Graphic Design on the Web

A great deal of attention is paid to the visual appearance of Web sites. However, it appears to have no impact on usability: the team "measured every aspect of graphic design we could think of [and] found no evidence that graphic design helps users retrieve information on a Web site."

The Role of Graphic Design

Conventional wisdom dictates that a more "graphical" site will be more interesting to users. Therefore, they will spend more time on the site; and hence, it will become more familiar and usable to them. The research they conducted indicated this theory "has no basis in reality."

The site that performed the best, by a wide margin, was a plain-text site with virtually no graphics. If graphic design had any impact, they would have expected this site to finish dead last.

The only correlation they could find to usability was link color: the default for most browsers is blue text for a link, purple for a visited link (both underscored). Sites that deviated from those colors were harder to use; sites that removed the underscoring more so.

EN: At the time the study was conducted, many sites did not deviate from the blue/purple default colors and underscoring for links. In the present day, many sites have implemented their own custom colors, and users have become accustomed to it. Except in instances where a site is internally inconsistent, or there is insufficient differentiation in color between text. links, and visited links, users don't have a problem figuring it out.

Download Time

Download speed is often touted as the chief complaint users have about the Web in general. However, users did not seem to be impeded or frustrated by download speed on the sites that they tested, which varied from plain-text to highly graphical interfaces.

EN: Bandwidth has increased dramatically since the time the study was conducted, so I tend to doubt that findings pertaining to load time remain valid.

Most users followed the same pattern during load time: they skimmed the text and image ALT text. If they found something of interest, they clicked away before the page finished loading. This indicates that most graphics are not helpful to completing a task.

There was, however, a problem when images without ALT tags were used as data labels. Users jumped to conclusions, and often provided an incorrect answer as a result of not having all the data labels available.

Users did express frustration at the use of graphics that did not add to the content, particularly when they waited for the images to load. However, there was not as much frustration when the images contributed to their understanding of page content.

Animation and Movement

Some of the sites tested used animation, but the subjects found it to be gratuitous: none of it contributed to understanding the information on the screen.

However, users find animation distracting - some would scroll it off screen in order to be able to focus on the content of a page. Others would actually cover animation with their hand in order to read the content without distraction.

Advertisements generally contained the most movement, presumably in order to attract user attention - but it achieved the opposite effect: users completely ignored it. Given the task of finding the lowest fare to a given destination on a site where there was an animated promotion touting exactly that (best fare to X), not one user followed the link provided by the promo.

The implication is that a user who is on-task will ignore animation, though it might still be useful to attract the information of someone who is browsing the site leisurely, without a specific purpose in mind.

EN: At the time the book was written, animation was fairly new to the Web, and was most often used as a novelty or a distraction. I'm disappointed that Spool's findings have been used to support the conclusion that all animation is bad: if used properly, an animation can be more illustrative than text, or facilitate the understanding of content. It would be more accurate to conclude that gratuitous animation on a page where the primary content is contained within the text is (obviously) bad.