So here's a question. What published previous-edition campaigns do this best? What are the classics that get it right? Which ones set up the best villain, have the villain most central, and keep the PCs involved with (or aware of) the villain all the way through?
It's not an easy thing to do, structurally speaking. The nature of levelling in the game very much works against it in a long campaign. If you have a campaign ultimate bad guy meet up with a low level party, he'll squash them like bugs if he notices them. If a campaign ultimate bad guy meets up with a medium level party, he probably WILL notice them and have sufficient reason to squash them. And PCs are resourceful critters, if they DO meet up with an ultimate bad guy early, they might just be able to kill him in some unexpected way and disable everything.
Normal way of doing this is having an ascending hierarchy of intermediate bad guys - in the classic DL series it's clear that Takhisis is the ultimate bad, but PCs have her cleric Verminaard to start with, then her armies at higher levels, then Ariakas and Takhisis herself at the top. Similarly, Savage Tide (while it does have other problems) does this pretty well, having Vanthus be a bad seed from level one, having him advance with the party up to the mid/high levels by which time the PCs realise he's just a tool of Demogorgon, the ultimate villain. But even then, in my playthrough the PCs got really intense about making sure Vanthus was dead dead dead and not coming back waaay before the modules had finished with him, and that was very awkward to manage.
What is the gold standard here?
I think you’ve hit the nail on the head. The standard is onion layer hierarchy to a central conspiracy with the PCs peeling back a layer at a time. Let’s take the answer to
@Distracted DM ‘s question as an example.
I often see success with villains, but there are great adventures that lack a central villain, or that lack one that the party is aware of/interacts with multiple times.
The Night Below is a big one- the party only becomes aware of villains of import a couple sessions before said leaders are defeated.
My lesson has really been to add more villains to games, and ensure that the party can interact with but not fight a number of times.
I don't put this lesson into practice as often as I should, though!
So I feel the complete opposite, Night Below is a case where there is an absolute villain who’s presence is felt throughout the campaign. I don’t the villain
must be interacted with directly for a great campaign but they do need to be felt. This can absolutely be done through agents. Tomb of Annihilation is another one.
From the disappearances that begin the adventure, to the orcs and their fishy control potions, to the illithid messengers, to the derro and Kuo Toa of the City of the Glass Pool, to Great Shaboath itself. The Grand Savant is behind the whole thing - with the assistance of Darlakanand. Incidentally Night below also includes free form exploration and some brilliant NPCs to interact meaningfully with which harkens to my OP.
Where Night Below struggles is that that the aboleth aren’t really given any individuality. They’re a bit like Ridley Scott’s aliens. There’s a queen but they’ll all pretty darn bad. The aboleth as a species certainly are brought to life though. It also struggles with the intermediate foes that
@GuyBoy mentioned. The bandit cleric Ranchefus is one, as is the King of the Glass Pool - but there is a hell of a lot of adventure in between these. If I was revising for 5e I would have more named villains that support that key aboleth plan and have them felt more within the campaign. Night below can become a slog and not all campaign finish (Matt Coleville’s for instance (and my own) my gut feeling is a lack of strong villains.
If being re-done I’d make the leader of the orcs far more well known - a famous raiding war chief with multiple atrocities to their name. I’d add in an Illithid fixer leading the patrols back and forth who would be quite slippery and meet and evade the players several times. I’d have the Kuo-Toa of the Glass Pool felt earlier on be suppressing all the other missing races - Drow, Duergar etc and have some key Aboleth personalities. Possibly one treacherous Aboleth who wants to see the grand plan fail.
One of Curse of Strahd’s geniuses is that the end villain does have reasons not to squash the PCs like bugs, and by the time he realizes he needs to it’s too late to do it easily. I do think 5e is very supportive of this (more so than any other edition) because of bounded accuracy and the likelihood of characters many levels lower lasting a couple of rounds at least. Whereas in 3e a single multiple target save or suck would be game over with no chance of reprieve.
Certainly in my mashup of Dragon Heist/Golden Vault those central four villains will be dancing in and out of the parties sphere of influence. My intention is that they each have a reason for not killing the PCs. Xanathar doesn’t care; Manshoon sees them as useful pawns; the Cassalanters want to recruit them, and Jarlaxle likes them. The art will be doing this while making the PCs still want to defeat them. Then again no-one said a great campaign was easy.