How to solve a mystery case: a detective's method
6 min read
A repeatable five-step method for reading evidence, ranking suspects and naming the culprit without guessing.
Start with the timeline, not the people
Before you look at a single suspect, rebuild the order of events. Write down when the victim was last seen, when the crime happened, and when the body or the loss was discovered. Every alibi you later read is only meaningful against that window.
Most players read the suspect list first and get anchored on whoever sounds suspicious. The timeline is neutral. It cannot be charming or nervous, and it quietly eliminates half of your options.
Separate facts from claims
A clue is something the scene tells you: a broken lock, mud on a carpet, a receipt with a time on it. A claim is something a person tells you. Facts constrain the case; claims can be lies.
Keep two mental columns. When a claim contradicts a fact, you have not found the culprit yet, but you have found someone who is hiding something, and that is where questions pay off.
Rank suspects by opportunity first
Motive is the most interesting part of a mystery and the least useful for solving one. Many people have a reason; far fewer had the physical opportunity.
Ask, for each suspect: could this person have been in that place, in that window, without being seen by anyone else in the file? Cross out everyone who could not. What remains is your real shortlist.
Ask questions that can be wrong
A weak question invites a story. A strong question invites a checkable answer. Instead of "what happened that night?", ask "who else was in the corridor when you left?" — an answer you can test against another statement.
Ask the same narrow question to two different suspects. Honest people give boring, matching answers. Liars improvise, and improvisation drifts.
Commit only when one story explains everything
Before you accuse, tell yourself the full sequence out loud: how the culprit got in, why the physical clues look the way they do, and why every other suspect's account still holds.
If your explanation needs a coincidence to survive, it is the wrong explanation. A well-built case has no leftovers.
Try the method on a real case now.
Open a case