For the past several encounters, I felt like my party has just wiped the enemies without issue. Taking inspiration from a one-shot I read, I had a new encounter put together that seemed to me like it would challenge the party well. And it did....to well. Most of my party is worried about a TPK and to be honest it seems to be within the realm of possibility. 

I have an idea on how the players on their own could take actions to make things alot easier on themselves, but it does require them to take a specific course of action. Otherwise, I would like to find a way to make the encounter more manageable without employing some sort of Deus Ex Machina. 

What are some ways I can secretly rebalance the encounter in the party's favor without making the encounter feel cheap. I obviously cant just reduce all the enemies to 1 HP, but perhaps a slight reduction in health could work. 

We ended the session in the middle of the fight because it was getting late, so any suggestions would be appreciated

Lower enemy hp and/or ac and/or damage. 
AC might be too obvious, if they start realizing that a 14 didn’t hit last week but now it does. So I’d go with hp/damage (you don’t have to make it 1hp, just low enough that they can actually win).

Or if you have a story reason for the King’s personal guards to be running past, or for there to be some other sort of ally in the area who can come help, that can work. If you weave it in well enough, it won’t be obvious what you’re doing. Such as they’ve arrived because they’ve been looking for the party because the king needs them for something, rather than they beat up the bad guy and keep moving on their way  

Fudge die rolls.

Or something scarier appears behind the PCs. The bad guys run, and they do , too. This can be risky since players gate to run from fights, so you need to make it clear that they are not going to beat the tarrasque today no matter what, but they might be able to get away from it. 

So, thing is... there's no way to rebalance the encounter without making it feel "cheap", because it would be cheap. Basically, you would as DM be making the decision that your players can't die, and if they ever seem like they might, you'd rescue them with a rebalance. What if you reduce the HP by a little bit... ...but then they roll poorly the first round, and don't hit, and the enemies roll well. Do you reduce the enemy's health even more, to put it within winnable range again? How far do you go with this? If you lower the difficulty by enough that you're pretty sure the players would win, it devalues all the fighting they just did over the last session - because it would be absolutely true that no matter what they did or didn't do last time, you would make the fight for this time be winnable but not too easy.

I've been down this path. If you give them a too big challenge and it looks like they're losing, you make it easier, if it seems they're succeeding too easily, then you make it harder... ...and then you realize that it feels like it *doesn't matter* what the players do, they always advance to the next story beat, always spending a bit of resources to advance but not too much.

This is a tough situation to get out of. You don't want to TPK the players because you messed up the challenge level that you threw at them, but you also don't want to create a situation where the players can't fail.

One alternate way out, without making this cheap, is a more meta approach. You can tell your players, more explicitly, straightforwardly - "hey, this fight looks like it's not going your way, and there's a good chance you're all going to die here. Your characters are be able to see this (describe how the what the characters see makes this clear to all of them). You might want to start strategizing about how you can escape alive rather than how to win, unless you've got some idea for how to radically change the nature of this fight."

This still gives them control over what happens here. Hopefully, once the battle is reframed from "fight to the death, TPK or win" to "how can we survive", with some explicit DM encouragement to reframe the win condition as "escape" the players should be able to find a way to escape - hopefully this fight isn't so hopeless that they can't even get away. (Or maybe they'll come up with something clever to beat the overwhelming odds! But they'll at least try this knowing they're facing overwhelming odds and knowing death is really, actually on the table for real if they mess up.)

Or you can just fudge the rest of the battle. Find an excuse for some of the enemies to run away, or have a third party join the fray (surprise! the fight drew a giant cave monster which is now fighting the players' enemies too) or maybe if the players don't know the HP of the enemies, make it so that all the remaining ones are weakish mooks which take just a hit or two to kill. Those are all options. You could do it. But having done this sort of thing before, in my opinion it does not lead to a long fulfilling DMing experience.

The basic problem with running challenging combats is that eventually the PCs won't be a match for the challenge (the dice might not be with them, they might be stupid, you might have just overestimated them). My basic advice is: rather than fudging, just make sure you consider "what happens if the PCs lose?". Not all enemies want to kill the PCs; for example, this is a perfect time to bring out the "You've been imprisoned now you have to figure out how to break your way out" adventure.

I kind of get annoyed when parties start talking about it being a TPK, or having back up characters if it all goes south. It means they aren’t engaged in the fight or do they even care about surviving it.

i have had fights where the enemy gets one good hit in and the party start that talk and I find it really frustrating. Partly due to how I build my campaigns as reactive to the players actions and character motives, it’s easy to feel like I am more invested in the characters survival than the players.

One thing to remember when running a combat scenario is that not all creatures will fight to the death. Even if a battle is going in their favor, any even mildly-intelligent creature will be continuously weighing its mission against is instinct for self-preservation. Sure, there are plenty of monsters that have no preservation instinct, and there are some enemies that are so completely committed to their task that martyrdom may actually be their goal. But most creatures (both monsters and humanoids alike) have at least some basic instinct of self-preservation. So once the enemy group is down to 50% total health, or once the party has killed at least 50% of the enemies, I would keep in mind that at the end of each round one enemy that has been injured may decide to flee. 

And while this may seem to be tipping the scales of fate in favor of the party, it's not. The one injured enemy who flees the fight doesn't just vanish from the planet. They're going back to their base to inform the BBEG of what happened and where the players are and what their capabilities are. So while the players might wipe their brow and be thankful that one or two enemies fled, whatever brief respite the party gained in that fight will make their next fight against the enemy forces that much harder. But hopefully by then the party will also have learned some new tricks and gotten a bit stronger. 

There's a cool website called The Monsters Know What They're Doing (themonstersknow.com). It's got some nice tips on running various monsters as more than simply "hit point attrition contests". 

There are a number of approaches depending on the monsters involved.

1) Are these intelligent creatures? What is motivating them? It can be satisfying for the players to see intelligent monsters decide to break and run if there is doubt that they can win. If the opponents have taken 1/2 losses or more and there is some possible escape route, some of them may decide to run.

2) If these are typically chaotic evil creatures they don't necessarily work well together (lawful evil like kobolds on the other hand often do). A group of chaotic or neutral creatures is much less likely to have one or two stand in front and die for the rest. Why should they? This kind of reasoning can sometimes lead to the opponents taking less than optimal actions in an attempt to preserve themselves over the group.

3) You could introduce an event happening simultaneously out of view of the characters that draws off some of the defenders (slave uprising, rebellion, escaped monster - maybe they have a captured basilisk that gets free in the confusion - attacks the bad guys from behind and then will move on to the party ... etc). The actual event depends on the circumstances but there is usually something that you can reasonably introduce to break up a combat that looks like it is going south as long as you don't leave it too late.

4) Capture. Although it should only be used rarely, in a lot of cases, captives can be worth more than dead adventurers. Ransom, information, revenge or even live food can be reasons that the bad guys want to keep them alive despite what they may have done already.

5) Adjust the actions the monsters choose to take. There are sub optimal choices that can seem reasonable and do less or no damage. A creature deciding to use their attack to grapple for example (I used this in an encounter a party I was running had with some vampire spawn. The party was blocking the doorway so the vampire grappled the character to move them out of the way ... the attack did no damage, delayed the combat by a round, and allowed the monsters into the room so that they could attack more player characters and spread the damage around ... in this case, I used the tactic to make the combat more interesting but it can easily be used to make it a bit easier). Another example is using the help action, this removes one creatures attacks while only giving one of the creatures advantage on one attack. Don't use it every round but it is one way to help reduce average damage.

6) Reduction in monster hit points assuming you don't tell the players up front. Monsters have a range of possible hit points typically defined by their hit dice. Many folks just use the average but a DM is free to increase or decrease as they see fit. There is no problem with monsters having different numbers of hit points. I tend to not adjust AC though since the players will often have a reasonable idea of what the number is from the combat or previous encounters and it can become obvious when a DM makes a change here. Another option though is "forgetting" to use monster special attacks or abilities - or get them out of the way and "forget" to recharge them.

All great so far

For context, the encounter is against a group of bandits** whose leader made a deal with a demon (so a mix of fighters supplemented with some fiendish creatures). 

From what I have read so far, it seems like the most "realistic" approach would be either reducing enemy HP some or having some of the bandit members start to scatter as they see the battle start to turn against them. 

If the party can manage to kill the leader (who is now on the field), I could realistically just have all of the non-fiendish fighters run for the hills (which is honestly something the party already thinks could happen). This would be the optimal course of action for the party to win the encounter. 

**the party is high enough level that I have replaced the bandits with more skilled fighters that have been reflavored and modified some. 

Change the remaining HP is the easiest, however even bigger is the idea of dumb monsters. Stop playing the monsters with tactics, have them choose less efficient attacks, or not recharge abilities. Have them dash past the fighter to go for the wizard just to force an opportunity attack. You could also fudge your rolls if you wish to hit slightly less often or not crit when you would normally. 

I mean the players don't know how many HP the monsters have so, you could adjust that and nobody will know.

But I think David42's methods are all superior to that.

It depends a bit on the fight, but for challenging, epic battles multi-stage fights can be both memorable and allow you to quickly rebalance it without feeling cheap.

Let's say your group is fighting against several heavily armored knights and they just can't hit them due to their high AC.

If a knight is sufficiently "wounded" (e.g. 70% HP), they throw their shield away and start 2h their longsword, foaming with rage and eager to kill the players as fast as possible. This increases their average damage by 1 point, but can decrease their AC to a level that makes them vulnerable to your players without it feeling cheap. Feel free to also give them reckless attack... increasing their potential damage while also significantly reducing their survivability. Something like this changes the pace of the fight a lot. The party's paladin could now be focused on keeping the "not-enraged" knights busy while the monk mocks the raging knight and kites him around and the ranger and mage quickly dispose of the angry knight when he's far enough away.

Or maybe they are fighting against a dragon. And during the fight they hear a squeaking sound from behind. When they turn around there is a very young dragon, maybe one of the CR 1 wyrmlings. Now suddenly the dragon can't use its breath weapon without risking to kill its own child in the process and the party can try to stay between the adult dragon and the child. Or leverage the child and use it as "hostage" while they retreat to come back another day.

Or have that same dragon in its lair. The dragon pulls back and breathes fire at the party (make sure not to hit everyone). Caused by the heat, rocks start falling from the ceiling and the group has to make DEX saves to avoid them. Be sure to narrate that the dragon is also hit by some rocks and tries to avoid them. If one of the players asks: yes, the ceiling is covered in stalactites that don't look too stable... and maybe that magic missile spell from the wizard suddenly seems much less useless.

Maybe the encounter is just normal enemies but there are just too many of them? The enrage from the knight works well for all kinds of humanoids. Bandits with low morale (and perhaps Goblinoids or similar) might become more interested in looting their dead companions than in killing the players. That could remove one or two selfish enemies from the combat for a couple of turns. 

Quote from Lyxen >>

The other suggestions are really cool, I just wanted to go along another tack, which is that apparently, it's recently that the encounters have started to be too easy. My take on this, as seen in other threads is that the encounter difficulty builder starts to be broken after a few levels if your players minmax at least a little, and especially if you are not doing the 6-8 encounters per day. As a result, for example now at level 6-8, I always build the encounters between deadly and impossible, and still the PCs usually overcome them without too much difficulty. And after that, unless the circumstances are dire, they will take a short rest or even a long one (although I try to restrict these, it's not always logical) if they feel that they are being winded.

So my advice would be to use tougher encounters using the builders, but to tailor the number of encounters and the rest capabilities more than fidgeting about the encounters themselves, because, indeed, the strings that you are using are much more obvious if you start adjusting during an encounter.

I totally second this. If your players aren't dealing with a good number of encounters per day, they're going to steamroll everything with ease, no matter how hard it seems, because they can burn all their abilities in one or two fights. And if they're suddenly worried about a TPK...that's not necessarily a bad thing. A bit of challenge and excitement is good, and I've almost always found that characters with their backs to the wall are tougher than either I or the players think they are. Plus, remind the players that there's no shame in running away to fight another day.

So I guess what I'm saying is that a possible TPK might not be such a bad thing once in a while! It's up to you as the DM though.

If the leader is new to the fight, he could see his minions needing help, fly into a rage and transform into something horrific. He could grab a bandit (an untouched or tough one) and eat him making him "pay for his failure". This will cause the other bandits to flee, leaving the party fewer opponents to deal with. It also gives the party the idea that maybe they should run away. 

Facing a demon, a bandit or two might try to fight it along with the party giving them unexpected allies against the leader.

Quote from Kaboom979 >>

 

If the party can manage to kill the leader (who is now on the field), I could realistically just have all of the non-fiendish fighters run for the hills (which is honestly something the party already thinks could happen). This would be the optimal course of action for the party to win the encounter.

 That would definitely work. Load-bearing boss doesn't feel "cheap" to defeat, kill the leader and the followers run away makes perfect sense.

Quote from Lyxen >>

One mechanic from 4e that was investing but did not make it to 5e, was the "bloodied" one, where a monster could recharge some abilities or activate new ones when below half hit points. It made some sense, and I have used it now and then in 5e to great effect.

 Bloodied was a great mechanic, we still use it for at least cosmetic purposes (e.g. just saying that a given monster or character is/is not 'bloodied' is a handy way of describing hit points without being too specific).

Quote from Lyxen >>

We do exactly that as well, for hit points and descriptive purposes, I'm just adding things like recharge on some monsters when it feels cool.

 Sure, that was a neat feature of 4e. There's still one inexplicable leftover in 5e: the champion's Survivor talent kicks in at half hp.

Have a surprise ally come to the party's aid. If it's a character they previously helped in some way, then it feels earned and doesn't feel deus ex machina.

Go ahead and TPK them. But they all just fell unconscious, and then the BBEG imprisoned them with an incompetent, monologuing henchman. You can still take some of their stuff to make them regret their mistakes.

There is a sudden change in the environment (sunlight, moonlight, darkness, weather, anti-magic field, etc.) that creates a disadvantage for the monsters or advantage for the players.

A bigger monster comes and eats the monster, but the players are too small for it to be interested in.

The monster offers to let the party go if they do it a favor.

If the battle is going too easy, bring in a second wave or a true form. The players will assume you planned it all along.

Add a ticking clock so that they not only have to defeat the monsters, they have to defeat them in a certain number of rounds.

Turn the battle into a chase, with the monsters trying to get away. Only the least intelligent or most motivated of monsters will continue a losing battle.