Note that this isn't the first step. In order: identify a problem, talk with people (first with those you think have the problem) to learn more about it: Is it actually a problem? How are people addressing/solving it? What are the pain points? Is it severe enough to warrant a better solution? Would anyone pay for a solution? How much would they pay? How many people would pay for that solution? Can I make the solution? Can I distribute the solution?

If you get a bad answer to any of these, you need to do some thinking and talk with more/different users, and if no change then you probably can't make a business on this (and don't need to continue along the chain until then). Once those are true, then yes, sell to the people who can write a check.