Most advice about serial fiction is about starting: how long the first chapter should be, whether to open with a prologue, what to promise in the blurb. Almost none of it is about stopping. That is a strange gap, because ending a serial is structurally harder than ending a novel. A novelist can rewrite the last third once they know what the ending is. A serial writer has already published the last third's setup, in public, to readers who remember it.
The ending is not a craft problem you solve at the end. It is a scheduling problem you solve a quarter of the way through, and a revision problem you solve by keeping room you have not spent.
The ending is a stretch of chapters, not a chapter
The first mistake is treating the finale as a single unit of work: the last chapter, written last, carrying everything. The reader does not experience the ending as one sitting. They experience it as four or six or ten arrivals spread over weeks, each of which has to feel worth showing up for.
Decide early how many chapters your ending is. Not which events go in them, just the count. Six chapters of ending is a different book from two, and the difference shows up in what you are allowed to introduce beforehand. With a number, the last stretch stops being a cliff and becomes a section you can plan and pace.
Pick the chapter where the ending starts, and write the date down
Every serial has a chapter where the story stops widening and starts closing. Readers feel it even when they cannot name it: new characters stop arriving, and the questions already asked start getting answered. That turn is a decision, and if you do not make it deliberately the story makes it for you, usually forty chapters too late.
Pick the chapter number, then convert it to a date using your real cadence, not your best week. Weekly, turn at chapter sixty, currently on thirty-one: the turn is seven months away. That number is useful in a way the chapter number is not. It tells you how much story you have left to fill, and whether the subplot you are about to open has time to pay off before the doors close.
Bank the final arc before you publish any of it
This is the same shape as the buffer you built at launch, but not the same argument. The launch buffer exists so a bad week does not become a missed update; if you have not built one, picking an update schedule you can actually keep is the place to start. The ending buffer exists for revision.
An ending is the part of a story most likely to change while you write it, because it is where everything has to agree at once. If chapter sixty-four is published and chapter sixty-seven turns out to need a different setup, you cannot go back. If all six are drafted before the first ships, you can move a revelation earlier, cut a scene that went redundant, and let the last line answer something planted in the first. Draft the whole ending, then publish it. The wait is the price of being able to revise.
Stop opening doors well before you need to close them
The most common way a serial ending goes wrong is not a weak final chapter. It is an overloaded one, carrying six threads opened because they were interesting at the time. Serials invite this: a new character who lands well gets more page time, a mystery readers speculate about gets fed. Good instincts mid-run, expensive near the end.
A useful rule: after the turn you may close, deepen and complicate threads, but you may not start one that needs more than two chapters to resolve. Anything arriving after the turn either pays off almost immediately or is scenery. If the cast has grown past what one ending can carry, keeping a growing serial cast legible covers demoting characters gracefully rather than writing them all a farewell.
Spend the payoff you have been banking
Serial writers get good at withholding, because withholding is what makes a two-week gap survivable. The habit does not switch off by itself, and the failure at the end is a story that keeps deferring a payoff it has already earned. In the last stretch the accounting reverses: you are spending, not saving. Long-running anime handles this trade explicitly, and the lesson transfers.
This changes how a chapter in the ending should close. Mid-run, a hook that survives a two-week wait is the right close for almost every chapter. In the final arc, some of them should close on something delivered instead: a question answered, a confrontation that actually happens, a character getting the scene they are owed. A finale made entirely of cliffhangers reads as stalling, because it is.
Give the last chapter a job that is not plot
If you paced the ending well, the plot is substantially resolved before the last chapter starts. That is correct, and it frightens people into stuffing one more twist in. Do not. The final chapter's job is to show what is different now.
That usually means a quiet scene: the character in the place they started, behaving in a way they could not have at the beginning. It can be short. It should be the chapter you revise most, because every reader who finishes reads it, and many of them decide in its final paragraphs whether to read your next one. Where an opening has to earn a click, a closing has to earn the next book.
Tell your readers the ending is coming
A serial that simply stops looks abandoned, and readers have seen enough abandoned serials to assume the worst. Say it plainly in a note at the turn: this story ends in about twelve chapters. Nobody drops a serial because they learned it has an ending; plenty drift away from one that seems to run forever. It also commits you, which is the real benefit. An announced ending is a promise with a number attached, and those are the ones that get kept.
If the story has outgrown the ending you planned
Sometimes the turn arrives and the ending you outlined months ago no longer fits what the story became. That is normal, and it is not a reason to keep writing until it resolves itself. Two honest options: end this story where it can end and treat the rest as a second book, or extend by a fixed number of chapters and announce the new count. What is not honest is an open-ended extension, which is how a serial becomes a thing the writer dreads and the reader stops opening.
If the problem sits earlier than the ending - setup that no longer matches where you went - there are ways to fix a serial you have already published without pretending the early chapters say something they do not.
Decide when the ending starts, draft it whole, stop opening doors, and say so out loud. The rest is writing the chapters you already know you owe.
Landing the ending is one serial challenge; sustaining the middle is another. Read how to keep a subplot alive in a serial.