How to Run a Sandbox Adventure

If it’s not already obvious by now, a lot of my articles are inspired by Reddit. The latest example is a discussion about GMs finding sandbox games hard to run relative to the trad style of plot-driven adventure or campaign. Which is really unfortunate because sandboxes should, in fact, be easier to run while offering far more player agency.

Justin Alexander illustrates this point in his 2009 article “Don’t Prep Plots.” In the article, he outlines the necessary steps to prep both types:

For Plots:

1. The PCs pursue the villains. (What if they don’t?)
2. The PCs have to choose to follow them by ship. (What if they decide to ride down the coast? Or teleport?)
3. The PCs have to spot the derelict. (What if they roll poorly on their Perception check?)
4. The PCs have to board the derelict. (What if they just sail past it?)
5. The PCs have to rescue the survivor. (What if they fail? Or choose to flee before realizing the survivor is there?)
6. The PCs have to question the survivor. (What if they decide not to pressure an injured man?)
7. The PCs have to go to the central sanctuary of the temple.
8. The assassination attempt on the PCs has to play out in a very specific way.

Vs.

1. The PCs have to pursue the villains. (This is the hook into the entire scenario. It’s a potential failure point shared by all scenarios. If the PCs aren’t interested in going to the red dragon’s lair, it doesn’t matter how you prep the lair.)
2. You need to design the city of Tharsis. (Where is it? What’s it like? What can the PCs do there? Et cetera.)
3. You need to design the derelict ship.
4. You need to design the Temple of Olympus.
5. You need to stat up the Tharsis navy, the villains, and (possibly) the survivor.
6. There needs to be a way for the PCs to know the villains are hiding out in the Temple of Olympus. (In the plot-based design, this is one of the failure points: They either question the survivor or they have no way of knowing where to go next. In situation-based design, you would use the Three Clue Rule and figure out two additional methods by which the PCs could reach this conclusion. This can be as simple as making a Gather Information check in Tharsis and/or questioning the captain/crew of the ship the villains took.)

In the first scenario, your design task is a series of railroaded encounters where any deviation is a fail-state requiring you as the GM to either say “No” and steer the players back on the rails or improvise new plot points to get them back there. In the second example, where you outline situations, the players are free to maintain their free will until they ultimately arrive at the destination. The “dirty secret,” as Justin calls it, is that you are doing all the same work (usually far less), but the whole thing doesn’t fall apart if the players don’t choose to do everything as you envisioned it. His metaphor of giving the players a map and then expecting them to follow “a specific route has been marked with invisible ink.” 

If you haven’t read the article, you really should; every GM can benefit from it, no matter what style of game you run.

Soundtrack to this article: “HDK 163 † Beneath the Violet Kingdom” by Hex Crawl

Types of sandboxes.

With the discussion of why handled by Justin, let’s look at what forms a sandbox adventure can take. 

• Hex Crawl: This tends to be the type of adventure that people think of when they hear “sandbox.” Traditionally, in a hex crawl, players will explore a hex-gridded map (often hex-gridded) with wilderness, points of interest, and random encounters.
• Point Crawl: which is a network of key locations (points) connected by paths, allowing players to choose their destination from a list of possibilities. This is prepped similarly to a hex crawl, it just places a little more focus on the points of interest rather than the world surrounding them.
• Urban Environments: Set in towns or cities, these adventures still have points of interest, but usually reveal deeper information through interaction with NPCs.

But here is the funny thing about all of these adventure types: they use the same pieces and parts. No matter which you choose, you need to prep the same things. They are:

• Points of Interest (Cities, Towns, Dungeons, Other Locations)
• Random Encounters (Puzzles, Socializing, Battles)
• Rumors or Clues (The threads that point toward new locations)
• Situations (Problems that arise and are resolved with or without the player’s involvement)
• NPCs (The people or monsters the characters meet along the way)

How do you create/gather them? 

Well, I have a whole article about a process I have used to put together a whole campaign in a single day or weekend. Check out “The Overachiever’s Guide to Lazy Hexcrawl Building.” And, if you want some tips about how to manage it all, take a look at “The Overachiever’s Lazy GM Binder.” The same process applies to point-crawls, usually with less of a focus on random exploration outside of the points of interest.

How are they used?

Generally speaking, your players will start at a point of interest, usually a town or city. While they are there, interacting with NPCs (usually 8-10 that you have sprinkled throughout a small town), they learn of rumors that point to other unknown points of interest or situations that are taking place. Sometimes these situations will require the PCs to gather more information to figure out how to approach them. If I have an idea of a higher-level threat that the PCs might face off against, but lack the resources to do so, I might begin seeding subtle hints in the rumors until they are ready. For example, a nearby village might be inexplicably getting colder, hinting at frost giants moving into the area. Finally, random encounters are the texture that you layer on top to make the world feel alive. They are threats to make travelling between locations dangerous or complications that get in the way of time-sensitive tasks.

How many of them should I prep? 

When I begin a sandbox campaign, I usually have a point of interest fully fleshed out (the starting town), five or so other points of interest, that I know where they are, and just a simple sentence detailing what is there. These are usually major locations that people in the area would be familiar with, like major cities or large dungeons frequented by adventurers. Three to five situations that the characters may wish to explore, and roughly 10 rumors pointing toward those situations. I like to have around 2 to 3 rumors per situation, so I can make a table out of them and roll against it when interacting with an NPC. Finally, I like to have 8 or so NPCs, each with a location in a town, a personality, and “wants” outlined to guide a conversation with the PCs. 

A tip on NPC personalities: If playing an NPC gives you anxiety, don’t worry about accurately reflecting a personality; instead think of a character from a TV show or movie that you know well and write that down next to the character’s name. Then, impersonate that character when embodying that NPC. I find it makes it far easier to hold that character in your mind when representing them.

But I’m not good at improvising.

Get over it. You are always improvising as a GM. If you truly can’t do it, then you shouldn’t be the GM. Because you aren’t even playing a TTRPG at that point, you are playing a guessing game where frustration is the most likely outcome. But here is the thing: I have never actually met a person who can’t improvise.

Maybe you’re not great with voices, but guess what, that doesn’t really matter. If someone is upset that you can’t swap between voices for them, too bad; get lost. 

Maybe you have a hard time giving clues to the players; well, you can prep those. For more tips on that, I suggest you read Sly Flourish’s Return of the Lazy Dungeon Master. It’s single-handedly the best book on GMing I have ever read, and it’s 90% of my method.

Maybe you fear the players getting outside of the stuff you have prepped. Well, there is an easy answer to that as well. At the end of every session, you are going to ask your players where their characters are headed next, with the understanding that you will be prepping that material. If they are unable to give you an answer, you can remind them of the rumors they have heard recently, and they can make a decision from amongst those. And if they really can’t reach a consensus decision from that, I turn to the player who never speaks up and ask them. You can never really go wrong doing that.

But here is the truth at the end of all this: you don’t like improvising because you are avoiding it. If you have prepped for improv, you won’t be making things up (which is hard for anyone to do on the spot), you will be selecting the right pieces and parts you have already created for the right moment. You will then feel empowered, and that fear drifts away. Trust me, you can do it. 

Looking forward.

I realize I didn’t really cover how to run a sandbox city, and I would like to cover that in a new article in the near future. I am currently building out a DCC Lahnkmar campaign using the official materials, and I think it’s a great opportunity to document my process as I work on it. Hopefully, this article has helped you take the leap into sandbox-style play. I continue to evolve and hone the specifics of my process, but maybe now you can understand why I find alternative methods flat.