User psychology dictates – or should dictate – many of the choices in interface design. There is no such thing as "one size fits all." Mechanical engineers don't want flashy graphics. Fourteen-year-olds insist on flashy graphics.

We have found qualitative testing of potential users to be the best process: showing small groups of potential users simple prototypes, getting their feedback, and then performing a couple more iterations.

Aside from being better, it's faster and cheaper.

(Quantitative testing tends to be rigid, and only answers a small range of questions. Because qualitative testing involves more dialogue between us and the users, they tell us things we wouldn't have thought of on our own.)