Updating your session…
Iloris
How to Write an Unreliable Narrator in Web Fiction
Opens in a new tab.Back to Writers

How to Write an Unreliable Narrator in Web Fiction

Jacob TamNovember 15, 2026

You write an unreliable narrator by making the gap between what the narrator says and what is actually true into a deliberate, readable second story. The reader has to be able to sense that the account is slanted, catch the places where it does not add up, and slowly assemble the truer version underneath, which means the unreliability has to be planted and consistent rather than sprung as a cheat at the end. The first decision is what kind of unreliable you are writing, because a narrator who is lying to the reader, one who is lying to themselves, one who is too young or too mad to understand what they are describing, and one who simply does not yet know the truth are four different machines with four different contracts. Web fiction sharpens all of this, because a serial narrator is cross-examined in public by a comment section that theorizes in real time, so a slant that is cheaply hidden gets caught while one that is honestly built gets rewarded.

What makes an unreliable narrator actually work?

An unreliable narrator works when the reader can feel the gap between what the narrator claims and what is really happening, and can do some of the work of closing that gap themselves. The term was coined by the critic Wayne C. Booth in The Rhetoric of Fiction in 1961, and his framing is still the most useful one, because he defined reliability against the values of the work rather than against mere facts. A narrator is unreliable when they speak or act against the norms the whole text quietly endorses, so the reader ends up reading two things at once, the narrator's version and the author's. That doubleness is the entire effect. If the reader only ever receives the narrator's version with no way to sense the slant, you have not written an unreliable narrator, you have just written a narrator who happens to be wrong, and the difference matters enormously. This is one of the hardest moves in the broader craft of writing web fiction, because it asks you to communicate past your own narrator to the reader without the narrator ever saying so.

My feeling about this is that the failure most new writers hit is treating unreliability as a reveal rather than a texture. They write a straightforward account and then, at the end, announce that the narrator was lying the whole time, as though the lie were a trapdoor. Agatha Christie's The Murder of Roger Ackroyd, published in 1926, is the famous case where the first-person narrator turns out to be the killer, and it works not because the ending is a shock but because the narration was scrupulously, technically honest the whole way through, withholding without ever stating a falsehood, so that rereading rewards you. The gap was always there to be measured. Compare that to a narrator who simply asserts false things and is later corrected, which produces no reread value at all, because there was nothing for the reader to notice in the moment. The craft is in the planting, not the punchline.

Decide what kind of unreliable you are writing

Before you write a line of the slanted narration, you have to know which kind of unreliability you are running, because each one makes a different promise to the reader. The deliberate liar knows the truth and chooses to deceive, which is the most dangerous kind to sustain, because a narrator who lies to the reader can feel like the author cheating unless the lies are fair. Gillian Flynn's Gone Girl, from 2012, handles this by splitting the book between two narrators and letting the diary half be a constructed fiction inside the story, so the lie belongs to a character with a motive rather than to the storyteller, which keeps it honest. The self-deceiver is gentler and often richer, a narrator who believes a flattering or protective version the text keeps undercutting around them. Kazuo Ishiguro's The Remains of the Day, from 1989, is the model here, where the butler Stevens narrates a life of dignified service while every scene he reports shows the reader the love and the complicity he cannot admit to himself.

Then there is the naif, whose account is accurate but whose understanding is not, because they are too young or too innocent to interpret what they are describing. Emma Donoghue's Room, from 2010, narrates captivity through a five-year-old who reports the facts faithfully and reads them completely wrong, and the horror lives entirely in that gap. The madman distorts perception itself, so the narration is unreliable at the level of what is even happening, as in Edgar Allan Poe's The Tell-Tale Heart from 1843, where the narrator's insistence on his own sanity is the proof against it. And finally there is the narrator who is simply in the dark, honest and sympathetic but missing a fact that recolors everything, which is the kind that shades most naturally into a twist. I think most memorable narrators blend two or three of these, and the point of naming them is not to pick a pure type but to know, scene by scene, whether your narrator is lying to you, lying to themselves, or just does not know, because the reader needs that to be stable even while they cannot yet articulate it.

Give the reader a yardstick to measure the gap

The single most practical technique is to plant a second, truer signal alongside the narration, something the reader can measure the narrator's account against. Unreliability only reads as unreliability if there is a reference point, so you build in evidence that contradicts, or at least complicates, what the narrator is telling you. This is usually dialogue and action, because a narrator can color their own interior monologue freely but has a harder time rewriting what another character plainly does on the page. When Stevens tells us a relationship was purely professional and then reports, flatly and without comment, the housekeeper crying behind a closed door, the scene is the yardstick. The reader trusts the reported action over the narrator's gloss on it, and the gap between the two becomes the real story.

A worked example helps. Say your first-person protagonist in a progression fantasy insists they are a pragmatic, reluctant fighter who only ever kills when forced. The weak version simply has the character say this and moves on, and the reader believes it. The strong version plants the yardstick, so three chapters running the narrator describes a cornered enemy, a clear chance to walk away, and a choice to strike anyway, each time narrated with a quick self-justification, he would only have come back for us, that the surrounding facts do not quite support. You never once write that the narrator is bloodthirsty. You let the pattern of chosen violence sit next to the repeated claim of reluctance, and the attentive reader assembles the truer picture on their own. This is closely related to the machinery of how to write a plot twist in web fiction, because an unreliable narrator is, in effect, a twist distributed across every page instead of concentrated in one, and the same rule holds, which is that you show the real thing plainly and let the narrator's framing do the misdirecting.

The web fiction problem: your narrator is cross-examined in public

Writing an unreliable narrator for a web serial is harder than for a novel, because a serial narrator is interrogated by a live comment section that reads in parallel, theorizes in real time, and never forgets. This is the thing that genuinely changes the craft, and it cuts both ways. A novelist's slanted narrator has to fool one reader reading alone and in a single pass, whereas a serial writer's narrator faces hundreds of readers pooling their noticing over months, with an archive they can scroll back through to check the account against itself. Running a web fiction platform I see this constantly, where the comment thread works out that the narrator is shading the truth long before the author planned to let it surface, because collective attention is a far better lie detector than any individual. Wildbow's Worm, serialized from 2011 to 2013, is a useful example of a narrator whose reliability its readers argued about at length, since Taylor narrates her own steady escalation with a running self-justification that the fandom learned to read against, weighing what she claims about her motives against what she actually does.

The liberating consequence is that you should stop trying to win by hiding and start trying to win by being fair. The slant that depends on readers forgetting a detail from eighty chapters ago will not survive, because someone has it bookmarked and screenshotted, and the reveal that the narrator was lying will land as a cheat if the comments can prove you gave them no honest way to suspect it. What does work is the opposite, a narrator whose unreliability is planted densely and consistently enough that the sharp readers who catch it early feel clever and rewarded, while the ones who took the account at face value get the slow, pleasurable click of rereading once they know. My advice, then, is to write the slant for the sharpest reader in the thread and assume it will be caught, so that being caught is part of the design rather than the thing you were hoping to avoid. The choice of how close your narration sits to the character is load-bearing here, and it is worth thinking through alongside first person versus third person in web fiction, because first person and very close third are where a believable slant can live, and detached omniscience is where it reads as the author lying.

Track the true version separately from the told version

An unreliable narrator is only fair if the truth underneath stays genuinely consistent across hundreds of chapters, and keeping the told version and the true version straight over a years-long serial is a real logistical problem, not just an artistic one. The danger is specific, because you establish a narrator's blind spot in one arc, write for another six months, and by the time the truth matters you have half-forgotten exactly what the narrator does and does not know, or let their self-image drift, so a careful reader finds the account contradicting itself in a way that reads as a mistake rather than a design. Readers will find it, and nothing collapses a carefully built narrator faster than proof that the author lost track of their own double bookkeeping. This is the most practical reason to record, explicitly, two layers for your narrator: what actually happened and what they believe or claim happened.

On IlorisNovel the connected wiki and timeline are built for exactly this kind of two-layer tracking, so you can keep an entry for your narrator that notes the truth of an event beside the version they have told the reader, and mark the chapter where each honest yardstick was planted, which turns the slant from something you hold in your head into something you can audit before you publish. Beyond the bookkeeping, the deeper lesson is that the strongest unreliable narration tends to grow out of a character you understand completely rather than a trick you decided to play. When you know precisely why the narrator cannot see or will not say the truth, the slant writes itself scene by scene and stays consistent without effort, because it is not an effect you are imposing from outside but a wound the character genuinely carries. Get that understanding right and keep the two layers honest, and the gap between what is said and what is true does most of the work for you, quietly, on every page, which is exactly where an unreliable narrator is supposed to live.

by Jacob Tam · November 15, 2026

I run IlorisNovel, a platform for web fiction writers. If you want to see what writing here feels like, you can try the editor with no account.