So I recently put a pause on GMing tabletop sessions due to scheduling conflicts getting out of hand, but I wanted some opinions anyway. I have a homebrew campaign that I originally planned to be sandbox-ish but ended up needing to turn into a guided experience because I actually needed to come up with stuff ahead of time, and figured that would be the more fun approach anyway. I made a general outline to better realize goals the players, based on their decisions and characterization in past sessions, would most likely follow, but I don't plan too far ahead due to the fact that wild card syndrome can, will, and has botched whatever questline I come up with.

My question is, what degree of planning should entail a campaign? Should I come up with a list of generic scenarios that I can then contextualize within the story? How much railroading should I enforce in order to get things going without disrespecting my players' actions? What's the best way to create on-the-fly situations?

Seems like some good questions.

So on the fly scenarios, probably just take things one at a time and ask various questions to the scenario. Say a party member decides to steal something at a market, he gets caught and the stall member yells 'THIEF!'. Is it crowded? Are there guards nearby? How visible/flamboyant is the player, are they well known? Have they saved the town?

While those questions will come up quickly, you answer and resolve based on if they are going to go to jail, or get a slap on the wrist and insist they give back the item and they won't let them off the hook a second time. Or in some countries, Sharia law says if you steal you get your hand cut off... That's the other extreme. (Or death...). But i doubt it will often come to that.

So, how much planning is needed. I would say an hour of planning is mostly what you need. Make a table for a random chart for events that come up, quests that get put up, things they find while traveling, etc. And in encounters put a note of how many that particular encounter may have. Maybe 30 guards are going from one section to another transporting a rich merchant, or maybe they are being hunted. I'd leave 5 spots open for infamy scenarios, and if the characters don't have enemies then bounty hunters that are just passing through, hell maybe they can join the bounty hunt for a criminal!

Planning for dungeons and larger things depends on how far they can travel. Say in a 3 hour session you expect they could only travel, say, 50 miles. So if there's any huge things within 50 miles they might stumble on, you can have some generated maps. Or borrow maps and plans from other modules that are completely unrelated like bandit camps or cave systems, just repurpose them, find a spot for treasure, and spread out a bunch of enemies who are by default unaware, or maybe they are on alert for something unrelated (say a bear attack).

As for railroading... Depends on the world. I'm playing in a game of Chult (not sure which game) where when i joined the person who hired the party has died do to them taking 8 weeks to do side-quests... Unless the scenario demands they be railroaded or have a time limit, probably don't add it.

Heh, if you want a ton of suggestions, ideas and other things, might listen to Mr Ripper, TtheWriter, Neckbeardia and a few others who regale their stories or read ones posted on reddit.They are great for ideas as well as discussing the mechanics and failures/successes of a few things.

Worse comes to worse i'm sure we can find some tables you can use for quick random encounters and events.

A general outline should do well; have an idea of where you want the party to ultimately end up, but also be prepared for them to smell the roses or otherwise derail things a bit.

Depending on the game system, it can be more fun for them to trod off the beaten path without railroading them back on.

If you have a bunch of players who will do random "funny" things it is not as important but if you are playing with people who believe they are their character in real life then you need an ultra serious story to keep them from casting spells on you in real life.

Campaign? Very little. Maybe a few long-term options that can inform decisions as you go along.
Arcs? A bit more. You should definitely have the overarching theme, plot hits, and movers and shakers planned. But nothing needs to be concrete.
Session-to-sesion? This is where the most comes in. Especially in D&D-style games where you have to put in the effort of statblocks and whatnot. However, I still personally prefer the "outline" format. Intended bullet lists.

Since you're doing homebrew, you have the option of active and reactive nemeses. Give them their goals and motivations and have them act accordingly, changing as the PCs do things.

But homebrew is A LOT of work. I prefer working in [semi-]established settings, like Golarion (Pathfinder) or Dragon Empire (13th Age). Ancient history and deities -- that's already written. I can tell my own story in that world. The worlds have cornerstones (and in the case of Golarion, voluminous available materials if you desire to use it and make it canon) to ground everyone, but you have plenty of room to tell your own story.

Planning too much can wreck a game just as much as planning too little.

"Even the best planned tabletop campaign won't survive the first contact with the players."
based loosely on Helmuth Karl Bernhard Graf von Moltke :)

But you know that already.

I'd say your campaign's story arc (beginning to end) needs to be planned in detail, but you have to prepare for the imevitable detours, that your players will take.

In general, these detours add to the game. Your players will enjoy it more, due to them.

However - if it becomes clear to you as the GM, that the detours will steer the game in a direction, that will ruin your campaign...feel free to throw some serious adversaries in the way.

All done with a sense for proportion, of course (you probably don't want to kill your players, just to keep 'em from straying off the pre-planned path too far), but let your players understand, that they won't succeed, if they try to push further in a certain direction.
If that means you have to (seriously) hurt (some of) them - so be it.

A little leeway is always good/nice to have - but the common thread that you as the GM have set up, must stay visible at all times.

Design a map, at least a few 100 square miles, and fill every acre with locations and scenarios, including player origins and goals. That way, no matter where the players go, the DM can always have a plan. Of course, some will probably make a b-line for the nearest border without even seeing the map.

While that would certainly ensure there's something everywhere, a problem is the DM can burn out just doing world planning, and may take weeks to fill the map out.

What might be better is to draw a map, decide towns and roads between towns, put forests streams, major features, etc and then leave pretty much it all empty and decide on the fly and fill the map in as you go.

Sometimes a forest is just a forest... And sometimes there's hidden caches in the forest left behind by those passing through. Apparently Lewis and Clark on their tour through the US would bury meat and supplies and caches and would plan on returning if they ran out of their food, ensuring they wouldn't starve while in the wilderness. Might be something similar. But those types of things should be random, not planned out unless you're following the trail of someone who might have done that.

Hey thanks guys, I can't quite say that every reply is some new illuminating thing about being a tabletop GM, but it does give me confidence that I am on the right track. This is actually the first campaign I've GM'd, trying to playtest a homebrew ruleset due to... less than satisfying alternatives suited to my tastes. And the setting is entirely my creation as well, so I have to do triple duty on writing the rules, the universe, and actually having to play it.

So far, I find that I have to stretch my creative muscles each session not just with the overarching plot but actually having to improvise on the spot where things will go. The first session I ended it early with a cliffhanger so I can figure out what to do afterward. The second session went better, where I started with the idea of going through a bootcamp to test the combat and ended with the party being captured by an enemy faction. The third session I actually planned the climactic battle in detail, but I had to come up with the context for going into combat and then needing to completely scrap whatever I planned afterward due to the players basically botching it.

Mind that these sessions are two hours long and are usually months apart so I can have a lot of time to plan, but the problem became that maybe I shouldn't plan with too much detail on the occasion that what is planned will never actually come to fruition.

I typically start with a map that's vague as hell. A general shape, plonk the starting location in there somewhere and worry about the rest later. I might put a few places of interest in there as well but worry about scale later, so the first two towns I put there might end up ten or 100 miles apart depending on what direction the campaign is going.

As for planning a campaign, I really don't as such. The most I do is start with the beginning and a possible end and flesh out almost everything in between as we go, because any plans I make will inevitably end up either in the crapper or adjusted to changing requirements anyway. I do plan for each session, mind, but never two sessions ahead (unless the players are committed to a long-term questline, and not always even then). Besides that I do have a Word document full of random NPC's, locations, ideas, encounters, quests, stupid gags and puns, previously abandoned story arcs and the like that I can either use for planning sessions or throw at the players on an ad hoc basis if they end up taking a route 180 degrees off where I thought they should.

Sounds pretty good. Other things might do is a handful of political things that are going on in the background as the main 'plot'. Say if the plot will be to eventually overthrow the evil king or kill the demon lord or take down the big bad.

Now if the players aren't hindering them then maybe there's a new base every month or two and things get a little more dangerous, and more bandit crews out there when they used to be townsfolk just surviving, more monsters are spawning due to the spread of evil, etc... If they ignore it long enough, well they might find something big headed their way.

But other than a few things going on in the background wouldn't concentrate on it too much. If/when they want to be serious they will get serious.