...
Traditional Usability Testing was about gathering a statistically significant result to determine which facet of the interface was more or less usable. This is not the kind of Usability Testing which I am proposing. In his article 1994 Guerilla Guerrilla HCI: Using Discount Usability Engineering to Penetrate the Intimidation Barrier Jacob Nielsen, a leader in the HCI/Usability field determined with another researcher (Tom Landauer that "the maximum benefit-cost ration is achieved when using between three and five subjects". This method is now considered to be a valid method of determining roughly 80% of the usability problems with a given product. It is not designed to yield research level results, but to identify usability problems to be corrected.
By using these kinds of "Guerilla Guerrilla HCI" we can accomplish simple, quick, relatively cheap Usability Testing. This testing will not identify every problem, but it's a "good enough" solution to help us optimize our efforts. Plus, since we are a self-contained department with a full suite of documentation tools at our disposal, the usability facts we learn in one project can and will carry over to other work and allow us to have fewer usability problems going forward.
...