Figure out the smaller jobs that help get the higher job done.
Define a target audience that fits the job story situation and context.
You only need to interview 5 users 1-on-1 to see the patterns.
Start with the basics before going deep in specific directions or looking into new developments in the industry.
Analyse current solutions. What's nice? What sucks? What can you steal?
Get in the field, experience the problem firsthand. Suffer so you can understand your users better.
Talk to experts
Try combining the best parts of some solutions to create robust concepts.
List all the viable solutions and vote on what to pursue. Kill the rest.
Gut check. Design is an art as well as a science. What do you feel you should pursue?
It's time to refine the survivors into fully fledged concepts.
List out your assumptions on each solution and figure out how you might test them.
Don't be perfect, make it good enough to get the idea across.
Focus on the 20% functionality used 80% of the time.
Repeat steps 5-7 until you have something that works.