Criteria for optimal web design (designing for usability)

By Michael L Bernard

For online shoppers, advertisements tend to be the most frustrating factor in using the Web.  According to Retail Forward (2002), the top five online shopping frustrations are:

 
General Web user surveys have found that  most dissatisfying web experiences are, a) not being able to find specific information, b) using websites that are confusing, and c) websites with slow download time, respectively (10th GVU survey, 1998). The fist two major annoyances are addressed in the navigation and the positioning of information discussion, respectively.

The third annoyance, slow downloads, are a very common complaint, which have been known to negatively affect user satisfaction, and ultimately sales.  In fact, it has been estimated that as much as $4.35 billion in e-commerce sales have been lost each year due to user frustration related to slow downloads (Zona Research, 1999). In fact, ZDNet reported that a survey of 12,000 online customers found that 48% of them gave up trying to purchase an item online because the web pages took too long to load (ZDNet, 2000).  Moreover, it has been found that slower web pages were significantly judged as being less interesting than their faster counterparts (Ramsay, 1998), and were thought to have lower quality products, as well as having compromised security (Bouch, et al., 2000).

Currently, the average connection speed is approximately 5Kbps (kilobytes per second). Thus, a 40 Kb web page will take approximately eight seconds to download. This just fits within the so-called 'eight second rule', which is considered the recommended loading time for web pages. That is, it has been suggested that users will tolerate no more than around eight seconds for a web page to download. 

There are some studies which back up this suggestion. For example, a study by Bouch et al. (2000) found that participants' average load-time tolerance was 8.57 seconds. However, the standard deviation was 5.9 seconds. Another study by Zona Research (1999), which examined a website with a homepage of 40 Kb, had a bailout rate of 30%, while other pages with 32-35 Kb ranges in the same site had bailout rates of 6% to 8%. Reducing the load time for the home page to 34 Kb decreased bailout rate to the same range as the other pages.

According to Dillart and Kahn (1999), the factors that determine users' frustration with a downloading a website are:

• How long they have to wait
    
  This is obviously a very important fact. Moreover, the frustration associated with waiting tends to be highest when there is a lot of  gratuitous graphics. Placing images that do not add to the site will decrease rather than increase user satisfaction with the site. However, there is no hard-fast rule about download speed and annoyance. Just factoring in the time it takes to load a web page does not take into account other variables that determine users' frustration with the site.
    

• Whether the consumer experiences uncertainty about the wait
    
  Not knowing how long to wait for a page to load is often very frustrating to users. In fact, Hui and Tse (1996) found that users reported negative effects on mood if they had no information concerning how long they have to wait, which affected their overall service evaluation. Thus, for longer waits it is recommended to inform the user how long the wait should generally take. 
    

• The kind of information that is provided about the wait
    
  Providing adequate information about their expected wait time is important. To do this, it is recommended to specify by the link the Kbs for those pages (especially large PDF pages). For very long waits, it is recommended to do this, plus use countdown information at regular intervals that indicated how much of the site has been downloaded, and thus how much longer they have to wait (Dillart & Kahn, 1998).
    

• Where the waiting occurs within the Internet episode
    
  According to Dillart and Kahn (1998), consumers can separate the download experience from their retrospective evaluation of sites if the the waiting occurs in expected times, such as the initial download stage, then the negative carry-over effects of waiting will be minimal.
    

• How long the wait is compared to the consumers' expectations
    
  The old expression, 'a watched pot never boils' seems appropriate for users perceptions of waiting time. That is, users more often than not just stare at the screen until the page is loaded. However, Dillart and Kahn suggest that the duration of the download often has little effect on retrospective evaluation of the site if the wait is not salient to the user. To lessen the saliency of the wait, it is recommended not to place large (Kb) images at top of the screen or imbed important text within the images.

Another cause of slow browsing is due to a poorly defined structure. A lack of careful consideration as to where information should be placed often produces unnecessary or unexpected delays ultimately slowing down browsing within the site -sometimes to a point where it would be faster to find the information by other means - which often causes the user to leave the site. For example, Selvidge found that the average time to complete an information search task across certain airline sites was about 17 minutes. Time to complete the same task by calling the airline by telephone averaged 3 minutes (Selvidge, 1999). Not surprisingly, these types of delays have a strong negative effect on the overall satisfaction of these sites (see Briley & Stoltz, 1999).

However, it does seem as though people are more tolerant in waiting than they were in the past (see 9th GVU survey, 1997). As discussed in Usability News, Selvidge found that users were frustrated by 30 and 60 second delays in page loading time, but would tolerate the 20 second delays (Selvidge, 1999). Again, many factors are at play, such as the provided information, the expected waiting time, their motivation, and their general tolerance for delays.

Users also consistently rate non-working hyperlinks as one of the most annoying aspects of web browsing.  An easy way to reduce this problem is to use free services such as websitegarage.com, which tests websites for such things as download time, non-working links, and browser capability.

References

Optimal web Home | Usability News | WSU HCI Lab | Human Factors | WSU Psychology

Contact: Michael Bernard with questions regarding this site.
Last update: March 31, 2003
URL: http://www.optimalweb.org
Disclaimer and terms of use