You end a web serial chapter by closing on an open question rather than a closed one, because the only job the final paragraph has is to make the reader want the next chapter. In practice that means cutting one beat earlier than feels natural, since most authors write past the real ending and bury it under a paragraph of tidying up. Vary the type of ending you use, rotating between a revelation, a decision, a threat arriving, and a quiet emotional turn, because a serial that ends every chapter on the same trick trains readers to stop feeling it. And a cliffhanger is a promise, so the next chapter has to pay it rather than open on something unrelated.
How do you end a chapter in a web serial?
You end a chapter by leaving a question open in the reader's head, and that is very nearly the whole of it. Everything else in this article is a variation on that one idea, because the structural situation of a web serial is different from the situation of a novel in a single specific way: your reader is not holding a physical book with three hundred pages of remaining weight in their right hand, they are looking at a screen with a "next chapter" button and roughly four other tabs competing for the same attention. In a printed novel a chapter break is a resting place, an invitation to put the book down and pick it up tomorrow, and that is a perfectly reasonable thing for a chapter break to be. In a serial it is a decision point, and the decision the reader makes there is the single most important number in your whole enterprise, because a serial that loses a percentage of its readers at every chapter break dies by arithmetic long before it dies by craft.
So the final paragraph carries a load that the rest of the chapter does not. It is not there to summarize, it is not there to give the scene a graceful landing, and it is emphatically not there to tell the reader how the protagonist felt about what just happened. It is there to open something. Running a web fiction platform, the pattern I see most often in chapters that lose readers is not weak prose or a slow plot, it is a chapter that answered every question it raised and then stopped, which is a structurally complete piece of writing and a structurally useless piece of a serial. My advice, then, is to read your last paragraph and ask a blunt question about it: after this line, does the reader want something they do not have? If the answer is no, the ending is not doing its job, however well written it is.
Cut one beat earlier than feels natural
The most reliable fix for a flat chapter ending is to delete the last paragraph you wrote, and I mean that fairly literally as a first-draft habit rather than as a metaphor. Almost every author writes past the ending, because the instinct that carries you through a scene does not switch off cleanly at the interesting moment; you land the revelation or the arrival or the decision, and then you keep going for another paragraph to let the character react, to describe the room going quiet, to gesture at what tomorrow might bring. That paragraph feels like craft while you are writing it, and it is the single most common thing standing between a serial chapter and a good hook, because its entire function is to release the tension that the previous paragraph just built.
Try it on your own draft. Take the last two paragraphs of a chapter you have already written and ask whether the second one gives the reader information they need right now, or whether it simply lets everyone catch their breath. If it is the breath-catching kind, it does not need to be cut from the story, it needs to be moved, because the top of the next chapter is exactly where a reaction beat belongs and it will read as a natural opening rather than as a deflating close. He looked at the door for a long moment, then turned back to the fire and told himself it was nothing is a fine sentence and a terrible chapter ending. The door was open. He was certain he had locked it is the same information cut one beat earlier, and the difference in whether the reader clicks through is not subtle. Matt Dinniman does this relentlessly in Dungeon Crawler Carl, ending chapters on the arrival of the next problem rather than on Carl's assessment of the last one, and the sheer momentum of that book is built substantially out of that one editing decision applied several hundred times.
Vary the kind of ending, not just the intensity
A serial that ends every chapter with the same move stops working, because readers adapt to any repeated pattern and a hook they can feel coming is not a hook. This is the part authors underestimate when they are told to end on cliffhangers, and it produces the specific failure mode where a story that seemed relentless in its first ten chapters starts to feel wearing by its fortieth: not because the endings got weaker, but because they got predictable, and predictability is the one thing a hook cannot survive. I think of the available endings as a small set of distinct types worth rotating through deliberately. There is the threat arriving, which is the classic hard cliffhanger and the one everyone thinks of first. There is the revelation, where the reader learns a fact that reframes the chapter they just read. There is the decision, where a character commits to a course of action and the reader wants to see it play out. And there is the quiet turn, where nothing external changes but a character understands something new, which pulls in a completely different register and is enormously useful after an action-heavy stretch.
The quiet turn is the one most web fiction authors underuse, and I think it is the most valuable of the four precisely because it is rare. Ending on a shift in understanding rather than a shift in danger leaves the reader holding a question about meaning rather than about survival, and that question is often stickier, because a threat resolves and a reframing does not. Will Wight's Cradle series, running from Unsouled in 2016 through Waybound in 2023, moves fluently between these registers, closing some chapters on an incoming fight and others on a single line that quietly recalibrates how much trouble Lindon is actually in. My feeling about this is that the variety is doing as much work as any individual ending, because it keeps the reader unable to predict the shape of what is coming, and an unpredictable rhythm is what keeps the hooks feeling live forty chapters in.
A cliffhanger is a promise, so pay it
The fastest way to lose a reader's trust across a long serial is to end a chapter on an apparent shock and then open the next one revealing it was nothing, and this is a mistake that is invisible in a single chapter and corrosive over fifty. Readers keep score, mostly unconsciously, and a serial that repeatedly cashes out its cliffhangers as false alarms teaches them that the hooks do not mean anything, at which point the endings stop working entirely and no amount of craft in the closing line will restore them. If the last line of chapter twelve says something is standing in the doorway, then something is standing in the doorway, and the reader who came back for chapter thirteen is owed the payoff reasonably promptly rather than after a nine-hundred-word flashback about the protagonist's childhood.
There is a related discipline about where the next chapter opens. A hook creates a small debt, and the natural place to pay it is at or near the top of the following chapter, which means the ending of one chapter and the opening of the next are really a single craft decision made in two places. Authors who cut hard at the end and then open the next chapter on an unrelated point-of-view or a leisurely scene-setting passage get the worst of both, since the hook's energy dissipates in the gap and the reader learns that the click was not rewarded. So when you cut a chapter early, immediately write the first two paragraphs of the next one, while you still remember what the reader is going to be holding in their head when they arrive. This is also, incidentally, one of the practical reasons to draft ahead into a buffer rather than chapter to chapter, since seeing the seam between two chapters as one unit is much easier when both of them already exist; drafting a run of chapters and shaping their edges together is the workflow the IlorisNovel editor is built around, and you can try it with no account if you want to feel how it changes the editing pass.
Chapter length and chapter endings are the same problem
Where you end a chapter and how long that chapter runs are not two separate questions, and treating them separately is why authors end up padding to a word count and then wondering why their endings feel limp. The chapter's length should be determined by where the natural hook falls, not the other way around, which in practice means that if your target is around two thousand words and the strongest available ending sits at seventeen hundred, you end at seventeen hundred. Writing three hundred more words to hit a number will, almost without exception, carry you past the good ending into exactly the tidying-up paragraph this article has been arguing against, and you will have traded your hook for word count that nobody asked for. I have written separately about how long a web fiction chapter should be, and the short version is that the range matters far less than the consistency, but the point relevant here is that the hook wins any conflict between the two.
The inverse case is real too. If you reach your target length and the scene has not yet arrived at anything worth stopping on, the answer is usually to keep writing rather than to manufacture a break, because a chapter that ends in the middle of nothing is worse than a chapter that runs long. Some of the strongest serial chapters I have read are noticeably longer or shorter than their neighbors for exactly this reason, and readers do not notice length variance nearly as much as authors fear they do; what readers notice, and respond to, is whether the ending made them click. Everything else in serialization craft, the pacing, the arc structure, the update schedule, sits downstream of that single behavior, which is why the closing line of a chapter deserves a disproportionate share of your editing attention relative to its length. If you want the wider frame around all of this, the complete guide to writing web fiction lays out how chapter craft fits into the rest of running a serial.
The habit worth building out of all this is small and mechanical, which is that after you finish a chapter you go back to the last three paragraphs and treat them as their own editing problem, separate from the rest of the draft. Find the last moment where something genuinely changed, cut to within a line or two of it, move whatever came after to the top of the next chapter, and then check what type of ending you have landed on against the last two chapters to make sure you are not running the same move three times in a row. It takes a few minutes per chapter and it is, as far as I can tell, the highest-return editing pass available to a serial author, because it operates directly on the one decision point where readers actually leave.
Common questions about ending web serial chapters
How do you end a chapter in a web serial?
You end a web serial chapter on an unanswered question rather than a resolved one, because the final paragraph exists to make the reader open the next chapter. The most reliable technique is to find the moment where the situation changes, a revelation lands, a decision is made, or a threat arrives, and stop within a line or two of it instead of writing through to the emotional settling afterward. Endings that tidy everything up feel complete, and a complete feeling is exactly what makes a reader close the tab.
Should every chapter of a web serial end on a cliffhanger?
No, and a serial that ends every chapter on a hard cliffhanger usually reads as exhausting rather than gripping, because readers adapt to a repeated trick and stop responding to it. The stronger pattern is variety, rotating between hard cliffhangers, quieter emotional turns, revelations, and decisions, so that the hard ones land harder for being rarer. What every chapter does need is an open question of some kind, even a small one, since an unanswered question rather than a shock is what actually pulls a reader forward.
Where should you cut a web serial chapter?
Cut one beat earlier than feels natural, which usually means deleting the final paragraph you wrote. Most authors write past the real ending, landing the interesting moment and then adding a paragraph of reaction, reflection, or scene-setting that releases the tension the moment just created. Look at your last two paragraphs and ask whether the second one adds information the reader needs immediately or simply lets everyone catch their breath, and if it is the latter, it belongs at the top of the next chapter rather than the bottom of this one.
How do you end a chapter without using a cliffhanger?
You end without a cliffhanger by closing on a change in understanding rather than a change in danger, letting a character realize something, decide something, or see an established fact in a new light. A line that reframes what came before leaves the reader holding a question about meaning rather than about survival, and that question pulls just as hard. Chapters that end on a resonant image or a quietly ominous observation also work well, particularly after an action-heavy chapter where a hard hook would feel relentless.
What makes a bad chapter ending in web fiction?
The most common bad ending is the one that resolves everything, tying off the chapter's tension so cleanly that the reader has no reason to continue. Close behind it is the fake cliffhanger, where the chapter ends on an apparent shock that the next chapter immediately reveals as nothing, which readers experience as being cheated and which erodes trust across a long serial. Endings that stop mid-sentence or mid-action for effect without a real story beat underneath also tend to read as a gimmick rather than as craft.
by Jacob Tam · July 18, 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.