• Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint

Lab 5.2 Exercises

5.2.1. Apply Web Page Usability Measurement

a) It has been said that only five users are needed to test out initial Web page designs. Discuss why you think this is valid.




Answer: Early Web page designs are so bad when applying the standards of today that only five users are needed to discover the faults. Because so many people still use the same techniques and approaches, the statement still applies to Web pages today.

Note that these user tests can be done quickly in the space of a day or so and are relatively inexpensive tests to conduct. However, as with any test, you need to have a good cross section of people testing. For example, you will get lopsided results if the only people testing are from the same area within the company. Highly trained IT who understand the technical aspects of the software give a different result from clerical staff who use software from a user perspective, who, in turn, give different results from managers, who only use their computer once a month.

The measure we are making here is the percentage of test users capable of accomplishing the specific task they were given. Many users will give up after a certain time in tests. Bear in mind that in real life they are likely to give up much earlier. This means that the percentage derived from tests will most likely be an optimistic number.

A useful side effect of the success rate measure is that a thinking-aloud study can be carried on in parallel. Making users think aloud as they attempt the tasks results in very useful qualitative feedback for the Web page designers.

Of particular value on the Web are usability measures for novice users. Users rarely become highly experienced with a particular Web site since they visit so many. Note that this will not be true for intranet use and some external sites that undertake crucial support for a user's everyday tasks. Here it will be necessary to acquire usability measures for expert users as well.

Nielsen tells us that if you are satisfied being an average site, then you could set a goal of having a success rate of slightly less than 50% for the most common things people want to do on your site. Being average on the Web is not usually a good strategy because user choice is so vast. Nevertheless, this might be a good way to go when design funds are very limited or a Web presence is vital at all costs.

Success rates of 90% or even 95% are overly ambitious and probably not realistic on the Web today. A more realistic success rate might be 80% for a company that values its customers very highly. Achieving a success rate of 60% is a reasonable target for most other projects.

b) What is the best usability measure for Web page designs?



Take a look at the Yahoo Web site at: http://www.yahoo.com


Answer: For Web designs the simplest usability metric is the success rate for a user performing a typical task, i.e., the percentage of test users able to complete the allotted task, the average time it takes a specific number of users to complete the task, and so forth.

Keep in mind what “usability” means. Too often usability is equated with “ease of use.” Although ease of use is an important factor is usability, it is by no means the only one. For example, something that is easy to use but does not allow the user to complete the task within the required amount of time has low usability.
c) Discuss what aspects allow the Yahoo site to load faster than other sites (such as the CNN Web site, http://www.cnn.com).




Answer: The first thing that should come to mind is the almost complete lack of any graphics. Although there are a few at Yahoo, they are fairly small. This means the page is almost completely text. There is a lot of information, but it can all be quickly loaded. In contrast, the CNN Web site relies heavily on graphics. There is a least one graphic for the lead article, advertisements, and many other graphics. Granted one picture is worth a thousand words, so it is expected that a Web site devoted to news would have pictures. The result is that takes longer to load the page.

If your visitors have wait too long for the page to load, it typically isn't usable. I know many applications written for traditional GUIs where there is an extremely long load-up time, but this decreases the access time later. For example, static information is loaded from a database, which takes a long time to transfer. However, once the data is on the local machine, reaction times are very fast. The tradeoff is slow loading times, but fast reaction later on.

The Web is different. First, you cannot expect to have all of the data on the local machine. Neither you nor your users want this, really. Therefore, you need to work on fast loading because the reaction time (i.e., database queries) is dependent on the connection (assuming that the programmers have optimized database activity).

Note that limiting the amount of graphics is not the only way to decrease the load time. Some sites (such as Yahoo) take advantages of something called “client-side image maps.” Most people are familiar with clicking on an image to load a particular page. You will also find almost as many sites where clicking on different places within the image loads different pages. Until recently this was done with server-side image maps. The browser was able to identify the location on the page where the user click and send that information to the server. The server then looked in its map file to find which URL to load. It then sent the URL back to the client, which actually made the request. This back and forth communication takes time.

Recently, more and more pages are using client-side maps now that more and more browsers support them. Instead of having to send messages back and forth to figure out what URL to load, all of the work is done at the client side. Here is how a client-side map looks on the Yahoo site:

<map name=m>
<area coords="0,0,52,52" href=r/a1>
<area coords="53,0,121,52" href=r/p1>
<area coords="122,0,191,52" href=r/m1>
<area coords="441,0,510,52" href=r/wn>
<area coords="511,0,579,52" href=r/i1>
<area coords="580,0,637,52" href=r/hw>

When you click on the rectangle enclosed by any of the listed coordinates, the client immediately knows which page to load (the href= tag). This means there is nothing send to the server except for the page request. Since this will happen in any case, the loading time is a lot less.

Another good example of this is the SCO Web site (http://www.sco.com). Although it still uses client-side maps, it couples them with JavaScript for some interesting effects. Some sites get the same effect using Java. However, this means retrieving even more data from the server, which takes even longer to load.

Graphics and Java applets are not the only reasons why pages take long to load. It might just be that there is too much text on the page. The author goes on and on and on and on and on and there are few breaks in the page and it is extremely hard to keep track of where you are in the text.

A different method for estimating average Web page size is described in Sullivan (1999). He does random probing of the Web. His results are shown in Table 5.2. Note the size is 66% larger than the Flanders' number for very popular sites. Maybe Web page user interface designers generally should take a closer look at these popular sites. Note also that the figure for 1999 has dropped by 1.6% compared with 1998. Perhaps page designers are actually responding to good Web page design guidelines.

Table 5.2. Average Page Size
 May 1999May 1998August 1997
Pages Sampled200 pages213 pages44 pages
Avg. Page Size60 KB61 KB44 KB

For an up-to-date sample, you should check out the Gif Wizard site (uswest.gifwizard.com). This site survey lists the top 100 sites using a number of different measurements, plus they have a wide range of tools to help make your own site more efficient.

5.2.2. Recognize Examples of Bad Web Page Design

Visit Flander's Page “Web Sites That Suck” at: http://www.webpagesthatsuck.com. Additionally, visit Stevyn Pompelio's site “How to Make Annoying Web Pages” at: www.users.nac.net/falken/annoying/main.html.



Not a subscriber?

Start A Free Trial

  • Creative Edge
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint