Writing web fiction is writing a story serially, in order, in public, on a schedule, and that one constraint reshapes almost every craft decision you make. You cannot revise chapter three after chapter three hundred is live, readers judge the whole book on the first installment, and the only thing that keeps them coming back is that every chapter ends wanting the next. So the craft optimizes for retention rather than for the perfect finished object: hook fast, keep the forward motion visible, end chapters on a pull, keep continuity in something you can search, and above all keep publishing, because a finished serial beats a perfect first arc that never reached chapter twenty.
What makes writing web fiction its own craft
Writing web fiction is the craft of telling a story in serial installments that go live as you write them, which is a genuinely different discipline from writing a novel, and treating it like novel-writing with a web address is the single most common mistake I see new authors make. The difference is not the platform or the genre. It is that a novelist gets to finish the whole book in private, discover in the second draft that the opening was slow and the middle sagged, and quietly fix both before a single reader turns page one. A web fiction author has no second draft in that sense, because chapter three was published months or years before chapter three hundred, thousands of people have already read it, and the story has to work while being built and shown at the same time. Running a web fiction platform, I watch this play out constantly: the authors who struggle are usually the ones writing a beautiful novel one chapter at a time, and the authors who thrive are the ones who understood that the serial is a different animal with different rules.
Those rules all descend from one fact, which is that the reader decides whether to continue after every single chapter, not just at the end of the book. A traditional reader who bought your novel is fairly likely to finish it; a serial reader is making a fresh choice to click "next" dozens or hundreds of times, and any chapter that does not earn the following one is where they leave for good. This is why the craft of web fiction bends so hard toward retention. It is also, to be fair, why the format has produced some of the most propulsive fiction of the last fifteen years, from Wildbow's Worm, a superhero web serial that ran to roughly 1.7 million words between 2011 and 2013, to Matt Dinniman's Dungeon Crawler Carl, which began on Royal Road and moved to traditional publishing in 2024 without losing an ounce of its serialized momentum. Everything below is a consequence of that one constraint, organized as the questions I would work through with an author starting their first serial.
The opening has to earn the second chapter, not the whole book
The most important craft decision in a web serial is how it opens, because in this format the opening does not just set the tone, it decides whether anyone reads past chapter one at all. A traditional novel can afford a slow, atmospheric first chapter, because the reader already committed by buying it and will usually give it fifty pages. A serial reader gives you one chapter, sometimes one scene, and the ranking systems that drive discovery on a site like Royal Road are largely downstream of how many readers make it from chapter one to chapter two. So the opening's job is narrow and specific: introduce a character worth following, establish the story's central promise, which in progression fantasy usually means the shape of the growth to come, and end on something that makes the next click feel necessary. My feeling is that most weak serials are not badly written, they are badly opened, spending their crucial first chapter on backstory and setting when they should be spending it on momentum.
The trap is the routine opening, where the author establishes the protagonist's ordinary world at length before the story actually starts, and I care about this one enough that I gave it a full treatment in whether your novel should open with a routine. The short version is that ordinary-world openings can work, but they are the highest-difficulty way to start a serial, because they ask the reader to invest in a status quo you are about to demolish. Compare that with how He Who Fights with Monsters by Shirtaloon opens with the protagonist waking up in an unfamiliar world already in motion, or how a good LitRPG drops its first system prompt within pages so the reader immediately understands the game they are being invited to play. You do not have to blow anything up on page one, but you do have to answer, fast, the reader's only real question, which is why should I still be here at the end of this chapter.
Progression is a promise, and pacing it is the core skill
The engine of most web fiction is visible forward motion, so pacing that progression, doling out growth fast enough to satisfy and slow enough to sustain, is the central ongoing craft skill of the format. This is most obvious in progression fantasy and LitRPG, where the protagonist's literal power level is the spine of reader investment, but it is true of almost every successful serial, because the serial contract is fundamentally a promise that things will keep developing. The failure mode on one side is a story that gives everything away too fast, where the protagonist is unstoppable by chapter thirty and the tension evaporates because nothing is at stake. The failure mode on the other side is the story that stalls, where two hundred chapters in the protagonist is grinding the same level and the reader stops believing the promise. Getting this curve right is hard enough that I devoted a whole piece to writing meaningful progression, because the trick is that progression has to feel earned and costly rather than merely numerical.
What separates the serials that hold readers for years is that their progression is legible and their gains carry weight. Will Wight's Cradle, which completed at twelve books in 2023, works because every advancement up its cultivation ladder is felt and constrained, so the reader always understands roughly what the protagonist can and cannot do, and each new tier arrives as a genuine event rather than a bigger number. When the growth is quantified, as it is in a LitRPG, the presentation of that growth becomes its own craft problem, which is why I wrote separately about how to write LitRPG system messages that inform without drowning the prose in blue boxes. My advice, then, is to treat your progression curve as deliberately as a novelist treats plot structure, because in web fiction it more or less is the plot structure, and a reader who can feel the ladder they are climbing will forgive a great deal of everything else.
Chapter craft: length, titles, and ending on a pull
The unit of web fiction is the chapter, and three chapter-level decisions, how long it runs, whether it is titled, and how it ends, do more day-to-day work than any of them individually seem to deserve. On length, most successful serials land somewhere between fifteen hundred and four thousand words, with a couple of thousand as a common sweet spot, because that is an installment a reader can finish on a phone in one sitting and still feel a beat of completion. There is no universal number, and Worm ran longer chapters while plenty of thriving serials run shorter, but the deeper rule is consistency, since readers attach to a rhythm and an erratic one costs you more than a slightly-wrong word count ever will. Titles are a smaller decision with a real effect on browsing and re-reading, and it is genuinely arguable in both directions, which is why I laid out the tradeoffs in whether web fiction chapters should have titles rather than pretending there is one right answer.
The decision that matters most, though, is how the chapter ends, because the last line is the hinge the whole retention machine turns on. Ending on a pull does not mean ending every chapter mid-swing on a literal cliffhanger, which gets exhausting and starts to feel like a cheap trick when overused; it means ending on something unresolved that the reader wants resolved, whether that is an imminent decision, a question just raised, a threat that has appeared, or simply visible progress toward a goal with the next step in sight. Matt Dinniman is a master of this in Dungeon Crawler Carl, where nearly every chapter closes on a note that makes stopping feel like leaving in the middle of a sentence. The habit to build is that you never end a chapter at a resting point if you can end it a beat earlier, on the tension, because a resting point is exactly where a serial reader closes the tab and does not come back.
Character, description, and continuity under serial pressure
Character work and description in web fiction follow the same underlying rule as everything else, which is that you introduce and deepen them in motion rather than in front-loaded blocks, because the serial reader will not wait through a static portrait. The instinct most new authors carry over from novels is to describe a character thoroughly the moment they appear, and in a serial that stalls the momentum you spent the opening building, which is why I argued for describing characters without info-dumping by weaving appearance and personality through action instead of pausing for a catalog. The same applies to competence and interiority: a protagonist the reader believes is genuinely clever, rather than one the narration merely insists is clever, is built through choices under pressure, which is a harder trick than it sounds and one I took apart in how to write a smart character. In both cases the serial constraint pushes you toward showing over telling not as an aesthetic preference but as a retention necessity.
Underneath all of this sits the least glamorous craft skill the format demands, which is continuity, and it breaks more long serials than bad prose ever does. When you are hundreds of chapters and a couple of years into a story you cannot revise, you will not reliably remember what you named a side character, how a particular ability worked, or which faction held which territory, and the moment you contradict yourself the comment section finds it within the hour. The authors I talk to who serialize successfully for years almost all keep an external, searchable record of their world rather than trusting memory, which is exactly the problem the world builder and wiki side of the craft is built around. A serial's world has to be a lookup, not a feat of memory, and building it somewhere you can search is what lets the story you invented in chapter three still hold up in chapter three hundred.
Finishing the thing: cadence, buffers, and the long middle
The hardest part of writing web fiction is not any single craft problem, it is sustaining the whole endeavor long enough to finish, because burnout and the sagging middle kill far more serials than weak writing does. The format asks something novels do not, which is that you keep producing on a schedule, in public, for months or years, often through a long stretch where the initial excitement has faded and the finale is still distant. That long middle is where most serials quietly die, and I wrote about climbing out of the mid-novel slump because it is so common that treating it as a normal phase rather than a personal failure is half the battle. The structural fix is to build arcs that each deliver a real payoff instead of deferring everything to the end, so both you and the reader get the satisfaction of completion at regular intervals rather than running on the fumes of a promise that never pays out until chapter four hundred.
The practical habits that let authors actually finish are less romantic than the craft advice but arguably more important. Write ahead of your posting schedule so a bad week does not break the streak, since a buffer of finished chapters is the difference between a missed personal deadline and a broken promise to readers. Protect a cadence you can sustain for the long haul rather than the sprint you can manage for a month, because readers reward reliability over speed, and a serial that updates twice a week for two years will outrun one that updates daily for six weeks and then vanishes. And give yourself permission to discovery-write within a known destination, which the serial format uniquely enables through reader feedback, a process I found worth defending in the case for discovery writing in web fiction. If you want to feel out how an opening chapter or a first system screen actually reads before you commit a serial's worth of continuity to it, you can open the editor and draft one with no account, which is the lowest-friction way to test whether the story in your head survives contact with the page. The craft of web fiction is real and learnable, but it rewards the writer who ships, revises forward, and keeps the promise to come back next week over the one waiting to get it perfect.
Common questions about how to write web fiction
How is writing web fiction different from writing a novel?
The difference is that web fiction is published serially and in order while you are still writing it, so you never get to revise an early chapter after readers have seen it. A novel is judged as a finished object you can polish in private; a web serial is judged chapter by chapter, in public, as it goes live. That single constraint reshapes the craft toward retention: fast openings, visible forward motion, chapters that end on a pull, and continuity you can look up rather than remember.
How long should a web fiction chapter be?
Most successful web serials run roughly 1,500 to 4,000 words per chapter, with 2,000 to 2,500 being a common sweet spot. The number matters less than consistency and rhythm: readers on Royal Road or a phone screen want an installment they can finish in one sitting that still ends on forward motion. Shorter chapters posted more often tend to build momentum better than long chapters posted rarely, because cadence is part of what keeps a serial audience attached.
How do you keep readers coming back to a web serial?
You end every chapter on a pull and you post on a schedule readers can trust. The pull does not have to be a cliffhanger; it can be an unanswered question, an imminent decision, or visible progress toward a goal the reader wants. Consistency compounds it: a serial that reliably updates twice a week trains readers to return, while an erratic one loses them no matter how good the prose. Retention, not polish, is the metric the format rewards.
Do you need to plan a web serial before you start writing it?
You need enough of a plan to write the first arc honestly and to know where the story is heading, but not a full outline. Many successful web serials are discovery-written, which the format actually enables because reader reactions become part of your feedback loop. What you cannot improvise safely is your power or magic system and your continuity, so pin those down and keep a searchable record; everything else can be discovered as you draft, provided you always know what the next chapter is.
How do you finish a long web serial without burning out?
You protect a sustainable cadence, build a buffer of finished chapters before you start posting, and design arcs so each one delivers a payoff rather than deferring everything to a distant finale. Burnout kills more serials than bad writing does. The authors who finish tend to write ahead of their posting schedule so a bad week does not break the streak, and they let the story breathe between intense arcs instead of running the protagonist at maximum stakes forever.
by Jacob Tam · July 8, 2026
I run IlorisNovel, a platform for writers of web fiction. If this kind of craft writing is your thing, you can try the editor with no account.