What is needed before you can do a walkthrough?

In order to perform a successful walkthrough you need:
  1. a fairly detailed description or prototype of the user-interface; (it should include button labels, placement, colors, etc.)
  2. a task description, either of a complete task or a part of it; (See the section on task-centered design for how to describe tasks.)
  3. a complete list of actions needed to complete the task using the interface;
  4. an idea of who the users will be and what "common knowledge" they have.
Before starting the walkthrough you should make sure that the list of individual actions is correct. A walkthrough is not an exploration of the interface or a "trial and error" approach towards performing the task.

Nonetheless the walkthrough may reveal that the user may have problems identifying or performing one of the actions, or interpreting the feedback.

In the walkthrough example of turning on the PC to start writing a letter, we already noted that the user may have problems identifying whether the computer is already on or not. So you should identify beforehand whether it is "common knowledge" to look for a green power light on the front of the computer case, and to check for the noise of a fan and/or hard disk. The complete list of actions may have to include a step:

(Note that we should not write "the button next to the green light", because when the light is not on it probably does not look green.)