I've been researching intently on different approaches and Krug's pragmatism is very appealling. I'm not a web usability consultant with a team and $100,000 contract (I've worked in that situation in the past - but as the client). I'm just a wee library tech head with wide ranging responsibilities and no dedicated staff. I can draw on my colleagues to help but they too have wide ranging responsibilities and little time to dedicate to other projects.
Simple points Krug makes:
- Usability testing can be done in house (it's not rocket surgerytm)
- Any testing is better than no testing
- 3 users will tell you 80% of what 20 users will tell you
- Don't sweat on demographics/user selection:
- Big problems will be big problems for everyone
- There is no such thing as a typical user
- Two male first year law students from a non-English speaking background could be as different to each other as they are from any other demographic slice
- One user test early is worth one hundred late
- Never stop testing
- Analytics: can tell you a great deal about what people are doing on your site, but not why they are doing it e.g. are they spending a lot of time on one page because it's exactly what they are looking for or because it's completely confusing and takes a long time to digest?
- Focus Groups: great for sampling users opinions and feelings in the abstract, but not great for determining whether your site actually works
- User Surveys: asking people to remember what they did, when, and how often will result in a cleaned up, fuzzy version of what actually happened. Nothing beats observing in real time.
- A continuous improvement model
- A user centred approach
- An evidence-based approach
Step One is defining the site's main purposes. Stay tuned for what we come up with and more about Step Two...