When solving a software-design problem or a web-design problem, you get the best results from following a design process. I’m not referring to something I made up. I’m referring to something that people who are trained in “Design” will recognise as a design process. And such processes inevitably include divergence and convergence.
Divergence is the stage when the designer or design participants come up with ideas. Lots of ideas. Not just comfortable ones, but ideas that push at the edge of the envelope. Ideas without judgement.
Convergence is the stage when ideas are judged, rapidly iterated, assessed, and checked against the requirements.
In my experience, divergence is best done by sketching, with a pen on paper:
Sketching on paper with a wide-nib pen is best because this “technology” is familiar—it’s easy—and inexpensive. Everyone can 
have a pen and paper, so there’s no need to wait your turn. Sketching is fast, so designers have little invested in any one sketch. Modifying or iterating any one of the ideas is easier to accept. Most importantly, for inexpert sketchers, the sketching process intrinsically discourages high-fidelity work and the wide-nib pen discourages sketching the fine detail that detracts design participants from getting the ideas out, fast.
Software that’s intended to help people “sketch” detracts from a good result. The software often lets users build prototypes with a great level of interaction detail. The problem with this: attention to detail is a distraction at the divergence stage. When filling the design space with ideas, it’s sufficient merely to evoke the idea and share it. Ambiguity actually helps the team see more than one idea in the same sketch; each interpretation of the idea can be iterated. Software that’s not intended to help people to “sketch” slows them down at the divergence stage—just when we need ideas to flow quickly. Divergence is about filling the design space with lots disposable ideas.
Read my blog post about why ideas are disposable.