Introduction:

Every year since 1983, I have reviewed (and summarized) most of the usability-related research literature that was published during the previous year. This has provided the basis for the popular, annual 3-day User Interface Update course. My annual two-month read and review activity provides me with a number of research-based insights into "what works" and "what does not work" in usability. I have listed some of these insights below. What makes these "Do's and Don'ts" unique is that they all have recent research to support them.

General Observations

1. Do evaluate the goodness of a Web site by evaluating content, navigation, visual design, functionality and interactivity, but realize that these elements are not totally independent (i.e., Web site evaluators have a difficult time separating them from each other).
2. Do make Web sites appear more credible by:
   • Making it easy to distinguish ads from content,
   • Providing quick responses to customer's service questions,
   • Listing the organizations physical address,
   • Having an organization that is well respected outside of the Internet,
   • Updating the site frequently with new content,
   • Allowing users to search past content, and
   • Providing a professional looking Web site.
3. Do use Flash to enhance interfaces knowing that 98% of browsers are capable of playing Flash without downloading.
4. Do design Web application (Weblications) interfaces different than traditional Web site interfaces.
5. Do use "parallel design" concepts when designing user interfaces, because participants:
   • Can reach a consensus quickly,
   • Will consider numerous different design alternatives, and
   • Respond to good ideas even when they are contained in poor designs.

Display Issues

1. Do use good headlines when designing newspaper, newsletter and other information-based sites because eye tracking studies show that users tend to look first at text, not images (a margin of nearly two to one favoring text).
2. Do consider the size of textual information to be used as entry points, because
   • When text and images are of similar size, text is more likely to be an entry point,
   • Images must be much larger than text to act as an entry point, and
   • Larger text dominates over smaller text (normal sized text rarely acts as an entry point – even if bold or a hyperlink).
3. Do put the most important prose text and bullet item information toward the top left of a page, because users tend to scan from left to right, and top to bottom when reading, but do not follow the same pattern between images, titles or links.
4. Do use common and contiguous background colors to help guide users in finding information on a page.
5. Do put the most important information on a Web page at the middle-top of the page because that text:
   • Will be seen first, and
   • Text at the bottom of a page is rarely seen.
6. Do use right and left panels for links, because users will investigate areas outside the center area:
   • When searching for a specific link, and
   • When visiting a Web site after the first time.
7. Do not use fonts on Web pages that are less than 10 points.
8. Do use 12-point fonts (size=3) on Web pages to elicit optimal reading speed for most adults.
9. Do use any of the most common font styles (e.g., Arial, Verdana, Georgia, Times New Roman), either serif or sans serif fonts, to elicit the fastest possible reading speed.
10. Do use sans serif fonts (e.g., Arial, Verdana) to satisfy user preferences.
11. Do use 14-point fonts for older users.

Interaction Issues

1. Do carefully consider the task being performed before making decisions about the breadth and depth of a Web site, because sites with numerous levels may or may not degrade performance.
2. Do carefully consider the task being performed before making decisions about using or not using frames, because some uses of frames will degrade performance.
3. Do not consider users' judgments about "ease of use" as accurately representing the speed with which a task can be performed.
4. Do use similar item justification and scroll bar orientation to produce reliably faster selection times, fewer errors and improved preferences in the use of list boxes (and possibly Web site scrollbars).
5. Do consider the experience level of users and "frequency of use" when selecting widgets in Web sites, particularly when deciding between using entry fields and dropdown lists (experienced users are faster with entry fields).
6. Do place commonly selected objects close to the edge of a window to expedite their selection.
7. Do not be concerned that frequent keyboard use will cause carpal tunnel syndrome.

Web Site Design Issues

1. Do design Web sites primarily for use with the Microsoft's Windows operating system, unless different (more specific) information about the operating systems being used by the target audience is available.
2. Do design Web sites primarily to be displayed by the Internet Explorer and Netscape browsers.
3. Do design Web sites to work best with screen resolutions of the majority of typical users, which worldwide is now 800x600 pixel resolution.
4. Do design Web sites for use on monitors that show at least 256 colors.
5. Do design pages to load in a few seconds; however, keep in mind that some users will rate the download speed of slow pages as "fast" if they are able to successfully accomplish what they intend to do.
6. Do design a Web site that encourages either linking or searching behaviors by users.
7. Do design a search capability so that most users will complete a successful search on their first attempt.
8. Do design to encourage users to access links (not search) if it is important for users to continue to browse the Web site after locating their primary target, because:
   • About 60% will continue to browse the site (only 20% of "searchers" continued to browse)
   • "Linkers" are three times more likely to find related (valuable) content
9. "Linkers" who started with links on the homepage looked at almost 10 times more content pages
10. Do design Web sites to have a "moderate amount" of white space, because:
    • Users prefer a moderate amount of white space, and
    • There is no reliable performance differences with differing amounts of white space.
11. Do design Web sites using a "Fluid" layout, because:
    • Users believed that the "Fluid" layout was best for reading and for finding information, and
    • There were no reliable performance differences among the methods (left justified, centered or fluid).
12. Do design Web sites with embedded links, because:
    • Users preferred having them embedded, and
    • There were no reliable performance differences between embedded links and those outside the text.
13. Do design Web sites with links that are both embedded and outside the text on the left margin, because redundant links were preferred by users.
14. Do design Web sites with the links in a frame on the left margin, because users preferred the frames rather than having the links scroll off the page.
15. Do design Web sites that enable textual chat (Instant Messaging) between users and Web site assistants.
16. Do design Web sites so that if users must be placed in a queue:
    • They can be continually updated, and
    • The updates can include sounds (so that users can work on other things).
17. Do write prose text in Web sites at appropriate reading levels for the target audience as calculated using a commonly used readability formula, because:
    • The average reading level in North America is at the eighth- to ninth-grade,
    • About one in five adults read at the fifth-grade level and below, and
    • Adults tend to read at least one or two grade levels below their last school grade completed.
18. Do use Rapid Serial Visual Presentation (RSVP) for presenting information to users who need to read prose text from small displays (PDAs, cell phones, wrist watches), because:
    • RSVP enables an acceptable level of reading performance, and
    • There is no performance or preference differences among 10-line, 3-line and RSVP.
19. Do use 20-point text presented at speeds of about 250 wpm on RSVP-based displays.

Usability Testing

1. Do design "low word count" Web sites (average of 65 words) using fewer words, smaller page sizes, fewer graphics, faster download times and more font variations (particularly between header and body text).
2. Do design "medium word count" Web sites (average of 230 words) using less body text, organized better into clusters (lists and shaded table areas), using lower percent graphics, and with more colors to help distinguish headers.
3. Do design "high word count" Web sites (average of 800 words) using less body text, organized well into clusters (lists and shaded table areas), with more text links, large page size, lower percent graphics, and using more colors to help distinguish headers.
4. Do use eye-tracking methods to determine confusion rates from users in usability testing.
5. Do require usability testing clients to make specific and focused requests, including usability objectives, specific definitions of success and clear time limits for task completion.
6. Do plan on inadvertently creating new usability problems when fixing problems identified by previous usability testing.
7. Do plan on finding and fixing about 37% of the problems in a system with each performance test (i.e., each iteration), which means that to detect and correct 95% of the usability problems it would take
   • At least six tests, and
   • 105 participants.
8. Do have members of the design team use a prototype Web site to create an estimate of the time it should take an experienced user to access information; then use this time to set usability objectives (e.g., 85% of the users shall not need more than 125% of the expert users' time for each task).
9. Do use a "large" number of users (e.g., over 100) for performance tests, because of:
   • The complexity of Web sites
   • Hundreds or thousands of pages
   • Hundreds of features and paths, and
   • The need to determine the prevalence of problems in target populations,
   • Only 35% of problems were identified after testing five users.
10. Do use automated testing methods for some, if not all, of the performance testing for a new system, because automated testing allows:
    • More rapid data collection and analysis,
    • The use of a large numbers of users, and
    • Quick evaluation of content that keeps changing.
11. Do provide online learning materials that primarily use a traditional prose text format, because it elicits better learning (however users preferred a hypertext format).
12. Do not use testers or test participants to be involved in a test when they have a cold, because:
    • They will respond more slowly,
    • They will detect fewer problems, and
    • They will have reduced energy and motivation levels

Users

1. Do use the following age categories when designing and using study information:
   • Old-old: 75 and older
   • Older: 60-74
   • Middle-aged: 40-59
   • Young: 18-39
2. Do ensure that important system (Web site) sounds are louder for older users.
3. Do use TFT screens rather than CRT screens to elicit the fastest possible reading performance from users (older users benefit even more than younger users).