There is a concept in economics called sunk costs. These are the dollars you have already invested and cannot recover. Bad behavior, like continuing to repair a car that's a lemon or pouring money into a doomed business, come out of the psychological attachment to these lost investments.
Revisions are the sunk costs of writing. The time you spent cleaning up grammar, making dialogue sparkle, and adding hooks to scenes and chapters that are not needed are lost, but the investment can make writers hold onto these with both hands. (The same is true for doomed books, but that's too depressing to consider.)
Of course, time spent is not completely lost. Learning takes place, especially early in a writer's life. But a productive writer finds way to approach the material that reduce big investments in time that won't pay off and are likely to create too much devotion to words that must be cut.
This is why are recommend a "big to small" approach to revision. Essentially, this means beginning by looking at the story as a whole -- whether it works at all and whether there are scenes or chapters that must be cut or added. Then following through with reexaminations of things like character, conflict, and scene construction. Looking at pacing, and ending with polish and the mechanical elements (spelling, grammar, wordiness, etc.).
Your order may not be my order. For instance, some people need to do a spellcheck of the roughest manuscript before they return to it. But, overall, going through dedicated drafts is much more efficient for most people than taking on all the elements of revision in one draft.
For me, the first step is preparing myself. I need to let the manuscript cool off for a month or so, otherwise, I can't see it clearly. I need to accept that lots of wonderful scenes I fell in love with are really a mess -- and fixing them is what revision is for. I need to be prepared to cut, cut, cut.
My second step is printing out the whole manuscript, and then sitting down and reading it from beginning to end in as short a time as possible. I take notes, but as few as possible. I actually budget myself to what I can fit on one page. I don't write on the manuscript itself.
My third step is firing up the computer and having the manuscript read to me (text-to-speech is available to everyone, and it is marvelous), about ten pages at a time, while I follow along and make small notes -- usually just check marks -- on the printed copy. I am looking for story, holes, unnecessary passages, inconsistencies, missed opportunities, dull parts, and failures of story logic. But I'm doing it at warp speed. I actually have text-to-speech cranked up to its fastest rate.
When I have done all of this, I make a copy of the manuscript file, label it with the revision number (2.0) and usually the date, and begin to change the text.
I'll talk about more steps next time.