Solving techniques · An interactive read
A Sudoku Strategy Roadmap: Choose the Next Useful Question
Build a practical solving loop from singles, reliable candidates and eliminations, with clear stopping points and no promises about universal difficulty levels.
The idea to take away
Use a search order as a guide, then return to simple moves whenever the board changes. Technique names do not form a universal difficulty ladder.
Use a route, not a rank
When a board stalls, a list of technique names is less useful than a question you can answer. Is there a cell with one candidate? Does a missing digit have one location? Do a few cells reserve a set? This roadmap gives those questions an order so you can search deliberately.
It is a suggested routine, not a promise that every puzzle will yield to the same short sequence. HoDoKu's index groups many techniques, from singles and intersections to chains and fish. [1] You can learn one family at a time without assuming its position in a list defines your skill or the puzzle's exact difficulty.
Follow the loop when the board changes
Begin by checking the active rules and the reliability of your notes. Look for forced placements. If none appear, inspect candidate relationships that justify deletions. After any deletion or placement, return to the affected cells and houses.
The return step is crucial. You do not graduate out of singles after learning pairs. A more elaborate deduction can create a simple placement immediately. HoDoKu's introduction describes candidate reduction as a way to expose singles. [3]
The diagram keeps this loop compact. It does not say that an unsuccessful pair search means an X-Wing must exist. The right next question depends on what information is present, and sometimes a hint is the most useful way to identify the missing idea.
First, identify an actual placement
Inspect a nearly complete house or follow one digit through intersecting rows and columns. If a target cell has one remaining digit, or a missing digit has one location in a house, name that reason and place it.
You may use notes from the start or add them when mental tracking becomes inconvenient. Neither choice changes the proof. What matters is knowing whether a candidate list is complete enough for the technique you are applying.
If you repeatedly miss a particular kind of single, review one example of that type. Avoid treating the entire unfinished puzzle as evidence that you need a more advanced strategy. A local misunderstanding needs a local explanation.
Check the information before the pattern
A pair built from incomplete notes is not a reliable pair. Before a candidate-based deduction, verify the relevant lists and the shared house. Before a location-based deduction, verify that the selected digit really has no extra location in the source house.
Separate a failure to find a pattern from a failure to maintain the board. If an old note survived a recent placement, update it before searching farther. If a candidate was deleted without a reason, restore a known correct state and revisit that step.
This checkpoint is especially helpful after a pause. Reconstruct what the notes mean before using them as premises. A quiet-looking cell may be partly annotated rather than tightly constrained.
Let one elimination lead back to singles
Our row example contains a naked pair {3,8} in c2 and c5. Another cell, c4, lists {3,6,8}. The pair reserves 3 and 8, so remove them from c4; 6 remains. You have moved from a two-cell relationship to a one-cell placement.
Place that 6 and then remove 6 from c8, which previously listed {6,7}. Now c8 becomes 7. The pair's own order remains unknown. The diagram deliberately stops after the first elimination so you can identify each later consequence separately.
This is the roadmap in miniature: verify candidates, notice a relationship, delete only what the proof allows, and return to singles. You do not need to keep hunting pairs while a forced 6 is waiting.
Treat difficulty labels as local conventions
Stuart explicitly calls difficulty ordering subjective in his strategy-family guide. [2] A pattern may be easy to recognise in one drawing and hard to locate among many unrelated candidates. Labels from different apps need not describe the same solving path.
A useful readiness check is whether you can explain a pattern's premise, target and failure case. For a pair, identify the two restricted cells and say what an extra candidate would change. For a locked candidate, identify the source house and the receiving house.
That check is about understanding, not a requirement to master every simple pattern perfectly before reading an interesting new one. Use curiosity to choose what to study, then verify the logic on a concrete example.
Review a decision instead of inventing a score
After a solve or a hint, record one useful observation: the missed house, the stale candidate, or the extra location that invalidated a tempting pattern. This is a suggested reflection exercise, not a validated performance metric.
The quiz follows the same approach. It asks what information is needed and when to return to a simpler move. For a next reading step, choose the guide that matches your current question rather than the one with the most imposing title.
Go to the evidence
Sources & further reading
Research checked on 14 October 2026. Follow the original sources to explore the evidence and its limits.
- Bernhard Hobiger — HoDoKu solving techniques
The solver author's technique index and caution that terminology is not universally standardized.
- Andrew Stuart — Strategy Families
A solver-specific ordering of techniques, useful as an example rather than a universal difficulty scale.
- Bernhard Hobiger — HoDoKu technique introduction
Original solver documentation explaining candidates and the terminology used in human solving.
Keep exploring