I think you really want to be marketing concurrently with doing the thing. Saving it all for the end is probably a mistake. And a lot of projects don't have defined "end"s, like open source projects, or websites.
That depends. If OP’s job is marketing then doing it before is doing the thing even if it pisses off the people who’ll have to do the thing OP made up.
Marketing can also lead to you receiving feedback for what the users actually want. Doing the thing does not imply doing the right thing. You can be super productive building the wrong thing that no one wants.