While the things we mostly discuss here as constructors and constructors-to-be are technical issues, solvers seem to accept all kinds of technically arguable things in a puzzle short of an inability to fill in the white squares, but a yawner is ignored. There has to be some element of what I call sparkle in a puzzle. Something has to be interesting about it. Sparkle can come from the theme, fill words, clues, whatever, but whatever it is, and however it may be defined, it has to be there or the puzzle passes unnoticed, even if it's published.
We can argue forever about partial phrases, numbers of black squares, French words, crosswordese, Roman numerals, abbreviations, and alphabetic runs. But unless these things are way overdone, they pale in significance when compared to the major issues: Is the puzzle solvable? Is it entertaining? Does it have sparkle? Of course, an editor's no-nos must be observed, but they all have different no-nos when it comes to the details. On the other hand, all the editors for the major puzzle markets want to see this sparkle.
We're living in a golden age of crossword puzzles with the introduction of the software computer programs, first by pioneer Mel Rosen and then with Antony Lewis's Crossword Compiler. More people are progressing from neophyte to published puzzle constructor in less time than in any previous time in puzzling history. I think that the competition will be progressively fierce, and what will distinguish those published in the top markets will be this element of sparkle. Filling grids with acceptable words, once the greatest of hurdles, is becoming the most easily mastered of the chores. Of course some new constructors will also find their way to success without these new tools as has always been the case in the past.
So my advice to new puzzlers is not what it would have been six years ago. Then I would have and did stress the technical issues of word fill and told everyone not to worry too much about clues because that was the least important issue. I no longer believe that. Editors are only human, and, particularly in the absence of a hugely novel theme, it's a lot easier to go with a puzzler who predictably sends clues and titles that require less work, all other things being equal. An acceptable fill is becoming an assumed feature of submitted puzzles even though it may appear otherwise in the questions addressed to cruciverb-l. At the time of this writing the highest paying editor for freelancers does not even require a finished fill for a submitted puzzle although he doesn't guarantee that he can finish the fill either. Achieving a satisfactory fill is no longer the formidable primary task that it once was, and achieving that hard-to-define sparkle is huge.