How to Design and Improvise NPCs and Groups of NPCs

Designing an NPC is generally simple enough in most cases. Often times, an NPC exists because a job needs done (tavernkeepers, guards, and apothecaries usually play more as business transactions than real people) or because the GM or a player has said something that would necessitate an NPC. In other words, an NPC exists in basically every case in order to fill a role. Knowing that is powerful: it allows us to design our NPCs to fill those roles as perfectly as possible. And most experienced GMs can tell you that designing a group of NPCs who know each other and especially work together--adventuring parties, gangs, groups of soldiers--is more complicated since the relationships between those NPCs will become a factor at some point most of the time. This guide is about how to design NPCs that fit your purposes while still allowing you to have some fun. 

Before we dig in entirely, I imagine there are a few surly GMs thinking to themselves, "Meh. I can improv NPC interactions just fine. There's no need to fully design an NPC." In some campaigns, this is true. I have long argued on this site that there are three overarching style of play in tabletop games. One is combat-focused--how can we strategically choose our actions to maximize our odds of survival? Another is more social in a comic sense--how can we at the table use the game as a way to make each other laugh and have a good time? The last in also more social, but in a more dramatic direction--how can we collectively tell a story involving our characters and the GM's story? For pure combat campaigns where NPCs are basically just vendors, fully designing a character is probably unnecessary, and the purest silliness campaigns may not always demand real NPC design; however, the final style basically requires NPC design, and all styles can benefit from it. 

The process is meant to be quick and simple so that you can do it on the fly as you GM. It's a few quick mental steps. First: what is the purpose of this NPC? Why do they exist? They may be an important character to the party of just a random person they bump into, but someone called them into existence, so identify that purpose. Second: what is an interesting way to fulfill that purpose? In the past, I've given small scale NPCs really colorful attitudes toward their purpose, such as having a vendor who is far too passionate about what they sell or a person who has a collection of related skills that will solve a problem as opposed to the direct approach. This is the first half of the process, and it's the harder half. Years from now, it's unlikely that your players will remember the name or exact details of every NPC, but adding something interesting to the NPC design can be memorable. 

I want to give some examples from my games in order to make this abstract idea more concrete. Let's first talk planned important characters. When I DMed my mystery campaign, I needed an inventor character who would feature prominently in the story. Her name was Lyssbetonk Cogswagon, and part of the campaign would take place in her lab, where her experiments were. Therefore, I also prepared her experiments to be prepared for the inevitable questions from my players. As I designed the experiments, I thought that they would have a huge effect on the city (a weather controller, for instance), and I needed a reason that Cogswagon would have been limited in using some of them to her full effect. In order to let her fulfill her role as inventor, I had created the new problem of needing to rationalize her failure despite her genius. So I chose to make Cogswagon a total isolationist, never leaving her house. This gave me an inventor who had an interesting tweak to how she pursued her role. 

A more improvised character was York the Orc. Many years before my silly Western campaign would run, my little brother played a character named York who was an orc in my first real campaign. And when I was DMing and my players demanded to have an interesting person come across them on the road, I needed an NPC to fill a role. I needed someone who would be entertaining, and I knew my group liked "random" humor and silliness, so I came up with a really sweet pie salesman who spends an inordinate amount of time making each pie who was out looking for only the best ingredients. His name was York the Orc, a silly name befitting this strange man. 

The next step of the process is basically the same as the first but applied to a different question. Third in our process: What is this NPC like? This can be broad if that suits your purposes, like if you're improvising them--they're surly, they're chipper, they're sleepy, what have you. Personally, if you're going broad, I recommend choosing two basic traits instead of just one--it's a small change, but it will really spice up the feeling of your characters. If they're meant to be planned characters, you'll want to get more specific. What are their values? What's their personality like? How do they spend their time? I follow the steps I use to develop player characters when I'm designing a really important NPC and I can go all out--that way, I have a real wealth of information and a solid sense of the character. The last step is this, and it's the trick that's gotten me countless compliments on my NPCs: why is the character like this?

What I mean by this is, there is the potential for a little story to be told with this NPC. There is a possible way for the game to play out where the players speak to this NPC for long enough to get a sense of who they are--many possible ways--but finding out what the "why" of that story is can be very rewarding. This is best explained via example, so let's return to Cogswagon and York. 

For step three, I needed to come up with personalities for them. Lyssbetonk was sweet and friendly but easily scared; she had virtuous aims with her inventions but was afraid to ask for help; she was calm and rational about everything but the outside world. Then I needed an explanation: she'd had an encounter with a burglar years ago and become a shut-in who feared the outside world. I liked this as a character design then, and I still like it. Cogswagon was a fun NPC, especially since she ended up being a suspect, a victim, exonerated, and helped, all while questions about her sanity brought doubt over her testimony and whether there had been a break-in. It was compelling stuff, and a lot of that came from taking the time to design Cogswagon beyond "kooky inventor."

York the Orc's personality emerged over time, as the party wanted silly York at all times and I was only able to slip in details every so often. But as York followed the party (they adopted him), his friendliness became part of a larger philosophy. Once I knew the party was keeping him around, I decided York carefully maximizes happiness. He acts consciously friendly to make other people happier; he's obsessed with the perfect pie so that he can make his customers happy; he does whatever he can to improve the lives of others. So ultimately the answer to the final question is that he follows the party around, abandoning his quest, because they are easy to make happy. He's at ease with them because they're always laughing and having a good time. And the deeper answer of all of it was that York had learned a bit of philosophy but not enough to know the downfall of his logic (York was himself not very happy). York could have been a strange little NPC on the road, but the players liked the basics of him, and he ended up becoming a major part of the campaign. 

It's true: this process takes time to do well. The answers sometimes take time to arrive at, and finding the time to prep is already a challenge, and improvising these NPCs with an aim at something your players will remember is even harder. That's why it takes practice and allowing yourself some grace when every NPC interaction isn't what you'd hoped. Here's my best advice with the process: give it a try a few times alone. Generate a few NPCs who have minor importance in your campaign, and try to quickly come up with answers to the process over about ten seconds. Give it another try later on. Try it during a session. Slowly work it into your GMing. The process works, and you don't always have to have answers right away--many of the details that will be remembered will be found in the moment. 

I've been using this process off and on for fifteen years, and after a while, it becomes second nature. As you make opening small talk with the player characters as some new NPC, you will already be thinking about ways to develop the NPC and coming up with interesting directions to take things in. And the beauty is that we are starting from a basic, common situation in tabletop games--an NPC is involved in the scene--figuring out how to achieve what the game situation calls for, and elevating the NPC to someone with real personality and depth (or at least the appearance of it). The process works for spinning quick NPCs to fill unremarkable role and for creating intense, complicated characters who can drive a story--the steps are the same, but the level of detail involved changes. 

So if that's how you design and improvise NPCs, it's tempting to think that designing a group of NPCs who all coexist is as simple as doing the process multiple times. In my experience, it can be if you're going for a characterization-light campaign such as you might when focusing on combat. However, to feel real and three-dimensional and compelling, groups should have dynamics between those involved. In my creative work, I've spent a lot of time and attention on building groups. Whether they're adventuring parties or gangs or soldiers or politicians or even merchants, there are bound to be tensions and rivalries and friendships between members of those groups, and it's often compelling for players to get to discover those dynamics. 

A bit about my groups before we talk about how to easily design a group. In my first campaign, I created many divided groups: the city was sectioned into distinct neighborhoods, the city and palace guards were independent of each other, and tensions about corruption created an us-them mentality. It's no surprise then that the group had a party split and the campaign ended in a civil war. My next campaign, an early D&D actual play podcast, prominently featuring a gang called the Keys to the City. Each member of the gang had been carefully designed, my first real foray into putting a group dynamic on center stage. There were some clunky moments, such as when a hot-tempered member of the gang couldn't coexist with some colorful player characters and abandoned the gang, or when the players didn't accept an ending I wrote for one member of the gang. But largely, it gave me a good sense of how to improve. I'd go on to develop a dozen narratively-related groups for later campaigns, and I put my skills to the test with Of Gods and Dragons, where essentially the whole campaign was the players negotiating with gods and dragons with existing relationships that stretched back for hundreds of years. That meant some really intensive work on how the characters all fit together. I'm very proud of how it came out, and I'm excited to tell you how I did it. 

The process is actually pretty similar to the NPC design process. What's the purpose of this group? What role do they fill in the world? What actions do they take as a group? What brings them together? This should be a fairly simple answer. They're bandits, or they're rival adventurers, or they're a council of politicians--whoever they are, it's important to remember what they're there for. Then we get into the fun stuff. How helpful to the players should this group generally be? It should make sense for the group (though there's always narrative wiggle room there), and it can also factor in your story goals. With an idea of how helpful a group will be, you can start to design individual attitudes. But before we move on, let's borrow an example from my novel series based on a D&D character I got to play. 

In the first novel, the main character, Asp, is a con artist trying to work with allies for the first time. Designing her gang, which I did using this NPC building process, was very important because they would be the main characters of the novel. So what was the gang's purpose? To live comfortably, to work infrequently, and to stay incognito. They merely meant to survive and grow ever more secure. And how helpful would they be? For narrative reasons, I wanted a pretty even split. And that made sense given the context--Asp was a promising but not perfect recruit, the gang was hard up for help on a looming job, and Asp was a con artist, which made the others slow to trust her. Knowing these two things unlocked pretty much the rest of the process. 

Next, we assign attitudes to individual members of the group. (If your group is so large that you don't have names or need for individual identities for all members, just work on about ten NPCs who the player characters will be in contact with.) This is where the general helpfulness of the group comes in. As newcomers to this group, your party will be greeted with some degree of friendliness and helpfulness (or tension and division) by the members of that group. But except in the most extreme of cases, you need variation. That's what makes the group seem like more than a bunch of copy-pasted NPCs. 

In general, there's a formula that works pretty well. I like to include a few outwardly friendly and helpful people, a few untrusting and distant people, and a fair amount of people who are somewhere in the middle as a base, with more helpful and less distant people in generally virtuous groups like a champion's guild and less helpful and more distant people in more dismal groups like an assassin's den. This may seem like a small detail, but in tabletop games, a helpful NPC and a non-helpful NPC are radically different experiences that can define a session or even a campaign, so knowing what you want ahead of time is really important. 

The next step is again similar to the first process: why do these individuals feel these ways about the party? Obviously, with lots of NPCs, this becomes more complicated. I recommend having at least fairly strong ideas of who these NPCs are using the process above first, but even a vague direction like "doesn't trust anybody because of trauma" is more than enough to go on. The trick here is to think about things from the group of NPCs' perspective. What does their organization feel about the party and their stated goals? Why might their organization help or hinder the party? What perspectives are dominant in this group?

In my novel, the initial introduction to the gang was rocky. A few kind members of the gang were supportive of a resourceful new member who could help get things done. A few skeptical members of the gang thought Asp was too unused to working with a group and couldn't see the bigger picture. A middle portion of the gang thought it was non-ideal, but ultimately, they didn't see much choice in the matter. And this became much of the narrative stakes of the novel--how could Asp convince everyone in the gang to approve of her, support her, treat her well? And that's the beauty of having varied NPC reactions to things--building relationships is rewarding, and changing minds is powerful. You can have that be a part of your campaign with well-designed groups. 

The last step is to ask how these NPCs, their opinions, and their existing relationships impact things. I know that sounds like a lot, and it honestly kind of is. But I promise it's worth it, and it's actually not that difficult. I start by looking at the obvious: outright support of the party versus outright opposition to the party. How do these people react to each others' starkly different responses? Are they surprised? Betrayed? Angry? Disappointed? Do the people with more moderate approaches look at those with extreme opinions as being overly emotional? Missing the point? Causing tension? Do those on the extremes see those in the middle as cowardly? Missing the point in their own way? Giving into stupidity? Any of these descriptors can be applied to members of a group, and the portrayal of the group now seems far larger, more varied, and more interesting than it would have been without these details. 

Another valuable question is about the relationships between group members. While in reality, the group only existed for as long as the players spoke to them, in the game, the group has existed for a while, maybe a long time. These people know each other. They are likely to be among friends, confidants, siblings-in-arms, lovers, respected colleagues, and found family. If two remarkably close people feel differently about the player characters, that is a meaningful moment. And the players seeing that they're straining relationships is powerful. If rivals in a group find themselves on the same side of the player issue, that's also an interesting moment. These don't all have to put on center stage or given serious time, but they're good small moments to add in order to give some color without dwelling on them (unless the players want to). 

One last way to think about these groups is reflecting on everything so far and asking: what do the members of the group think about the current situation of their group? In other words, how do they feel about the direction their group is heading? Is leadership popular? Has the group been successful lately? How might these impact the group's willingness to take risks versus play it safe? Or trust outsiders? Or lend a hand? The player characters might be right in front of them, but their organization is probably still their primary priority, so these ideas will help inform how group members act. 

And the best part: when all of the planning and prep work and brainstorming and good intentions still go awry, you can just fudge it. There's always a rationalization for why someone might do something--if you're clever enough to come up with it, it can be yours. In my novel, Asp's greatest rival in the gang supports her and her plans much to her surprise, acting as the swing vote that got her heist adopted by the gang as their next plan--he reasoned that she she had a plan with a payoff he was interested in, something I could back up with the fact that he was a very rational and calculating person. As long as the general average attitude remains the same, you can give pretty much any given member of the group any attitude you choose if you can back it up. 

You can go further and design the group in more formal terms--what codes and by-laws do they live by if any, what are their relationships to other groups and the community they live in, what is their history, why does the group exist in the first place--and I do recommend that. In my series on clans in my homebrew setting, I did just that for all of the cultural groups in my world: Daltoners, Faninites, dwarves, orcs, elves, half-elves, gnomes, and a wildcard guide to halfling social values since they reject all formal organization. But you don't have to go that in depth, and if you're improvising, it's probably not realistic or necessary to quickly devise an entire code of ethics for the champion's guild your players are headed towards. But knowing the dynamics of the group will be enough to play the scene out in an interesting way. 

It's something I've observed often that things in tabletop games get filled in the more time players spend with them. NPCs who are around often are usually the most complex and well-explored characters outside of the player characters themselves. Locations that are visited often are more vividly imagined than scenes seen only once. Story elements, themes, even jokes that come back become richer over time with the repeated attention. It's important to remember that if you need a group of eight politicians on a council because your players just arbitrarily decided to visit city hall, you don't have to improvise biographies for the council members and the blueprints to the building. 

For real improvising emergencies like this city council situation, I quickly hatch three main speaking NPCs. One is helpful to the players, one is suspicious of them, and one is more moderate for whatever reason is convenient. The other NPCs in the group add colorful details in support of one of these three speakers. I mention the repeated attention on things enriching them because the truth is, if you charge in with three broad ideas of group members (the four quick steps at the beginning), then plug those NPCs into the second process's framework for a group, and you can play out a scene with that much, and it will get more detailed and interesting as you go. Should you ever return, you will have the foundation you laid before to build on. Every second spent playing out time with your NPCs or groups of NPCs is time that further defines them, and that's incredibly powerful. 

Designing characters is one of the most fun parts of tabletop games. Some of us imagine strange and interesting people we can pretend to be for a while. Some of us like to pore over the rules to build a character who can best any challenge. Some of us brainstorm hilarious backstories and silly names to make our friends laugh. In my first years as a D&D player, I would endlessly design characters I would never play. Some of my friends still engage with tabletop games that way. It's fun. And I'm not going to tell you how you should have your fun. But if you struggle with designing NPCs who can engage with your story or with improvising NPC interactions, these tips can help you get past those challenges and focus on what you do find fun. That is the point, after all--we're all here to have fun. So get comfortable with making NPCs with these steps, and they will become so quick as to be automatic; then you can offer an interesting experience without losing your mind. 

That's all for now. Coming soon: what exactly a tabletop game is, how to balance kindness and antagonism as a GM, and an argument for downtime. Until next time, happy gaming!

How to React When Your Players Think of Something You Didn't

You've put in all the legwork. You made a campaign idea into a campaign; you might have developed a homebrew setting; you've worked on the setting for the campaign; you've populated the world with characters. You're playing, and your story ideas are going well. And then, suddenly, it all goes awry: the players come up with something you didn't anticipate, and now everything is threatened. What do you do?

I have faced this situation hundreds of times on a scale small to large. My favorite solution is to improvise. But it's a very specific kind of improvisation in these situations, and I want to devote a full guide to this exact situation. In order to show it in action, I'll share some times that I was foiled by players but refused to concede and kept the action going in fun ways. 

The first example is a modest one. It comes from my mystery campaign. In the very first session, my players were tasked with investigating a murder. They went to the murder scene and found the body there. Montana, a bard, immediately cast "Speak with Dead" on the body. I hadn't really anticipated this. But I had already decided that the murder victim was mute, and so I ruled that she couldn't respond because she couldn't speak. But then Montana had another idea. He cast "Speak with Plants" on a dandelion in the alleyway where the victim was found. Derek the Dandelion gave a perfect description of the murderer. It was only fair. But I improvised; testimony from a dandelion would not pass legal muster, and so they would need to get another witness. 

In this small moment, I had to juggle a few considerations. First, I had game balance to contend with. If the players solved the crime in the first scene, the whole campaign would be over before it began. I needed a way to change the situation (or our perspective of the situation) to allow the game to continue. My goal was to both reward the good idea and prevent the story from being compromised. 

But I also had to consider fairness. If a GM is unfair, the fun of the experience disappears quickly. So I needed a really good reason that things would not work exactly as intended. Holding that Derek wouldn't completely satisfy the witness requirement was lightened by the fact that the questgiver had explicitly asked for two witnesses to begin with. The important thing, though, was that my players feel that "Speak with Plants" had been a victory even if I couldn't give them the outcome they imagined would come from such a good idea. 

And finally, I needed something that was interesting. When I had decided that the dandelion tactic would work, I leaned into it. I had the dandelion be a colorful character, a young dandelion with little idea of his place in the world. The players loved him, and they named him Derek. Derek ended up being featured in our final session and epilogue as a beloved part of the campaign. 

Of course, this is the kind of speedbump that we often have to deal with as GMs. But sometimes a situation might have larger implications. Long ago now, I was running my first serious campaign as a DM. The campaign centered around the party completing various assassination contracts, eventually taking out the entire royal government including king, queen, and princess. The big twist was to be that the contracts came from the princess herself, who was committed enough to her plan to overthrow the kingdom to lose her own life in the process. As it turned out, the campaign didn't really go like that. 

My list of contracts was 20 targets. They completed one, kidnapped the second, and then turned on the questgiver before proceeding any further. The party split into pro-kingdom and anti-kingdom factions. Things got complicated, especially when one player on the anti-kingdom side broke into the palace late one night and assassinated the king. A pro-kingdom player procured a high healer, who cast "Resurrection" on the king. This created a huge issue: it placed the outcome of perhaps the entire campaign on a moment that only one player initiated. 

I solved it by thinking about the above considerations. I needed to think about balance. How would this affect the game? It either meant the anti-kingdom team had effectively won, or that their efforts to kill the king were so easily undone as to be meaningless. This is incredibly delicate. I needed something that would be a good compromise between the two extremes. 

I also had to consider fairness. The player who had assassinated the king had obtained an invisibility spell, passed dozens of armed guards, passed half a dozen stealth checks, and rolled a critical strike on the king. To deny success after all of this would be very unfair to that player. On the other hand, the player who got the cleric was using common sense to solve the problem, and of course the king would have a high healer around for just such occasions if he had already increased guard patrols. I couldn't deny any of this either. Again, I needed a compromise. 

The true solution came when I thought about the interesting angle. I needed some compromise that would be memorable and make sense. When I first realized the gravity of the situation, I asked my players for a minute to think things over. I highly recommend you do this any time you think prolonged thought would be necessary to determine a situation's outcome. In the minute or so that I took to respond, I ended up with this: "Resurrection" works, but only for a few days, and it can't be cast on the same person again. This allowed the pro-kingdom party time to stabilize new leadership under the king and to have meaningfully responded to the assassination, and it also gave the anti-kingdom group a big advantage, but one which wouldn't be complete or immediate. I asked the group if they all agreed that my ruling was fair, and they all agreed. This is another thing I recommend you do with big decisions. A simple "Does that seem fair to everyone?" goes a long way. 

But of course, these examples are specific to the situations in which they occurred. Someone with an almost identical problem could need to make entirely different calculations. I know from experience. 

About ten years after I had to deal with "Resurrection" as a surprise in the kingdom-toppling campaign, I was running the mystery campaign I referred to above. At one point, I had planned for a main character in the campaign to be murdered. This character, an inventor, would die before completing an invention which could solve a huge problem in the city and potentially solve the mystery. I had planned for the players to complete the invention themselves from puzzle-laden clues left behind by the inventor. That's not how it went, and for a familiar reason. 

Montana cast "Raise Dead" on the inventor. In my defense, most of my experience in D&D came from 3.5. In 3.5, you had to be a high level cleric if you wanted to resurrect someone. Those rules have since eased up a bit. In 5E, a 6th level bard can cast "Raise Dead." I thought I had accounted for this, but I hadn't. Once again, I had a dead NPC with a resurrection spell making narrative wrinkles. 

But this time, it was different. I wasn't dealing with a split party. This meant that the balance of the situation was measured differently. I'm okay with the party occasionally having a boon that wasn't as earned as others. I can accept letting the party, using a creative and meaningful solution to a big problem, make a step bigger than I planned without compromising the game. It is, after all, only cutting out a few puzzles. 

But there was an associated problem. In the act of finding the inventor dead, the party had encountered the one of the archvillains of the campaign: the murderer they had been seeking since session one. I had planned for the murderer to acrobatically escape the laboratory and have a real showdown later on. But the party used spells to bind the murderer in place. I rolled for her resistance, and it was pretty low. I had to either fudge the roll in the murderer's favor (which would take agency away from the players) or accept the roll and lose my dramatic showdown. For fairness' sake, I needed to lean towards accepting the roll. 

And I needed something interesting. I remember that as those moments unfolded and it became clear that Montana would raise the inventor, I was searching for something dramatic to add. Some flair that was missing from the situation. But then I remember surveying the faces of my players. They were riveted. They couldn't wait to know whether the inventor would be alright. And I realized that I didn't need to add anything. They had stopped a robbery, captured their antagonist, and found an ally dead. I didn't need to spice the situation up. Because it honored all three considerations, I had the inventor revive and complete the invention. It proved to be one of the more memorable moments in the campaign. 

So let's review the considerations and what they can do to help us come up with the best solution to unforeseen problems:

• Balance: Tabletop games are carefully calibrated for a specific distribution of powers and weaknesses to keep playing it rewarding and meaningfully predictable in terms of outcomes. When we try to resolve a situation that impacts the balance of the game, it's important to consider every consequence of a decision. This can require time to fully think about. Make use of time between sessions, and don't be afraid to ask your players for time to consider. 
• Fairness: Your players want and need to feel like there's a real fairness to the game. If things are too easy or too hard, the players suffer. Ask yourself about whether or not your decision will inordinately punish or reward a player for something, and avoid situations where players are pitted against each other. 
• Interest: The situations we're considering here tend to have a certain amount of contention around them. You are, after all, negotiating between the player's creativity and your own oversight in planning; to have them leave the situation happy, we need to include something enjoyable. But as noted above with the inventor/murderer situation, sometimes all a situation needs is already present. Strive for a satisfying feeling if you can. 

There you have it: my best advice (from experience) for dealing with the unforeseen player actions which inevitably pop up when you play for long enough. I hope that these examples serve to illustrate the ideas here, and I wish you the best of luck in responding to surprises. 

That's all for now. Coming soon: how to deal with DM mistakes, a branching-path mystery one-shot, and a free spell slot tracker. Until next time, happy gaming!

Back to the homepage (where you can find everything!)