Most of my early campaigns were a combination of what the other GMs like to call “sandbox” or were episodic. The old Dungeons & Dragons games were a dungeon or monster kill of the week — fun, but not narratively pleasing, and there was little reason or room for character growth. The characters were avatars for the players “being awesome.” Each game episode was discrete, with little connection to the next, save the characters involved, and often GM duties were swapped. Toward the end of the last D&D campaign of our high school period, I experimented with an actual story arc involving the stereotypical, Tolkeinesque fight to stop the evil baddie from destroying the world. It went well and is the only bit of the D&D gaming we did that I and the other players from the period remember because there were consequences to our actions, the characters changed (some died), and there was an actual plot to follow.

But that plot was improvised. It grew organically out of a bunch of episodes that were very loosely connected. The story arc wasn’t so much planned as I simply set down an ending — we would fight the evil god-thing (the name of which I forget, but let’s face it, it’s the devil/Sauron/Walking Dude/whateveryoucallit) and either win and become the new masters of the world or cork it spectacularly.

The nice thing about an improvisational approach to campaign design is that it allows for a lot of changes on the fly. You don’t have every detail planned out, and loosing a player for a week or two because of a car accident, or because they have a new job, isn’t as much of a blow as something more structured. the character development is a lot more freeform, instead of having specific incidents you’ve planned to lay on them to see what the players do and how they craft changes (if any) in their hero(ine.)

The downside is you’ve got a lot of tap-dancing to do. NPCs that aren’t fleshed out at all can catch the player’s imaginations. Now that barkeep that has no name needs one. And a backstory — if not then, soon since the players insist on using the guy as an intel source. You need to be ready to explain on the fly the politics of a town or region that you thought wouldn’t matter, as they were simply supposed to go kill the monster and steal the treasure (and find a clue for the next episode, perhaps) at a dungeon or what have you.

This is even more challenging for settings where you might need a bit more knowledge — say a Victorian period campaign in London. Hopefully you boned up on the city in the period, because in a sandboxed/improvisational campaign, the characters are going to wander off an force you to try and stay ahead of them.

I like the “sandbox” as a means to give the characters more flexibility in their actions, but I’ve found a lot of players wind up not taking the bait for missions you might have in mind and wander off to do not much but get in barfights. For me, creating a setting gives the GM more ability to create richer adventures, and have a few fall-back side missions to distract the characters from killing off your cool villain too quickly, or as a means to nudge them toward the big mission you have laid out.

A lot of my early gaming was episodic, as mentioned before: comic book games, spy games where the adventures could be described as “movies” or serials where, once again, the only connection was the characters. The story arcs evolved, if they were there at all. this changed when I decided to run a sci-fi campaign during my time at the Defense Language Institute that was one part Babylon 5, on part Varley’s Titan series. (If you haven’t read the latter, stop everything you’re doing and do so — it’s some of the most inventive and screwy worldbuilding in sci-fi.) There was to be a specific beginning, middle and end, and I had a timeline: 16 months to roll the whole thing before we were off to our next duty assignment.

There were what the TV types call “push episodes” that moved the story along, and others that were there for filler or to let the characters wander off and do a bit of exploring in the world created. It was fun and I managed to pull it off with a few weeks to spare. Everyone had a blast, including me. What I discovered is that having a much tighter plan for the story arc is much more fulfilling for the GM and the players, but it is — by definition — ore restrictive. You do’t have to railroad the characters, necessarily, but the nature of the narrative should propel them — generally — in the direction you want.

Ever since that campaign, I’ve become a planner. There’s usually a story arc and a string of push episodes that are interspersed with side missions and a bit of sandboxing to keep things interesting, but I have a much tighter eye on the prize: telling a story with the help of the players.

My espionage campaigns are the easiest to make this happen: you get orders from the powers-that-be that set you on your way. They define the parameters of success, what you can and can’t do legally, and the potential consequences for failure. There’s still plenty of room for the characters to decide how to make this happen, and they can go off the reservation to follow leads that crop up, leading to missions that are player, rather than GM-generated. Having a chain of command or hierarchy that the characters serve is one of the easiest (and less intrusively railroading) ways to get them to take on a mission and follow a plot without you smacking them over the head with the story plan. They’re just doing their jobs.

I used this technique on my next camapign, a straight Babylon 5 game, that also had the added difficulty that I wanted to hew to the show canon, but not have the characters directly involved in the show’s events. When using licensed material (pre created worlds with metaplots not necessarily connected to your players) this is the big trick. For the B5 game, I made them another front in the Shadow War, only vaguely connected to the space station and the show characters. It worked and was fun, and I managed to roll the campaign up in two years successfully — even though we lost a few of the players toward the end. I followed the same method for a Star Trek campaign, setting it between series to allow for a connection to them, but giving the characters room to be the heroes.

The toughest of these licensed universes to work in so far is Battlestar Galactica. There’s a pretty solid metaplot you should have going on if you are using the universe: the Cylons destroy the colonies…now what? In the first iteration of the game, I did the “second fleet” game — the players were on another vessel that survived the holocaust and were attempting to rescue folks from the colonies before they lit out for greener pastures. It roughly followed the show, with events showing up in the background, but they never found Galactica or Pegasus and were eventually going to set up in another tar system remarkably like the colonies, with multiple habitable worlds, so the grand story could begin again. (They were working off another “prophecy”, to stick with the semi-mythic feel of the first few seasons.) This go ’round, I’m doing a lot more before the attack, creating a more Cold War feel — the Cylons are still out there, the higher-ups know but are keeping the population in the dark for stability sake, and eventually, the character might or might not turn the attacks. Another option is they are the big heros, leading a rag-tag fleet.

As you can see, I have a general plan for the campaign, but at this time, it’s more improvisational, to allow the characters to really be the heroes. I’m using the “all of this has happened before” idea to allow them to be the Adamas and Apollos of this iteration of the story, but if they don’t stop the Cylons, there will be a general arc like the show: find Earth, save mankind.

I’ve found that the secret to a successful campaign is a balance between planning — not just the adventure plot at hand, but having a few NPCs in the back pocket for quick use, and a knowledge of your setting, even if it’s just broad strokes. No matter how well you plan, however, the players are going to go off script or will get distracted by a background NPC or a “clue” they thought they saw that was just some descriptive fluff for you. That’s when you need to be ready to bang together a quick caricature of an NPC, expand on the look of a building or the setting their in, or prepare to create a whole red herring or real clue trail for them to follow when they see the gaming equivalent of “SQUIRREL!”