Studies : User Testing

1. Find users that represent the target demographic of the site

According to Neilsen (2000), a group of five users is sufficient to run a test. This makes it easier to keep the sample groups small and run more frequent tests throughout the design process and ensure that the design changes recommended achieve the desired effect. A true, proper sampling of users is 15, however Nielsen recommends sampling five users three times for iterative testing rather than 15 users one time.
Medlock et al (2002) confirm that in most cases four to five users can detect 80% of the problems with an interface. Therefore, it is most cost effective and productive to run smaller, more frequent user tests than larger tests less often. There may be times in which it is appropriate to use a large number of test users, for instance, when the target audience is a widely diverse group of individuals with specific distinctions.
2. Ask the users to perform tasks that are within the scope of the intended function of the site
3. Observe the users and do not interfere with their performance