This is essentially about lay-out and visual structure, and using a couple of more or less random people to help thinking ("Random" people can mean someone passing your office, your neighbour, cousin, boyfriend).
Building interfaces is a tricky business. All the wonderful designs in our heads are sparkly and dripping diamonds of course, but when it comes down to it, it is about how the users interact with it that will tell wether it is a success or not.
I know some people use video and eye-tracking in analysing interfaces and user interaction; that is not the level I am talking about. What I am after is simple user testing long before a system is rolled out. Quick and dirty they call it: user testing from very early on. It is mainly a tool to make you think, and ensure you have considered as much as possible.
The benefits are many; some glaring blunders might be sorted early, some of the users input can of course also be ignored. But it is always better to know you are ignoring something, than to not know. It is all about informed, intelligent choices.
One that I have used sometimes is what I call the post-it note game. For structural visual architecture: you give people a bunch of post-its with key words written on them, and ask them to sort them. It is also helpful to get them to sort the post-its on a piece of paper/wall, and pretend that it is the application. It could be functionality, menue-points, keywords. Often, some interesting anomalies crop up, that helps a quick rethink. Sometimes the fact that the guinea pig does not know where or how to categorise and place something, is a useful hint.
Another one is making a very primitive sketch, preferably on a whiteboard. It is important that this is something that does not looks remotely finished, and a long way from the digital product. Simple, primitive boxes for content, buttons and other elements. Tell the guinea pigs what the application will be all about, give them a list of keywords, and ask them to write in where the various functionality should be visually placed. They should also be free to add things that they feel are missing. Let people fiddle as much as they want with it.
(In my experience, wireframes and mockups are rarely that helpful. It might make the designer feel that she is doing something, but a lot of people do not quite grasp the fact that they are ideas, not products)
My question is: what other methods have you used, that proved helpful? I would be very much interested in tools, methods and experiments that have given interesting and unexpected results, other ways of pestering users way before coding and colour-schemes comes into it.