A common misstep I see dungeon masters make, myself especially, is to put off their good ideas. They, or rather I, take their craziest, most heroic adventures, monsters, dungeons, what have you and plan to run them sometime down the road. For myself at least, there are several reasons for this, working together to make my campaigns somewhat less than they ought to be.
The first is the idea of a campaign progression. The idea that I start with small, mundane challenges before diving into the realm of the fantastic. This makes sense, on both a storytelling and gameplay level- you need to set up the world before the characters see its extremes, and you need appropriate challenges for low level adventurers. The problem is that I often overdo it. I’ve never run a liche or a rakshasa, never ran an adventure in the lower planes, even though I’ve run more than one campaign that has lasted over a year. I don’t want to throw out my best stuff right away, and often never get to it. Especially since most of my campaigns don’t last more than a year.
Then there is the concept of worldbuilding and verisimilitude. The idea that there is a sorcerer lord every twenty miles, the forests are filled with giants and dinosaurs, makes it kind of hard to believe that the npcs could survive at all. This is even worse when I decide to run the travel and hold back on the random encounters. Sure, as you are travelling the roads from one end of the country to the other your characters probably shouldn’t encounter anything more than bandits, goblins, maybe a pack of unnaturally vicious wolves. But the problem is obvious, my players have fought a lot of bandits and wolves. One big headache I've had as a player is a DM throwing waves of mooks at my high level PC, since it would make sense that you're more likely to encounter an army of humanoids than a menagerie of monsters.
Then there's balance. Having started in 3rd edition, the books drilled into my head that balance was part of the game. CRs, encounter building guides and whatnot really pushed the idea of a balanced progression of challenges for characters. This, of course, is absolute rubbish. This isn’t a video game, it’s a story- characters shouldn’t encounter a series of creatures and challenges designed to match their current level of ability, they should go headlong into danger, not knowing if they can win for sure. They shouldn’t feel like they can handle everything thrown at them head first because they can expect to only use a quarter of their resources in a fight. This is something I realized first as a player after I knew the system, which I learned very well. I knew when I wasn’t in any real danger, which was most of the time with my soft, daycare DMs. This isn’t to say I’m fully against CRs as a whole- they are a useful tool for a DM to know just what they are doing to their players. No DM wants to force a combat encounter that the players can’t win and can’t run from without knowing they did just that, but the idea that encounters should always be an appropriate challenge is a bad one.
The last issue might have some merit- though not as much as it might seem. You don’t want to make the fantastic the mundane. If the players fight nothing but dragons, then dragons are just orcs, and that's not good. But again, the players shouldn’t only fight orcs. While you shouldn’t have nothing but epic encounters though, you do need epic encounters. Songs for the bards to sing and for the players to tell for years. The worry that you’ll burn out the fantastic is a dangerous one, especially if you have short campaigns or an irregular schedule. If you aren’t playing the same campaign every single week, you really should go for all killer no filler, becuase the idea that the exciting session is in the future doesn’t make a slog session any better.
So what am I suggesting really? Every adventure should be a tale. Every adventuring area, whether its a dungeon, or an atlas hex (usually a thirty mile hex made up of five mile hexes) should have something mythic. A dragon, a true giant, a liche or vampire king. Something grand and fully capable of a total party kill. It shouldn’t just be there though, plant the seeds, leave evidence of it around the dungeon or wilderness, let the players know about it- they can avoid it or hide from it, or choose to die gloriously, or figure out a way to succeed. On a hex map this could be a dungeon or well defended fortress rather than a single entity, but it has to be a fool's errand to take it on. Because foolishness and bravery are usually the same thing.
And use more bloody dragons! Not level appropriate juvenile ones- proper strange and cunning beasts of lore-it’s in the name. Make them too strong to take on. Leave ruins created by their wrath. Make them cunning and forethinking, have them make deals with the party, have them purposefully manipulating the countries they dwell in. Have false dragons, stories exaggerated by townsfolk trying to make wyverns and pythons scarier than they are. Not in every dungeon, not in every hex, but don’t hold off until they are a “balanced” challenge to the players. Because either the campaign won’t get that far or the players will, at that point, be able to beat it soundly, which makes a dragon more mundane than having too many. If each dragon is a unique memorable npc, and a glorious battle where the players scrape by through guile and blood, then it won’t matter that it’s the third one this campaign. Life is too short to not fight dragons, even if it’s shorter if you do.