How to write a mystery: clues and red herrings
The implicit contract with a mystery reader is stricter than in almost any other genre: the reader agrees to be misled, not to be cheated. The difference lies entirely in fair play — the unwritten rule that the solution must be reconstructible from the clues given, even if the reader, like the characters, doesn't get there before the last page.
The culprit needs to arrive early, not be conjured at the last minute
The most discrediting mistake in a mystery is revealing as the culprit a character who first appeared in the second-to-last chapter. The reader rightly perceives it as a cheat: they had no chance to suspect them, because they didn't even know they existed.
The culprit needs to show up early, move through the story with an apparent function of their own — a witness, a relative, a sidekick — and have, from the start, the logistical possibility of having committed the crime. The reader has to have met them enough to be able to suspect them, even if they don't.
Clues: hidden, not absent
A well-placed clue isn't invisible: it's there, but set in a context that distracts from its weight. The classic technique is to bury it among neutral details, or have a character distracted by something else notice it, so the reader registers it without giving it importance.
The test is simple: at the end of the book, the reader should be able to go back and find the clue exactly where they thought it was just scene-setting. If the clue wasn't there at all, the solution is a trick. If it was too obvious, the solution is predictable.
Red herrings: plausible suspicions, not lies from the author
An effective red herring doesn't lie to the reader: it makes them notice something true — a motive, an opportunity, suspicious behaviour — that points the wrong way because the reader misreads it, not because the author misrepresents it.
The difference is subtle but crucial. "The butler had a motive" is an honest red herring if the butler genuinely had one. "The butler confesses to a murder he didn't commit for no plausible reason" is a trick, and the reader feels it as such once the real solution arrives.
The detective: what they know, what they deduce, what's withheld from the reader
If the point of view follows the detective, managing their deductions is delicate: revealing them all as they happen drains tension; hiding them systematically feels unfair. The most common solution is to show what the detective observes without always showing what they conclude from it — the reader sees the clue alongside them, but the conclusion arrives only at the end.
A narrator who follows a companion of the detective — as in the tradition starting with Watson — solves the problem at the root: the narrator themselves knows no more than the reader does.
The solution: surprising yet, on rereading, inevitable
The moment of revelation needs to produce two feelings at once, apparently in contradiction: surprise, because the reader didn't see it coming, and inevitability, because reviewing the story every piece falls into place. Missing the surprise, the plot was too predictable. Missing the inevitability, the reader feels cheated.
A good revision exercise is to reread the book from the end knowing who the culprit is, and check that every one of their actions, seen with that knowledge, is consistent with what the reader read the first time.
Internal consistency: the mystery's silent enemy
More than any other genre, a mystery doesn't forgive inconsistencies: a timeline that doesn't add up, a character who knows something they couldn't have known, a distance covered in an impossible time. Mystery readers are attentive readers by definition, and they notice what would slip by unnoticed elsewhere.
Keeping a precise chronology of where every character is at each key moment, and exactly what they know at that point, isn't a technical detail: it's the genre's load-bearing structure. In NovLore, the story timeline and character sheets live in the dashboard for exactly this reason, and stay within reach when the AI assistant helps check whether a clue or an alibi holds up against the rest of the book.
Frequently asked questions
Should the reader be able to solve the case before the detective?
It isn't mandatory, but they should be able to in principle: every necessary clue has to be given, even if well hidden. Whether the reader gets there or not depends on their attention, not on missing information.
How many suspects does a mystery need?
There's no fixed number, but three or four credible suspects, each with a plausible motive and opportunity, are usually enough to sustain the plot without scattering the reader's attention across too many characters.
Can a character lie to the reader?
A character can lie to other characters — that's normal and often necessary. The problem arises if the narrator itself, not a character, withholds or distorts a fact the reader would have a right to know: that breaks the fair-play contract.