A nice book about web usability. But also applicable to other user interface design cases.

This book points out that, users do not do what the designers are expecting. Designer spent a long time to make the detail correct but the users don’t watch it. Simply,

  • users don’t read a webpage, just scan it.
  • users don’t make optimal choices, just take the first reasonable option they found
  • users don’t study how things work, just muddle through it

Because of that, a UI design shall be trivial in the sense that its design suggests the usage, but not the words or details.

Some concrete design rules mentioned in the books:

  • make a clear visual hierarchy: visually nested, visually related
  • align to conventions if possible: leverage on people’s habits
  • keep area borders clear, make clickables obvious
  • keep the noise down: cut background noise, omit needless words, no one read instructions
    • try use bold/color to make things stand out if needed
  • use mindless choices: different choices shall have clear distinctions

Compare to other media, web is unique as the reader has no sense of the scale, direction, and location. That is, the reader does not know how much materials are there, where a page is located amongst many pages, and how far it is from another site. Therefore, a navigation bar (e.g. color-coded tabs of Amazon.com) is a characteristic element on a web page. Usually it contains the site ID, a home button, a search box, sections of the site, and the “you are here” indicator.

To test if a web page is usable enough, do the trunk test: based on the appearance, but not text, we should be able to tell the site ID, page name, sections, local navigation, search box, and the “you are here” indicators.

Finally, the book tells how to do usability testing: It is not like the marketing focus group. Just to observe one man using the web for several minutes doing something. This is because the focus group may want to collect quantitative data (how much percentage of something) but usability problem either exists or not, it is qualitative. Thus if a use case reveals a problem, you don’t need another tester to confirm it. The other book of the author (Rocket Surgery Made Easy. New Riders, 2010) is a manual on doing such usability testing, but less interesting to read.

Bibliographic data

@book{
   title = "Don't Make Me Think",
   edition = "2nd",
   author = "Steve Krug",
   publish = "New Riders",
   year = "2006",
   classification = "TK5105.888 K78 2006",
}