Practical guide
A practical solving routine for difficult puzzles
A repeatable scan-and-update routine for more challenging Sudoku puzzles.
Pass 1: collect the easy facts
Start with the boxes, rows or columns that contain the most given numbers. Find naked singles: a square with exactly one candidate. Then check for hidden singles: a digit that fits only one square in a region.
Whenever you fill a square, update its row, column and box. Repeat this scan before applying more advanced techniques, since a new placement often reveals another single.
Pass 2: compare candidate patterns
After the basic singles stop producing progress, write candidate lists for unresolved cells. Check for naked pairs, especially two squares in a shared unit with identical two-digit lists. When you can eliminate paired digits from the other cells in that unit, update candidates and scan for new singles again.
Keep the candidate lists accurate. In a particularly dense board, making a table of missing digits by row can prevent repeatedly doing the same mental work.
| 5 | 3 | 7 | ||||||
| 6 | 1 | 9 | 5 | |||||
| 9 | 8 | 6 | ||||||
| 8 | 6 | 3 | ||||||
| 4 | 8 | 3 | 1 | |||||
| 7 | 2 | 6 | ||||||
| 6 | 2 | 8 | ||||||
| 4 | 1 | 9 | 5 | |||||
| 8 | 7 | 9 |
Pass 3: make every decision explainable
For each proposed entry, ask: Which row, column, box or candidate pattern forces it? If you cannot answer, mark it as a possibility and move on. A short break can help you return with a fresh scan rather than repeatedly retrying the same square.
Use the hint tool to compare your reasoning with a simple naked or hidden single. It will say when those basic methods find no move rather than inventing an explanation.
Additional worked reasoning and practice
Use a repeatable cycle: update all candidate lists, solve naked and hidden singles, inspect locked candidates and pairs, then move to wings or fish only when justified.
When an advanced deduction removes one pencil mark, return to simpler techniques. A single new elimination can unlock a full chain of easy placements.
Deeper explanation and guided self-check
A staged routine for hard grids
First scan for naked and hidden singles in short rows and constrained boxes. Next look for locked candidates and pairs, since those eliminations often create new singles. Use triples or fish patterns only when their exact supporting candidate positions have been verified.
Record eliminations before placements
A correctly applied X-Wing may remove a candidate without solving a cell immediately. That is still progress. Make the elimination, update the affected notes, and rescan for simpler deductions before applying another advanced pattern.
A checkpoint when stuck
Ask whether the grid has been updated consistently and whether a row, column or box now has only two places for a digit. If nothing follows, review the explanations for the last technique you used; a missed candidate may invalidate every later elimination.
Go further: detailed reasoning and practical examples
Use a predictable scan order
Begin with rows, columns and boxes that have many fixed values, then look for a missing digit whose possible locations are few. Recompute candidates only in the units affected by proved placements. When neither cell-first nor digit-first scanning yields a single, move to pairs and locked candidates before seeking advanced patterns.
Keep the method proportional to the puzzle
Do not start searching for Swordfish in a puzzle with obvious singles. Harder techniques depend on complete candidate information, so an early advanced-pattern search with sparse notes is more likely to produce an incorrect inference. If an advanced pattern is found, verify every source cell and target before erasing a candidate.
Build a repeatable review cycle
After each deduction: state the rule, make the one justified change, update the peers, and recheck singles. If no progress appears, select a different unit and inspect its missing digits. A useful session can end after demonstrating one complex elimination, even if the whole puzzle remains unfinished.
Illustrative candidate positions are not universal puzzle solutions; verify all constraints on your current board.