Climb the ladder in order
Sudoku techniques form a ladder, and the rungs are ordered by how often they actually finish a grid: the low rungs — scanning and singles — resolve most of an easy or medium puzzle, while the exotic patterns exist for the grids the low rungs cannot crack. Skipping rungs does not just slow you down; it hides which technique the grid is actually asking for.
The catalog's Sudoku Master carries a Hard difficulty label, which in practice means the mid-ladder rungs below appear regularly. Learn them in this order and each one builds directly on the marks the previous one left behind.
Scanning and singles
Cross-hatching is the first rung: pick a digit, look at where it already sits, and rule out the rows, columns and boxes those positions touch. A three-by-three box with only one remaining cell for that digit takes it; a row or column works the same way. Most solvers alternate a few digits rather than sweeping one to exhaustion.
The second pattern is the last cell in a unit: count what a row, column or box is missing, and if a cell has exactly one legal candidate left, place it. These two patterns alone clear the opening of nearly every grid, and they leave no pencil marks to maintain.
Pencil-mark what you proved
Marks are fuel for every later rung, but honest marks matter: write a candidate only when you can point at the reason it is possible. Copying every possibility into every cell produces noise that hides exactly the patterns the mid-game depends on.
A workable middle path: mark a row, column or box only when it resists scanning, and update the marks the moment you place a digit. If a marked cell drops to one candidate, that is a single you already paid for.
Naked pairs clean house
When two cells in the same row, column or box both hold exactly the same two candidates, those two digits belong to those two cells in some order — so no other cell in that unit can hold either of them. Removing the pair's digits from the neighbors is often what collapses a stalled unit into singles.
The same logic scales to triples: three cells sharing three candidates among them, even in messy combinations like 2-7, 7-9 and 2-9, claim those three digits exclusively. Triples are rarer but worth checking whenever a pair removal leaves a unit still stuck.
Pointing pairs and claiming
When a digit's candidates inside a box all sit in one row or column, the digit must land on that line — so the line's cells outside the box lose the digit. The mirror image works too: if a digit in a row is confined to one box, the rest of that box loses it.
Pointing is the bridge between box-thinking and line-thinking, and it is the rung most self-taught players skip. On Hard-labeled grids it frequently unlocks two or three cells in a row before any pattern more exotic is needed.
X-Wing, and where the ladder ends
An X-Wing forms when a digit has exactly two candidate positions in each of two rows, and those positions share the same two columns. Whichever way the digit lands, it occupies one cell in each column — so every other candidate for that digit in those two columns can be erased. Swordfish extends the same rectangle logic to three rows and three columns.
These patterns are genuinely rare on human-difficulty grids, which is why they sit at the top of the ladder: reach for them only after pairs and pointing have swept the board clean. If a grid resists the whole ladder, the honest move is the undo button and a re-scan — a mis-typed early mark is more common than a missing fish.
Knowing when not to climb
The ladder has a stopping rule most guides skip: climb only when the grid asks for it. A grid that still yields to scanning and singles gives no feedback for X-Wing hunting — there is nothing to find, and the search feels like the puzzle resisting you when it is simply already solved at your current altitude. The tell is pencil marks: if honest marks produce no naked pairs and no pointing pairs, the grid is asking for the next rung; until then, every rung above the locked pair is a waste of the one resource the ladder consumes — attention.
This is also the honest answer to the intermediate plateau. Players who plateau are usually players who jumped to the top of the ladder and now apply exotic patterns to grids that singles would have finished, missing the simple placements the patterns were supposed to enable. Descend before you ascend: re-run the low rungs on every new grid, and climb only when the low rungs genuinely go quiet.
