We are all master procrastinators. At any opportunity, we'll re-write that chorus hook, replace, repair, re-track, reinvent. To some extent, this is what making a record is all about, but it comes with a little advice. With modern computer-based production, there is the potential for songs to sit unfinished for a long, long time. Although today's recording landscape allows almost limitless revision, editing and arranging possibilities at the click of a mouse, it is always worth considering what's important, what is necessary, and how the song is best served. Some of the work I am most proud of, was done very quickly, instinctively, with time or tool constraints. I don't think this is just a coincidence.
There are many reasons for wanting to go back and forward on ideas, I understand. Many fear failing to impress with their work - they don't want to be judged by this snapshot of their ability, or they set the bar mega-high, or they focus on the negatives in a track instead of moving forward on the positives. Perfectionism is OK, but learn to filter it. One finished track, on which you can learn lessons and take them into your future work, is way more valuable to you as an artiste than a thousand potential masterpieces at the bottom of the trash can.
Now we come to the hard part - how do we know what to focus on, and what to skip by? Well, that comes down to working on those skills over time (or using a good producer who already has them) but in either case make sure you think about it. Work out what elements of your track are actually contributing to what you are trying to get across. Work out what it can't live without. Determine where the meat is, and which elements are the salt and pepper. Lastly think of the "big picture", because ultimately that's all-important. You don't need to spend hours crafting one backing vocal that hides in the back. Fire it in, and move on. See it away!
The old songwriting advice comes to mind here: write 20 songs a day, and throw 19 away. Not all things are meant to be finished, but most of the time, there's no lesson learned if they aren't.