Mistakes were made
Natural 20's are one roll in every 20, or so. 5% chance. They happen fairly frequently. So why were your players excited about this one?
- It happened after the Hag's vessel failed a saving throw, leaving the villain confronted by the heroes in a conversational situation (a very common and much loved trope of the genres of fiction D&D games are drawn from)
- The sorcerer, who is charismatic, was the one who rolled the natural 20 on a charisma based check. In a situation where a charisma based check was entirely thematic and appropriate.
Many newer players also get excited over 'natural twenties' any time they happen for any reason. This isn't a bad thing. Getting excited over a game you're playing is a great sign that you're having fun.
In this scenario, where a lucky roll coincided with a specific situation where it should be relevant and that was already a tense scene, your players got excited because they expected these factors to lead to something interesting occurring. At least, that's what I surmise from information given - perhaps the rest of the party was just tapping on their phones etc and not interested at all and the sorcerer rolls 18 social checks at every mouse the party encounters along the way.
But the situation sounds dramatic. I'd guess players were interested. So that's the first mistake.
Not utilizing player interest. If players are interested in a situation, or its outcome, going 'uh, nothing happens' is a sure way to not only lose that interest but also make them less interested in the game as a whole. Even if you don't know why they are interested, try to make use of that interest in some way. Introduce a character, have something exciting happen, drop some lore, whatever. You have the attention of the group. Use it.
The second thing i'm more sure about, and that's
Refusing a dramatic offer. The sorcerer player gave you a dramatic offer, a narrative offer, a roleplaying offer, whatever you want to call it - they laid a premise, started a scene (via their actions) and then tossed the ball to you to continue it. Your continuation 'she doesn't seem intimidated' is lacking detail and offers no easy recourse to continue the scene. It's like shutting the door in the face of someone - they will, mostly, leave. If someone then climbs in a window that's pretty rare. Sometimes you should refuse player narrative offers - like when you don't have time, they've had too much spotlight, or the rest of the party isn't interested etc. But in this scenario? When the other party is a major villain, the whole party is there, and it's a classic type of scene that people love? Do something with it. Anything. Spit in the sorcerer's face, have the possessed body bite its tongue, whatever, just don't do 'nothing'.
Finally, there's;
Success or failure is binary. 'Either the hag gets 'intimidated' and then she is ??? defeated? something, or the hag doesn't and isn't'. That's the binary thinking. It goes further. 'If the player succeeds on the roll, good things happen, if not, bad things'. That's another binary thinkingism and they often both happen at the same time. Success always good, failure always bad, single state outcomes, dependent on die roll. That's a major mistake. It makes the game much, much more boring than otherwise. You can make lots of mistakes and still run a fun game, but this one sucks so much fun out of the game it is quite infamous.
Succeeding at a task can make the character's life harder. Climbing into a prison window for example, and then through the window... the character is now in a deep prison cell with no way out but a harder check. Hiding from the sounds they heard (that is their friend come to stealthily find them), success causes them to hide.. from their friend. Great.
Failing at a task can make a character's life easier. They want to Deceive the baron into thinking they aren't sympathetic to the orcs? They fail, and the baron, seeing through this sellsword's ruse, tells them that he too would like to negotiate a peace rather than kill off the orcs. Succeeding there would have left them without a powerful ally (who would think they were just more bloodthirsty racists, and not include them in his plans).
In this situation, you have all kinds of options. Success at intimidation could lead to all kinds of outcomes for the party.
the Hag calls in a favour and a Young Black Dragon is sent after the party to remove this threat to her wellbeing. She's scared! So she uses up that valuable favour. And now the party has 2 boss fights to contend with instead of 1. (although she might leg it if they turn out to be strong enough to kill a Dragon)
the Hag surrenders, claims she has some excuses for what she was doing, but repents and is now willing to help the party in exchange for mercy. Still a fight scene if they say no, but now they could maybe end up with a kooky potionmaker npc Hag they keep having to keep an eye on so she doesn't steal children
the Hag leaves the area. She's still a danger, still developing that mushroom plague or whatever, but now the party can't easily find her. She's a problem for later and will turn up somehow - ideally in some interesting way, like the actual big bad has enslaved her and that's where his poison fog doom timebomb thing came from. A 'your choices in the past created this future situation' thing.
the Hag gets very worried! She creates a plan to kill the PCs as soon as possible. Basically what you were going to do anyway but instead of cackling about how she is unkillable she looks terrified and yells 'die! die!' a lot and focuses her attacks on the sorcerer.
the Hag gets very worried! Instead of dismissing the PCs as impotent fools, she sets up a plan to poison them using children and a special mushrooom and a complex ruse involving stew.
There's lots of places you can go with it that aren't 'uh... you win' or 'nothing happens', and all of them are far more interesting than either of those options.
Finally, outcomes of checks being binary (yes or no) is bad thinking. You want degrees of success for all kinds of reasons, and you want different outcomes possible dependent on circumstances and also to create interest. If every time someone rolls Intimidate the opponent is either cowed and helpful or angry at you and nothing in between it's weird, awkward, and not interesting. If they show a range of reactions from servility to anger to confusion to 'yeah sure I guess, I don't want trouble' to changing their opinion of the character (like, positively) to assuming you're related to some other threatening force in their life (like a criminal organization) and sending the police after you, etc, it is both more interesting and more believable and helps extend the barebones 5e skill system.
Bad reasoning for having the skill automatically fail - This isn't exactly a 'mistake' per se, but your reasoning indicates that your game might lack verisimilitude - aka, making sense. Lack of verisimilitude is one of those major mistakes that tends to torpedo games.
This is due to your reason for having the intimidate check autofail - the Hag not being physically present. I can threaten someone over the phone. That's what is happening. The sorcerer is threatening the hag over the phone. It's still scary. 'I'll come over to your house and kill you' is scary. They don't need to be physically there for it to be scary. That's something that players would generally expect from a world - that threatening to come over to someone's house to kill them is scary. If that's not the case, and you haven't shown them why this person is unfazed by credible death threats, it seems off.
Lacking verisimilitude is brutal for any game, and you should generally go as far as you are able to try to make sure things in your game make sense to basic logic.
The party is going to go places, and as the DM it is your job to build the roads. Those don't have to be roads of success and glory and fame and fortune. They can be roads of bitter failure. But they have to exist. Players doing things should be part of the story -they are protagonists, their actions, choices, and even thoughts, feelings and attitudes - should be important. If players do things and it does not affect the story, you've dropped the ball as a DM.