Alibi Strategy
A solver cracks an Alibi case in milliseconds using two ideas a person can copy exactly. This page is those two ideas, plus what to do when a case stops giving.
Alibi is an ordering game
Almost everyone opens an Alibi case looking for the statements that put someone somewhere. That instinct is aimed at the wrong clues. Count up every statement across the cases published so far and roughly four in five are one of two kinds: somewhere left of and somewhere above. Statements that put two things together are uncommon, and statements that keep two things apart are close to a curiosity — under one percent of everything ever shipped.
So Alibi is not really a game about who was with whom. It is a game about the order things sit in along a row and down a column, and the arrangement is what falls out of getting that order right. Read the rail with that expectation and the case opens much faster.
Every spatial clue is two clues
The wording of a statement is deliberately natural, which means it under-sells what it asserts. The rules page covers the two big ones: somewhere left of also means in the same row, and somewhere above also means in the same column. Neither is a hint about direction alone — each one pins an axis outright.
Here is the one nobody tells you. All three spatial statements — left of, above, next to — are impossible to satisfy when both things are in the same room. So every one of them is quietly also an apart statement. When the case says “Hartshorn is next to the Knife,” it has told you two separate things: the two are orthogonally touching, and Hartshorn is not holding the Knife. That second half is free information, it arrives on a clue you were already reading, and most solvers never bank it.
Four moves, in the order a machine would use them
Alibi’s solver cracks a case in a few milliseconds, and the interesting part is that it does so with a very small bag of tricks — small enough to copy by hand. Here they are, in the order it applies them. The first three never guess at anything, and on most days they are the entire game: roughly seven in ten published cases can be finished by pure propagation, without ever hypothesising once. On Mondays and Tuesdays it is very nearly all of them.
1. Support-checking — and check it backwards too
Take one statement and one room you think a suspect could occupy. Ask: if they were here, is there anywhere left for the other half of this clue to be? If the answer is no, they cannot be in that room — and notice you have just eliminated a room without placing anything.
The half everyone skips is running it in reverse. Having tested every room for the suspect, test every room for the other item in the same clue. The machine does both directions on every clue, every pass, and the backward direction is where most of its early progress comes from.
2. Keep two lists, not one
Almost every solver tracks the board one way: for each room, who could be in it. That misses half the deductions available. The engine keeps the mirror list too — for each person, which rooms they could still be in — and it pays off constantly.
With the first list you catch the obvious: a room down to one candidate is settled. With the second you catch the one people miss: if a suspect has been squeezed out of every room but one, that room is theirs — even if that room still looks busy with three other candidates. Nothing about the room told you. Only the person’s own list did.
3. After every placement, start the rail again
A statement that did nothing the first time you read it is not spent. It was waiting on a fact you didn’t have yet. The engine never marks a clue as used — it re-applies all of them, top to bottom, and keeps going around until a full pass changes nothing.
Copy that literally. The moment you lock anything, go back to statement one. The temptation is to keep pushing the clue you were in the middle of; the value of a new fact is almost always somewhere else on the rail.
4. When you truly stall, attack the tightest slot
Sometimes propagation genuinely runs out — on a Saturday or Sunday case, often. Don’t reach for the most interesting-looking suspect. Find the slot that is closest to already decided — two candidates, not five — and work there. Scan both the people and the evidence when you look for it; the tightest slot on the board is very often a piece of evidence while you were staring at suspects.
Then take one of the two candidates, assume it, and push hard until something breaks. If it breaks, you have not wasted the branch — you have proved the other candidate, which is a real deduction and not a guess. The board is built for this: moving chips around costs nothing and only a commit is a commit, so you can run the whole experiment on the board itself rather than in your head.
A real Monday, start to finish
This is the case published on Monday 13 July 2026 — a real one, not an illustration. It is solved completely below, so skip this section if you want to play it cold. Six rooms, two rows: Study, Pantry and Solarium along the top; Gallery, Library and Kitchen beneath them. Seven statements. Watch how little of it involves knowing anything about anybody.
Chain the orderings
Two statements share a name: the Pliers are left of Crowley and Crowley is left of Pemberton. Because left of means same row, these chain into one fact: Pliers, then Crowley, then Pemberton, all in a single row, in that order.
The grid is three wide. A strict three-long chain therefore fills the row completely and fixes all three columns at once — first, middle, last. Which row it is, you still have no idea. That is fine.
Notice the board is only two rows tall
The Pliers are above the Bow — same column, Pliers higher. On a two-row grid there is nothing to work out: above means top row, and the Bow is directly beneath. The Pliers were already in the first column from step one, so the Pliers are in the Study and the Bow is in the Gallery.
The chain now seats itself: Crowley in the Pantry, Pemberton in the Solarium. Three statements have decided half the board, and nobody has been told where anybody is.
Support-check, and let the absence do the work
The Brick is left of Hartshorn. Try each room Hartshorn could still occupy and ask where the Brick would go. Top row? Only the Study is free, and nothing is left of the first column. Gallery? Also first column — nothing to its left. Library? The Brick would have to be in the Gallery, which is already the Bow.
Only the Kitchen survives, which puts the Brick in the Library. Read that back: you just placed Hartshorn using a statement about the Brick, purely because everywhere else left the Brick homeless. That is the single most useful habit in the game.
Same move, other axis — then let the rooms fill themselves
The Revolver is above Plumm collapses the same way: the Kitchen is taken, and if Plumm were in the Gallery the Revolver would have to be in the Study, which is the Pliers. So Plumm is in the Library and the Revolver in the Pantry.
That leaves the last two statements to finish it. Larkin is next to the Pliers. The Study’s neighbours are the Pantry and the Gallery, and the Pantry is Crowley, so Larkin is in the Gallery and Bell takes the Study by elimination. Then Hartshorn is next to the Knife, which does the same on the evidence side: the Kitchen’s neighbours are the Solarium and the Library, the Library is the Brick, so the Knife is in the Solarium and the Rope in the Kitchen.
Both of those closing statements are the two-for-one from earlier — each also quietly told you Larkin wasn’t carrying the Pliers and Hartshorn wasn’t holding the Knife, which is exactly why the neighbour sets resolved to one room instead of two.
Done, with no hypothesising at any point. That is not luck — the case’s own build record confirms it was solvable by propagation alone, and most published cases are.
What to do when a case stops giving
Stalling in Alibi has a small number of causes and they are worth learning to tell apart, because the fix is different for each.
You read an ordering clue as a proximity clue. By far the most common. Left of does not mean beside — rooms can sit between. But it does mean the same row, which is the stronger half and the one people drop.
You’re tracking rooms and not people. If nothing on the board looks forced, try the other list. Ask each suspect and each piece of evidence how many rooms it has left. Somebody is usually down to one and nothing about their room advertises it.
You’re waiting for a clue that isn’t coming. There is no statement left to find for the last room or two. The final placements in most cases are made by elimination alone — nothing will ever name them. Placement by elimination is a real move, not a guess.
Two clues seem to contradict each other. They don’t. Every statement in the file is true simultaneously and exactly one arrangement satisfies all of them. A contradiction always means one of your own placements is wrong — back out the most recent thing you committed rather than re-reading the rail from the top.
And nothing here is on a clock or a life count. There is no fail state in Alibi: a wrong call simply doesn’t lock, the timer counts up rather than down, and the only thing a mistake costs you is the flawless finish. Being stuck is free, which means the thorough move is always the correct one.
Harder days have fewer clues, not more
This surprises people, and it changes how you should approach the back half of the week. Wednesday through Saturday are all played on the same nine-room board. Wednesday hands you twelve to fourteen statements. Saturday hands you ten or eleven — on an identical grid.
Saturday is harder precisely because it gives you less. More clues mean more slack: several routes into the case, and a few statements that are nice openers you could have done without. Strip that away and every remaining statement has to carry weight, the entry points narrow to one or two, and the chains you have to build get longer before they pay out. The same step happens on Tuesday, which is a tighter Monday on the very same six-room board.
The practical consequence: on a late-week case, do not go looking for the easy opening. There often isn’t one. Start support-checking immediately, and expect the first real placement to take three or four statements working together rather than one. Sunday is the only day that grows the board — twelve rooms — and it is the one day where the extra statements are a reflection of size rather than generosity.
If you want the reasoning without the casework, the grid version next door runs the same instinct on a smaller surface, and the no-guessing rule behind both is written up separately.