I do not claim that this is a new recipe, but one of my belief's is that people (mostly me) place too much emphasis on "new" while thinking about startups and not enough focus on solving people's problems. The goal is not to do something that nobody has ever thought of, but to find a meaningful problem that you can help people overcome. Somebody has thought of pretty much everything, and pushing further into the unknown is generally unadviseable unless you have some deep pockets with a massive timeline financing you.
With no claim of novelty, here is my recipe:
1) find meaningful problem (meaning = big problem for few people OR little problem for many people)
2) do as little work as humanly possible to build a solution for the problem
3) validate that your solution addresses the problem with actual customers
4) repeat (but finding and solving a more meaningful problem with each iteration)
I will go on the record as saying this: if nobody has thought of your idea, it probably sucks. I am not wed to my recipe, but I do believe that it is important to overcome this harmful belief that you have to be in unchartered territory to build something meaningful.