Everything you think about Game Mastering is probably wrong: Part v: You can't force Epic
So many pieces of GM advice I come across are about trying to set up epic and memorable encounters. Drama! A sudden reveal! A turn-coat! A sinking ship that is on fire while you sword fight on the sails! Wouldn't that be great? Absolutely it would! Unless it's forced.
I don't think I need to belabour this point too much, I think everyone really does agree with this (or would if they think about it). Rolling a natural 20 to do double damage wasn't originally a rule, but it pretty quickly got house ruled in because it led to memorable moments that people remembered. If a GM tried to force the impressive feeling for rolling a natural 20 by just saying every hit is a natural 20 critical hit it wouldn't extend that feeling, it would just be damage inflation and make natural 20s less memorable. So it's usually more about "leading the players to an awesome situation", which I posit is equally as pointless. The interesting and memorable nature of the event is its rarity and the difficulty in enabling it to take place. If you only ever run one campaign with one group of players, then maybe you can pull the wool over their eyes. I think its a jerk thing to do, but beyond that it could work. Unless you are gaming with very daft people, they will notice a pattern forming after a few campaigns. These awesome sword fights atop burning church rooftops will cease to be seen as special, and worse may make it so that if you don't put those things in it's seen as incomplete. They aren't special, they aren't memorable, they are expected.
This is where I may also seem like a bit of a hypocrite to people who play in my games, because awesome memorable things happen frequently. The distinction is, I don't plan for those events, I merely ensure the right pieces are in the sandbox so that they can happen. What do I mean by that?
One of the criticisms I get a lot for my d8/d6/d4 tables is that most of the table results are boring minor encounters. Three farmers in a field with a dilapidated hut they are investigating for boring local reasons. You can either just keep walking by or investigate and probably find out one of them lost a button, maybe you help them look any they thank you. A little vignette that took 10 real minutes of game time. The party will find similar encounters a dozen times. One of those times though, the dice will line up just right and the farmers will have cornered a vampire in the hut. If the players help they can easily defeat a vampire (or perhaps capture it or who knows what else they may want to do with it). If they don't it will likely destroy the hapless farmers and emerge at night to terrorize the nearest people without a home to protect them from the vampire (likely the PC's camp). That is more memorable.
You get more and more things like that in different tables. There are "suspiciously narrow passes" , scouting parties that call large forces to the next encounter, flying enemies, and random caves. On their own, no piece is really that much more than of minor interest, but when you randomly combine them then eventually a perfect storm will form.
This is increased in chance BECAUSE most encounters are only somewhat interesting. If every cave is full of some horrid hellbeast then no one is going to camp in one. But if most are simply caves, then the party will be perfectly willing to camp in the one with a secret door that leads to the Goblin Kingdom. If you want animate statues to surprise a party, then most statues have to be inanimate so that the sight of a statue doesn't immediately bring in the use of a sledgehammer "just in case".
Because most statues were just blocks of stone, a party used one to tie their ropes off and now has had a memorable encounter where half the party are trying to climb back up a rope over a chasm while a statue is trying to untie them and being interrupted by the other party members who have to both not die, but also not kill the statue before the party climbs back up. That is memorable, and it required no planning. It occurred because all the pieces were there in the game, swirling about, eventually they formed something interesting. Memory has a survivor bias in terms of what is retained, so be sure to use it to your advantage as a GM.
You can combine this survivor bias with the mind's natural inclination to make a story using the simple trick of having a variable game length. If a game is between 3.5 and 4.5 hours long you'll be able to hit a "natural endpoint" to any particular chapter in the story. Neoclassical Geek Revival was designed to lean into this with its awesomeness mechanic, specifically that at the end of the session when XP is awarded everyone reiterates what they did that they felt was awesome. This helps reinforce it within memory. At no point is there any planned story, or plot, or set pieces yet the games I run are full of memorable stories with impressive scenes that get retold by using these simple tools.