Griductive Strategy
Griductive will not let you guess, which means every grid is winnable by pure reading and none of it is luck. The skill is spotting the verdict the clues have already decided — usually before you notice they decided it.
Every statement on the board is load-bearing
There is no dead weight in a Griductive grid. If a card makes a claim about somebody, that claim is one the grid cannot be solved without — it survived a pass whose whole job is to find statements that aren’t pulling their weight and take them off the board. So you never have to wonder whether a clue is a red herring. It isn’t. Nothing on the rail is there to waste your time.
The corollary is the thing to carry into a grid: not every card is a statement, but every statement is one you need. Some of the cast are simply talking — saying something in character that doesn’t constrain anything. That’s the cost of the household feeling like a household rather than a spreadsheet. Learn to tell the two apart quickly, because time spent squeezing a deduction out of small talk is the one kind of effort a grid will never repay.
“Neighbours” and “connected” are not the same shape
This is the single richest source of difficulty in the game and the thing most likely to make a grid feel impossible when it isn’t. Griductive uses two different ideas of adjacency, and a clue can use both at once.
Neighbours means all eight — the full ring around a person, diagonals included, and never the person themselves. Connected means four — up, down, left, right. Diagonal contact does not connect.
Now read a statement like all of Lucy’s innocent neighbours are connected again. The group it is talking about is Lucy’s eight-ring. The test it applies to that group is four-directional. So two innocents sitting directly above and directly right of Lucy are both her neighbours — and they are diagonal to each other, so they are not connected, and the statement is false. Whole grids turn on that distinction. It is worth re-reading any clue with both words in it twice.
Five ways a cell gets forced
The board decides whether a verdict is provable by enumerating every line-up still consistent with what you have unlocked — which is exponential work, and not something you can do at the table. You don’t need to. Every forced cell has a local reason, and in practice it is nearly always one of these five.
1. Read the subject, not the count
Some statements name a person and a number in the same breath: Hazel is one of the 2 criminals neighbouring Lucy. The number is the eye-catching part and it is the less useful part. That sentence asserts, flatly and before any counting, that Hazel is a criminal.
This is the only kind of clue that pins a named person outright, so harvest it the instant it appears. Then take the count as a second, separate fact — it usually tells you how many of the remaining neighbours are innocent, which is the setup for the next technique.
2. Intersect the two groups, then check for saturation
The highest-yield move in the game. Any statement about two groups — exactly two of the three innocents around Lucy are in row four — is really a statement about the cells the two groups share. Work out that overlap first, before anything else.
If the overlap contains exactly as many cells as the number demands, every cell in it is pinned. If the number is zero, every cell in it is pinned the other way. Clues of this shape are about a quarter of everything on a board, and the overlap is often only two or three cells wide — so this converts a vague-sounding sentence into two locked verdicts in one step.
3. Treat exact counts as budgets, then starve the outsiders
Every bare number in Griductive is exact. Not “at least” — exact. So a statement about a group is a budget: this many, no more.
Spend the budget on what you already know, then see who is left claiming it. If the remaining budget is entirely consumed by some sub-region — because another clue demands it there — every cell outside that sub-region is forced to the opposite. You proved a verdict for people no clue ever mentioned. That is the move that cracks the middle of a hard grid.
4. Turn shape statements into position statements
Both innocents in column B are connected: two facts at once — exactly two, and orthogonally touching. In a single column, “touching” can only mean consecutive. So the clue has quietly become “the two of them sit next door to each other,” which is a much more concrete thing to work with.
List the surviving adjacent pairs and read them two ways. Anyone who appears in every surviving pair is forced in. Anyone with no possible partner left is forced out. Both fire constantly, and the second one is the one people miss.
5. Parity is always odd
Parity clues never name anybody, which makes them easy to skip. Don’t — they halve the space, and paired with any count they pin the last unknown outright.
One shortcut worth knowing: the only parity you will ever see is odd. Even-numbered parity statements exist in the engine but are retired and never ship. If a clue mentions parity, it is telling you the count is odd.
Three things that are not clues, and one that is
Professions are noise. A judge is no more suspicious than a singer. Jobs exist so clues have something to point at; unless a statement names a profession, it tells you nothing.
The difficulty chip is the weekday, not a measurement. It says what tier was requested when the grid was built, not how the finished puzzle graded. A “Tricky” label means it’s Thursday.
The grid never changes size — Monday and Sunday are both four across and five down, and both start you with exactly one card face up. Sunday is not a bigger board with more to read. It is the same board asking for longer chains.
But the names are a tool. They run alphabetically, left to right and top to bottom. When a clue names someone you can find them instantly instead of scanning — which matters more than it sounds when you are re-reading a rail for the fifth time.
One more asymmetry worth carrying: the clues talk about innocence about twice as often as guilt. If you have been hunting for criminals, you have been reading against the grain of the evidence.
What a refused call actually tells you
Griductive will not let you lock a verdict the clues don’t force, and it refuses in two different ways. The difference is worth learning, because one of them is feedback and the other is nearly an answer.
“Conclusion not possible” means this cell is not forced yet. It says nothing whatsoever about whether you were right — a correct-but-early call and a wild guess produce exactly the same message. What it tells you is that your target was wrong, not your verdict: stop working this person and go unlock testimony elsewhere.
“That’s not it” is a different animal, and it is the more useful of the two. It means the proof was already on the board and you had everything you needed — so this is not a gap in your evidence, it’s a misreading. Nine times out of ten the word is connected, or a number you read as “at least” when it was exact. Go back to the clue you were leaning on and read it again, literally.
One thing to know, because it is genuinely counter-intuitive: a correct call made too early is recorded exactly like a wrong one. Being right before you could prove it still marks that suspect on your share card. The grid is not scoring your accuracy — it is scoring whether you did the work. There are no lives behind any of this and nothing ends the run; the share card is simply the only thing that remembers.
Four rounds of a real grid
The grid from 1 June 2026 — a Monday, twenty people, and exactly one card face up at the start: Noah’s. The opening is solved below; skip it if you’d rather not know.
The only card face up hands you a person
Noah says Hazel is one of the 2 criminals neighbouring Lucy. Read the subject: Hazel is a criminal. Done — no counting, no grid work, on move one.
Then bank the count separately. Lucy sits against the right edge, so her ring is five people, not eight. Two of them are criminals and Hazel is one — so exactly one criminal hides among the other four, and three of them are innocent.
Flip Hazel, then intersect
Calling Hazel turns her card: exactly 2 of the 3 innocents neighbouring Lucy are in row four. Take the overlap. Lucy’s ring meets row four in exactly two cells — Orla and Paul.
The clue demands two innocents in that overlap, and the overlap holds exactly two people. It is saturated. Orla and Paul are both innocent, forced, in one step.
A shape clue, read in both directions
Paul’s card: both innocents in column B are connected. Column B is a vertical stack, so connected means consecutive, and “both” means exactly two. Noah is in column B and already known to be a criminal, which leaves four candidates — and only two consecutive pairs survive among them.
Fern appears in both surviving pairs, so Fern is innocent. And Roy sits at the bottom with only Noah above him — a criminal — so Roy has no possible partner at all, which forces Roy criminal. Two verdicts, opposite directions, same clue.
Starve the outsiders
Roy’s card constrains Brian’s ring to exactly two innocents, and Fern — just proved innocent — is one of them. So one innocent remains among four people. A second clue pins that last innocent inside a two-cell overlap.
The budget is now fully spent inside that overlap, so everyone outside it is starved: Anya and Enzo are both criminals. No clue ever mentioned either of them.
Four rounds, four different techniques, and the whole grid unlocks in twelve. Every card you flip is both an answer and a new premise — which is why the rail is worth re-reading from the top after every single lock.
Same grid, longer chains
Griductive’s ladder does something unusual: nothing about the board changes. Every daily is four across and five down, from Monday to Sunday, with the same twenty people, the same one starting card and the same clue vocabulary. Easy and Evil draw from an identical set of sentence types.
What changes is how the steps interlock. An easy grid is built to favour moves where a single newly-revealed clue forces the next verdict on its own — plain, legible, one thing at a time. A hard grid is built to favour the opposite: steps where the new clue does nothing by itself and only bites when combined with another card you just turned over. Same vocabulary, deeper chains, later payoffs. That is the entire difference between Monday and Sunday.
Which means the technique doesn’t change across the week — the patience does. On a late-week grid, a clue that appears to do nothing is the normal case, not a dead end. Pair it with something else before you dismiss it.
The full range of difficulty opens on 1 August 2026; before that every grid is built at tricky or below. The rules page has the complete clue glossary, and the no-guessing rule behind the gate is written up separately. If you want the same reasoning set in a house rather than on a grid, Alibi is next door.