When you send a manuscript to an agent or editor, it should already be the absolute best you can make it. That means you have revised it at least once, preferably several times and with the help of first-readers if at all possible. By the time you submit it, you can't think of anything else to do that would improve it. You're ready for someone else to make suggestions based on their greater skills, or demands that they are paying for the privilege to make.
"Cut 10-20% of your manuscript" is not professional advice. Any idiot can say that, whether it is relevant or not. What it usually means is something like this: "I am overworked and/or lazy, and a brief glance at your manuscript makes me think that it isn't very polished, so I'm just going to spit out a generic piece of advice instead of doing my job."
Yes, ladies and gentlebeings, part of an editor's job is to make manuscripts better, and ideally, teach writers how to work better. When you work with a skilled and diligent editor, you can learn all kinds of things about writing in general, the professional end of it, how to collaborate with other folks in the process, and how to revise your work. Some editors don't like parts of this; some aren't good at it, having different editorial skills instead. But it really ticks me off when this kind of laziness becomes widespread, because it can do a lot of damage. I understand, because I've been there, that editors get overworked. Here's a piece of advice to editors: if you're not willing to kill the chicken over the manuscript yourself, keep your mouth shut and let the writer go find someone else who is. You know where the rejection slips are, so use 'em. Unguided advice can do more harm than good.
See, it doesn't tell you how to fix your manuscript. It's about as helpful as a mechanic saying, "You need a bunch of new parts." Which parts? Determining that, or at least giving you a good hint, is their job. Not yours at this stage, unless you shirked your part earlier and sent them something unrevised. (Really don't. It wastes their time and yours. Editors hate it when writers send raw material, and it's a common gaffe.) So let's consider what it looks like when an editor does this correctly, which will further illuminate why the quick-and-dirty version is bad.
"This subplot is distracting/confusing/adds nothing to the plot. Delete it." This is an example of macro-editing, and it's one of my fortes. You go through the whole story in a vertical fashion, removing a complete thread because it's not really important. This tends to cut a lot of material. You can often do this if you tend towards fancy plots and intrigue.
"This section/chapter/chunk of paragraphs is excessive. The reader will get bored. Revise the first/the last/this little bit here, and cut the rest." This is another example of macro-editing where you cut chunks, and it's another good way to get lots of new breathing room. You can often do this if you run heavy on description in the way of detailing places or listing things that people see or go past.
"Your story really begins/ends here. Cut the rest of the beginning/ending and revise the severed edge into a tidy boundary." Plenty of writers find it difficult to pinpoint the best place to start or stop a particular story, especially in a series. This is macro-editing, and it can cut a few paragraphs or many thousands of words. (I once had someone cut one of my stories in half and that worked. I hadn't noticed that I'd essentially stuck two stories together; the editor did.) If you've made this mistake once or twice, really watch for it, because it's often a habit.
"You have a pattern problem. You keep repeating X. Fix Problem X by applying Solution Y. Go through your whole manuscript and do this." Here is an example of micro-editing. It goes through the manuscript horizontally and snips out a bit here, a tad there. Maybe you're overdoing descriptions, stacking adjectives, or using filler phrases. Delete them. This approach is useful if you're just trying to knock off a few hundred or few thousand words to get under a threshold, or if you tend to write fluffy and need to condense it. You're not really losing anything of substance. Doing this throughout a long manuscript can be very tedious but can usually be described briefly. This is not one of my strong suits, but I know other editors who are brilliant at it.
Of course, some writers tend to produce fluffy rough drafts. If this is you, pay attention to what editors want to cut when they make specific requests. You'll probably make the same mistake the same way in a lot of stories; that makes it easier for you to find and fix. Look at the above examples to see if any of them sound familiar. Other writers habitually write lean. They probably don't need to cut anything, and may have the opposite problem, needing to add details to place the reader firmly in the story or to provide support for later developments. Learn where you fall on this scale, by comparing your writing to that of other people and by comparing the feedback from many editors over time.
Now let's consider your story. There are many types of fiction, but for this context, we're concerned with the Story of Idea, the Story of Character, and the Milieu Story.
The Story of Idea is all about this cool thing that happens. It doesn't matter much where it happens or who gets swept up in it; the focus is on the action and events, or in science fiction, the nifty gadget or discovery, or some other Idea. People read these stories because they're intrigued by the idea and want to see where it goes. These stories should be lean and mean. Details usually distract from the main idea and bog down the action. You may be able to spot much of this yourself and weed it out during early revisions. Much classic SF is Story of Idea stuff; so are many mysteries, which are puzzle stories. If you're writing this type of story, cutting may indeed be a good idea, but a responsible editor will at least give you a clue what to cut and ideally why.
The Story of Character is all about this cool person and their adventures. People read these stories because they want to spend time with that person. They'll be interested in what the protagonist does, says, thinks; their background, their family, their hopes and fears. So these stories tend to be more involved and detailed, particularly in the human aspect -- relationships, feelings, personal growth, facing challenges and succeeding or failing, etc. You may or may not benefit from cuts here. If you're shooting the scenery, you may be overdoing it; cut what isn't supporting the character. But if you cut anything about the character, be very careful, because that can easily undermine later developments. This is where an expert's assistance is precious to you.
The Milieu Story is all about this cool place and the people and activities therein. People read these stories because they love the place and want to wander around in it. It's like a vacation in print, and you're the tour guide. So show them the sights! Taking a red pen to a Milieu Story is very risky; it's easy for one person to cut stuff that bores them but other readers might love. Sometimes these stories do need to be trimmed, but it's best done with expert advice -- say, if you're running a page or more describing a single location or object, that's usually too much, but in a crucial scene it might not be. Then again, you might want to cut some stuff out and save it for another story in the same setting, because people love returning to a favorite milieu (hence fanfic).
So how did we get into this mess? Well, editors are super busy and usually looking for ways to save time. Giving generic advice does that, but it's hard on writers and can be ruinous to fiction. Then too, there's the problem of book bloat, which has been getting worse gradually for some time and has become downright horrid in the last 5-10 years. Many a fine book has been wrecked by a writer trying to fluff it up to meet some idiot publisher's minimum threshold of 100,000 words or whatever. Forget that nonsense. Write each story to the length it needs to be, because hacking it down or bloating it will damage it, and savvy readers will be justifiably disgusted with the results. Novels start at around 40,000 words depending on whose guidelines you use. There is nothing wrong with a 40,000 word novel. Some stories really need 100,000 words or more, but most do not. So I can see where editors got tired of seeing the resultant bloat, but dudes, you are not going to fix that problem from that end; to banish bloat requires publishers accepting a wider range of sizes (or writers finding other venues besides publishers...) instead.
In conclusion, there are no Absolute Rules in writing or editing. There are a great many things that are prudent and effective. There are a great many things which are popular, fashionable, and widely expected. But if you look at a few dozen sets of submission guidelines, you will discover that every editor has their own quirks and their own bundle of expectations. And as you write, you will discover that bending and breaking rules can be great fun and terrific inspiration, once you know what you're doing -- or heck, even just as an exercise.
So the truly useful advice is this: Know the skills and parameters of your craft. Make your story true to itself. Revise it to the best of your ability, with the help of friends if possible. Then find an editor who believes in it enough to assist you with the final polishing by explaining what needs to be fixed, and preferably why and how. If they're not willing to do that, they're probably not going to be very helpful to you or your story. Move along; if your story is good, someone else will be willing.