the process of designing and evaluating alternative online stores by identifying those features with the greatest impact on traffic and sales.

The study focused on super stores, which are Web-based stores with a large selection of products, such as L.L. Bean, Land's End and Spiegel. They collected information from 28 different stores.

The authors evaluated 32 interface features using stepwise regression. Nineteen of the 32 features had no impact on predicting either store traffic or sales. These included virtually all of the traditional display-oriented screen design issues. Among the features found to have no impact were:

• choice of background colors or patterns,
• availability of help screens,
• consistency of selecting new pages from the top, side or bottom,
• homogeneity of product listings,
• the number of buttons on the home page,
• use of icon-only, icon plus text or text-only for buttons, and
• the depth of navigation (number of clicks) to an item.

The remaining 13 were identified as predictor variables. The variables in the traffic model explain about 89% of the variance, and the variables in the sales model explain about 87% of the variance. Most of these variables had little predictive power, but a few were very strong.

The two variables that had the strongest predictive value for increasing store traffic were:

• having a well-designed FAQ (explained about 45% of all variance), and
• having a large numbers of products (explained about 17% of all variance).

Each additional product in the store seemed to yield additional store traffic. In both of the above cases, however, the additional traffic did not result in higher sales.

The variable that had the strongest predictive value with sales was how the products were listed (61% of the variance in monthly sales). In fact, any improvement over the basic list box yielded increased sales. The best lists of items were those with added text descriptions, graphics and navigation buttons to additional material.

The authors concluded that a user interface that facilitates browsing product lists is more important for generating sales than a fancier storefront. This means that, "no amount of 'sparkle' in the presentation of information could overcome a website designed with poor navigation features."

e-Commerce Pricing

When selling from an e-commerce site, is it better to advertise everyday low prices or sale prices?

Tom and Ruiz (1997) conducted a survey and found that the choice of pricing strategy is one of the most important decisions a store must make. They asked study participants several questions such as "should a store advertise watches at an everyday low price ($150), or a sale price (regularly $200, but on sale for $150)?"

They reported that 76% prefer sale prices and only 24% prefer everyday low prices. About 66% of their participants indicated that they felt they were paying too much with "everyday low prices." Overall, most participants believed that they were saving more with sale prices, regardless of the store and product.