To create factions in a fantasy world, start with what each group wants and how far it will go to get it, not with its heraldry or its robes. Give at least two factions goals that cannot both be satisfied, and conflict becomes structural instead of something you invent scene by scene. Then fracture each faction from the inside, keep the active count small enough to hold in your head, and record each group's aims and alliances somewhere searchable so they stay consistent across a long serial.
What actually makes a fantasy faction feel real?
A faction feels real when its goal makes sense from the inside, which means the very first thing to decide is not what a group looks like but what it wants and what it is willing to do to get it. New writers tend to build factions the other way around, starting with a color scheme, a cool name, and a one-word morality (the noble knights, the evil empire), and then wondering why those groups sit inert on the page. The banner is the last thing that matters. What matters is desire, method, and limit: what this group is reaching for, how it justifies the reaching, and the one line it will not cross even when crossing it would win. Get those three answers down and the culture, the symbols, and the hierarchy fall out of them almost on their own, because a sect that will torture initiates to accelerate their cultivation has already told you what its temples smell like.
My feeling about this is that the "internal logic" test is the whole game. If you cannot state a faction's goal in a sentence that a member would nod along to, the faction is decoration. Dune, from 1965, is the cleanest teaching text I know here, because House Atreides, House Harkonnen, the Bene Gesserit, the Spacing Guild, and the Fremen each want something concrete and mutually obstructive, and Frank Herbert never has to remind you who the villains are, because every group is the hero of its own arithmetic. Progression fantasy and cultivation inherit this directly. The sects in a xianxia setting are factions in their purest form, a web of groups each hoarding techniques and territory, and the genre works precisely because a reader understands that a sect guards its methods for the same reason a real institution guards a patent.
Start with what each faction wants, not its aesthetic
The reliable way to generate a faction is to write its goal before you write anything else, then derive the rest downward from that goal. Ask what this group is trying to secure (survival, a resource, revenge, legitimacy, the continuation of a bloodline, control of a magic system), then ask what its preferred method is, then ask what taboo constrains it. A merchant guild that wants a monopoly and pursues it through law and bribery rather than open war is a completely different story engine than a warband that wants the same monopoly and takes it by burning caravans, even though their goal is identical on paper. The method is where the personality lives. And the taboo, the thing a faction will not do, is what keeps it from being a cartoon, because a group with no limits reads as a weather system rather than a set of people making choices.
This is also where world-building and faction design stop being separate tasks. Once you have decided what your factions fight over, you have implicitly decided what is scarce and valuable in your world, which is most of the setting's economy and much of its magic. I wrote more about that downward flow in world-building for web fiction, and the short version is that factions are the fastest way to discover a setting, because you learn a world by learning who is fighting over it and why.
Two factions have to want incompatible things
Conflict becomes structural, rather than something you script scene by scene, the moment two of your factions want things that cannot both come true. This is the single most consequential decision in faction design, and it is the one most settings skip. A world can have a dozen richly imagined groups and still feel strangely peaceful because none of their goals actually collide, whereas a world with only three factions crackles if those three are locked in a genuine zero-sum bind. So when you sketch each faction's goal, immediately check it against the others and ask, can these two both get what they want? If the honest answer is yes, you have built neighbors, not antagonists, and you will spend the whole serial manufacturing reasons for them to fight.
A Game of Thrones, from 1996, is the textbook case, because the great houses are not evil, they simply want the same throne, and only one body can sit on it. That is the entire engine. In progression fantasy the incompatibility is often a resource on the critical path of power, so the sect that controls the only source of a rare pill and the rising protagonist who needs that pill to advance are in conflict by construction, no villainy required. This is the same principle that underlies good kingdom-building in web fiction, where a realm generates plot precisely because its expansion takes something a neighbor was counting on. Build the incompatibility first, and the scenes write themselves.
Fracture each faction from the inside
A faction should never be a monolith with a single opinion, because a bloc that agrees with itself is a wall, and walls do not generate drama. The move that turns a faction from set-dressing into a story engine is giving it an internal fault line: an ambitious second-in-command who thinks the leader is too cautious, a moderate wing that wants to negotiate while the hardliners want to march, a splinter that has quietly reinterpreted the founding doctrine into something the elders would call heresy. Now the faction can act against itself, betray from within, and change over time, and a reader can watch a group evolve rather than just win or lose. Worm, by the author known as Wildbow, does this relentlessly, because its parahuman factions are constantly cannibalising and reforming, and the tension inside a group is usually sharper than the tension between groups.
I think the internal fault line is also your best insurance against the most common web-serial failure, which is the antagonist faction that gets defeated and takes the story's momentum with it. If a faction is a monolith, killing its leader ends it. If it is fractured, the leader's fall promotes the ambitious lieutenant, and the conflict escalates instead of collapsing. This is really a special case of writing antagonists who survive contact with a strengthening protagonist, which I get into in how to write a web fiction antagonist; the faction-level version is to make sure every group has a bench, so defeating the starter is a promotion for something worse.
How many factions does a web serial actually need?
Fewer than most writers put in, and each introduced only when the story reaches it. A reader can comfortably track three or four active factions at once, and past that the setting turns into a list of names that all blur together, which is death for a serial where chapters arrive days apart and memory is already strained. Sprawling settings can feel populous without violating this, because the point of view does the filtering; A Song of Ice and Fire has dozens of houses on the map, but any given stretch narrows the reader's attention to the two or three whose collision is live right now. That is the trick worth stealing, which is not "have few factions" but "foreground few factions at a time" and let the rest wait offstage until their turn.
Running a web fiction platform, the pattern I see most in first drafts from otherwise strong authors is front-loading, dumping six faction names and their entire political history into an early chapter to prove the world is deep. It reads as noise, because the reader has no character to attach any of it to yet. Introduce a faction the way you would introduce a person, on the page, in a scene, doing something, and the reader will remember it. The rest of the political map can accrue quietly as the serial grows, which is also the honest way most long web serials are actually written, discovered arc by arc rather than planned entire.
Keeping your factions straight across a long serial
The practical problem, once your factions are good, is that a long web serial will contradict itself unless you keep the political map somewhere outside your own memory. Alliances shift, a rival becomes a reluctant ally, a leader dies and a splinter renames itself, and because serialized fiction is written over months or years, the odds that you will misremember which faction was aligned with whom in an arc you drafted last spring are roughly certain. A reader, meanwhile, is tracking those relationships closely, and the moment a sect you framed as an enemy in one arc is casually described as an old ally in the next, the trust that took a hundred chapters to build takes a hit.
This is exactly the kind of thing a wiki is built to carry, so on IlorisNovel each faction can live as an entry with its current goal, its named members, and its standing alliances, and the always-on relationship view shows how those groups actually connect rather than leaving it in your head. When a war ends or a leader falls, you update the entry, and an @mention ties every appearance in your draft back to it, while the timeline records when each alliance formed and broke so the sequence stays coherent even across a serial written out of order. If you want to see how that feels before committing anything, you can open the editor and world builder with no account and rough out a faction map in a couple of minutes. The AI world builder can even take a premise and sketch a starting web of rival groups for you to edit, which is a faster way in than staring at a blank page trying to invent six sects from nothing. Track the political map as diligently as you track your protagonist's power, because in the kind of story where factions matter, the reader is watching the map, and continuity there is what makes a big world feel true.
by Jacob Tam · August 11, 2026
I run IlorisNovel, a platform for web fiction writers. If this kind of world-building writing is your thing, the rest of the blog lives here.