Sudoku Strategy: The Order to Apply Techniques In

The thing that separates fast solvers from stuck ones is not knowing more techniques — it's applying the ones they already know in a fixed, repeatable order. Most people solve sudoku by scanning the grid semi-randomly, hoping something jumps out. That works on easy puzzles and falls apart the moment a puzzle has any real bite to it.

A strategy is an order of operations: cheapest technique first, most expensive last, and never move down the list while something higher up is still producing results.

Why Order Matters More Than Technique Count

Every sudoku technique has a cost — how much of the grid you have to hold in your head, and how long it takes to spot. Scanning a single box for a missing number is cheap. Hunting for an X-Wing across the whole board is expensive.

The mistake almost everyone makes is jumping to an expensive technique while cheap ones are still available. You spend four minutes hunting for a pattern that isn't there, when a thirty-second scan would have handed you three cells for free — and each of those cells makes the rest of the puzzle easier.

There's a compounding effect here that's easy to underestimate. Every number you place removes a candidate from up to twenty other cells. Place three numbers and you've altered the constraint picture across a third of the grid. A pattern that genuinely did not exist five moves ago may now be sitting in plain sight.

The rule: after every single placement, go back to the top of the list. This feels inefficient and is the opposite. Solvers who restart religiously finish faster than solvers who push forward through a technique they've already committed to searching for.

One refinement makes the restart cheaper. Before sweeping the whole grid again, check the three units the digit you just placed belongs to — its row, its column, its box. Placements come in chains, and the next link is nearly always in one of those three units. Clear the immediate cascade first, then restart at the top.

The Order of Operations

1. Full scan for obvious placements (cheapest)

Go number by number, 1 through 9. For each one, look at the boxes where it's still missing and check whether there's only one legal cell left. Start with whichever numbers already appear most often on the board, since they have the fewest remaining homes.

Do this for all nine digits before you do anything else. On an easy or medium puzzle, this alone may finish the grid entirely.

The technique underneath this is usually called cross-hatching: for a given number and a given box, mentally strike out every row and column that already contains that number. Whatever cells survive are the candidates. If exactly one survives, you have a placement.

2. Fill in complete pencil marks

Once scanning stops producing placements, stop scanning. Write every legal candidate into every empty cell.

This is the single highest-value ten minutes you will spend on a hard puzzle, because every technique below this line depends on accurate candidates. Do it systematically — cell by cell, left to right, top to bottom — rather than jumping around. Missed candidates cause errors that surface twenty moves later, and by then you have no idea which placement was wrong.

Resist the urge to start solving while marking. Complete the pass first. Spotting a naked single halfway through and chasing it means the remaining cells get marked against a grid that has since changed.

Write the candidates in fixed positions while you're at it — 1 always in the same corner of the cell, 9 always in the same corner. It costs nothing and it converts the next three steps from reading into pattern recognition: two cells with marks in identical positions announce a naked pair from across the grid, without you having to compare their contents digit by digit.

3. Naked singles

Any cell with exactly one candidate left gets filled. Then go back to step 1.

These are the cheapest deductions available once marks are in, and a full pencil-marking pass usually produces several immediately.

4. Hidden singles

Within each row, column and box, look for a number that only appears as a candidate in one cell — even if that cell has four other candidates listed.

This is the technique people miss most often, because it requires looking at where a number can go rather than at what a cell can be. Your instinct after marking is to scan cells; hidden singles require scanning numbers. It's a deliberate change of direction and it has to be consciously chosen.

The drill: take the digit 1, check all nine rows, all nine columns, all nine boxes. Then 2. Tedious the first few times, fast within a week.

5. Naked pairs, triples and pointing sets

Two cells in the same unit sharing exactly the same two candidates means those two numbers belong to those two cells, and can be erased everywhere else in that unit. The same logic extends to triples, including the partial form where three cells hold {1,4}, {4,9} and {1,9} between them.

Pointing pairs work the other way: if a candidate within a box only appears in cells that all share a row or column, that number can be erased from the rest of that row or column outside the box. Box-line reduction is the mirror image — a candidate confined within a row to cells inside one box can be erased from the rest of that box.

Two things make this step faster. Hunt in the fullest units first — a box with seven digits already placed has few empty cells, so a pair there is easy to see and eliminates a lot. And don't skip hidden pairs on the grounds that they're hard to spot: a hidden pair strips the other candidates out of two cells, which usually turns them into a naked pair, which then eliminates again. They do more work than they look like they do.

More detail on all of these in hidden singles, pairs and triples.

6. Advanced patterns (most expensive)

X-Wing, swordfish, empty rectangle, chains. These work across multiple rows and columns at once and take real effort to spot. Only reach for them when everything above has genuinely run dry — see advanced techniques for how each pattern works.

The X-Wing is the one worth being able to reconstruct from first principles, because it's the first genuinely two-dimensional pattern and most descriptions of it are needlessly opaque. Pick a digit — say 4. Find two rows in which 4 has exactly two candidate cells each. If those two cells in the first row sit in the same two columns as the two cells in the second row, the four cells form the corners of a rectangle and you have an X-Wing. It works because in the first row the 4 is in one of two corners, and in the second row the 4 is in one of two corners, and they cannot both land in the same column without leaving one of the rows without a 4. So whichever way it resolves, each of those two columns already has its 4 accounted for, and 4 can be erased from every other cell in both columns. Run the same argument with rows and columns swapped and it works identically.

How to actually find one, without paying full price for the search. Do not scan the whole grid for rectangles. When you're stuck, take the digit with the fewest remaining candidates overall and list only the rows where it has exactly two candidate cells. There are usually three or four such rows, and checking them against each other takes seconds. That single habit is the difference between the X-Wing being a party trick and being a working tool, and it is what keeps step 6 from costing the fifteen minutes the cost table budgets for it.

The Cost Table

Matching Strategy to Difficulty

Not every puzzle needs the full sequence, and knowing where to start saves real time.

Easy. Step 1 alone. If you find yourself pencil-marking an easy puzzle, you've either missed something in the scan or the puzzle is misrated.

Medium. Step 1, then partial marks on the most constrained region rather than the whole grid. Full marking is usually overkill.

Hard. Steps 1 through 5. Full pencil marks are worth the time here and the puzzle is unlikely to yield without them.

Expert. Same as hard, but expect to cycle through steps 3-5 repeatedly. Each pass produces one or two placements rather than a cascade.

Extreme. All six steps, with step 6 required at least once. Budget an hour and expect to be stuck more than once.

The practical failure is applying hard-puzzle strategy to an easy puzzle (slow, tedious) or easy-puzzle strategy to a hard one (stuck within five minutes, no idea why).

The Three Habits That Cause Most Failures

Not restarting after a placement. You place a 7, then keep hunting for the advanced pattern you were mid-search on. Meanwhile that 7 just created two naked singles you walked straight past. Always restart at step 1.

Sloppy candidate maintenance. When you place a number, you must immediately erase it from every cell sharing its row, column and box. Skip this once and every deduction afterwards is built on a lie. This single habit accounts for the large majority of "the puzzle is broken" complaints — the puzzle is fine; the pencil marks aren't.

Guessing when stuck. A properly constructed puzzle never requires a guess. Being stuck means a technique hasn't been applied, not that the logic has run out. Guessing on a hard puzzle usually produces a grid that looks fine for twenty more moves and then contradicts itself, with no way to tell where it went wrong.

What to Do When You're Genuinely Stuck

Work down this list before considering the puzzle beaten:

1. Re-verify your pencil marks on one full box. If you find even one error, redo them all — an error in one cell means you may have made errors elsewhere the same way.
2. Do a hidden-single sweep by number, not by cell. Take the digit 1, check all nine rows, all nine columns, all nine boxes. Then 2. Most missed deductions live here.
3. Look for the most constrained unit — the row, column or box with the fewest empty cells — and work only inside it.
4. Check the digits you've placed least often. A number appearing only three times on the board has six homes left and is more likely to yield than one that's nearly complete.
5. Verify that every digit already on the board is legal. Roughly half of genuinely "impossible" positions are transcription errors — a digit written into the wrong cell forty moves ago. Check every placement against its row, column and box before you go hunting for an exotic pattern that the grid may no longer contain.
6. Rotate the grid ninety degrees. On paper, physically turn it; on screen, tilt your head. Column patterns then read like row patterns, and human vision is measurably better at scanning horizontally. It sounds silly and it works.
7. Walk away for ten minutes. This is not a joke; pattern recognition genuinely recovers with a short break, and experienced solvers report that a puzzle abandoned in frustration often falls apart in under a minute on a fresh look.

Speed vs. Accuracy

Worth being explicit about, because they pull against each other.

Solvers chasing speed skip the full pencil-marking pass and rely on partial marks and intuition. It works on medium puzzles and produces errors on hard ones — and an error on a hard puzzle costs far more time than marking would have.

Solvers chasing accuracy mark everything, verify twice, and finish slowly but reliably.

For anything rated hard or above, accuracy is faster. The time cost of discovering a contradiction thirty moves in, with no way to identify the bad placement, dwarfs the ten minutes marking would have taken. Speed strategy belongs on puzzles where a mistake is cheap to recover from.

Reading the Grid Before You Start

Experienced solvers spend the first twenty seconds looking rather than solving, and it changes where they begin.

Count the givens. Somewhere in the high twenties or low thirties suggests an easy or medium puzzle. Low-to-mid twenties suggests hard or worse. It's a rough signal but it tells you whether to expect a scan-only solve or to prepare for marking.

Look at the distribution. Clues spread evenly across all nine boxes generally mean a gentler puzzle. Clues clustered heavily in three or four boxes, leaving others nearly empty, usually mean the empty regions will resist for a long time and the solve will run through the dense areas first.

Find the most-placed digit. Whichever number already appears six or seven times has only two or three homes left. That's your opening move, every time, and it's a far better start than beginning at 1 out of habit.

Spot the fullest box. A box with seven of nine cells filled will often resolve immediately and cascade into its neighbors.

None of this is solving. It's deciding where to point the first technique, and it reliably saves more time than it costs.

Building the Habit

The sudoku techniques and tips above only become an advantage once the order is automatic. The fastest way to get there is to solve deliberately slowly for a dozen puzzles — announce to yourself which step you're on, finish it completely, then move down. It feels tedious and it will roughly halve your time within a couple of weeks, because you stop wasting minutes on expensive searches while free cells are sitting there.

Practicing against puzzles of a known, consistent difficulty helps too. A sudoku puzzle maker lets you generate sets at one specific level so you can drill the same technique repeatedly instead of getting a random mix that never builds the pattern. Ten hard puzzles in a row teaches more than ten puzzles spanning easy to extreme.

Target the practice at one rung at a time. Hidden singles turn up throughout easy and medium grids; pointing pairs appear reliably in medium-to-hard; an X-Wing is a hard-grid feature and you can go several puzzles without meeting one, which is exactly why generating a batch beats waiting for the daily paper to cooperate. Free online sudoku makes that possible in a way a printed book never did — ten medium grids in one sitting, hunting the same technique in all of them, is a session you simply could not assemble when free sudoku meant one puzzle a morning.

One last habit, and it's the cheapest of all. When you finish a hard grid, go back and name the single move that broke it open. There is almost always exactly one — a hidden single in box 4, a pointing pair on row 7. Naming it is what converts a puzzle you solved into a technique you own.

A Sudoku Cheat Sheet: Trigger and Action for Every Technique

The cost table above says when each technique is worth the time. This one says what to look for and what you are allowed to erase when you find it — the two things people actually forget mid-solve. It fits on an index card, and keeping it beside the grid for the first few weeks is a legitimate shortcut to making the order automatic.

Two reminders belong on the card itself: after any placement, return to the top of the list; and after any elimination, re-check singles before hunting for another pattern.

Step-by-Step Solvers: Using One Without Blunting Your Own Solving

Automatic solvers that show their working, usually sold as a sudoku solver step by step, are genuinely useful and also the fastest way to stop improving. The difference is entirely in when you reach for one.

Most of the tools people use to play sudoku online now include one. They accept a grid and then reveal the solution one deduction at a time, naming the technique used — "hidden single, row 4," "pointing pair in box 7." The good ones apply techniques in roughly the order described here, which means the step they show you is usually the cheapest available deduction rather than an exotic one.

Use it after the solve, not during it. The productive workflow is to solve until you are genuinely stuck, note the grid state, then let the solver take one step. Nine times out of ten it names a technique you already know and points at a spot you failed to scan — which tells you something specific about your own routine rather than just handing over a digit.

Set the hint level as low as it will go. "There is a deduction available in box 3" teaches far more than a filled-in cell. If the tool only offers full answers, cover the answer and use the location alone.

Use it to verify, not to rescue. Feeding in a half-solved grid to check whether you have already made an error is the single most valuable thing these tools do, because a broken grid can absorb an hour that no technique will ever fix.

Frequently Asked Questions

What is the single most useful sudoku strategy? Working through techniques in a fixed order, cheapest first, and restarting at the top after every placement. Solvers who do this consistently outperform solvers who know more techniques but apply them haphazardly.

Should I pencil mark every puzzle? No. On easy and medium puzzles, scanning alone usually finishes the grid, and marking wastes time. Start pencil marks the moment plain scanning stops producing placements — which on a hard puzzle is usually after the first few minutes.

Why do I keep making mistakes on hard puzzles? Almost always because a candidate wasn't erased after a placement. Every number you write must be immediately removed from the candidates of every cell in its row, column and box — no exceptions, every single time.

How long should a hard sudoku take? For most people, somewhere between fifteen and forty minutes once the strategy is habitual. Expert and extreme puzzles requiring advanced patterns can reasonably take an hour or more, and that's normal rather than a sign of doing something wrong.

Is it faster to solve quickly or carefully? On hard puzzles, carefully is faster. An error discovered thirty moves later usually can't be traced back to its source, which means restarting — a far bigger cost than the ten minutes that complete pencil marks would have taken up front.

What are the most important sudoku strategies to learn? In order of payoff: cross-hatch scanning, complete pencil marks, hidden singles, naked and hidden pairs, and pointing pairs with box-line reduction. Those five carry almost every puzzle up to hard. X-Wing and beyond only start earning their keep at expert level.

Are step-by-step sudoku solvers worth using? Yes, if used after you are stuck rather than instead of getting stuck. Take one hint at the lowest detail level, see which technique it names, and finish the grid yourself — the useful information is which deduction you missed, not the digit itself.

Is it worth printing a sudoku cheat sheet? For the first month or two of learning intermediate techniques, genuinely yes. A single card listing each technique's trigger and its permitted elimination prevents the most common failure, which is spotting a pattern correctly and then erasing the wrong candidates.