The board has stopped moving and you've read every square twice. That's the moment everyone means by "stuck."
It isn't one thing. It's five, they have different fixes, and staring harder only helps with one of them. Here's how to tell which you've got, cheapest test first.
1. You haven't written anything down
The test: look at the grid. Are there pencil marks in it?
If not, this is almost certainly your problem, and it's the most common one by a wide margin. Holding candidates in your head works for about six squares. After that you're re-deriving the same eliminations every pass, forgetting them, and doing it again — which feels exactly like a puzzle with nothing left in it.
The fix: pick the emptiest-looking region and write out the candidates for every square in it. Not the whole grid, just one region. Most of the time something falls out before you finish.
The reason this works isn't mystical. A square holding "2 or 5" is invisible when it's in your head and obvious when it's on the board, because now you can see that the only other square that wanted a 5 just got filled.
2. You're asking the only question you know
The test: for the last five minutes, have you been going square by square asking "what goes here?"
That's one of two questions and it dries up first. The other one is: pick a region, pick a number, and ask "where in here can this number go?"
Same board, same rules, and a completely different set of things you notice. A square with four candidates is invisible to the first question and can be the answer to the second — that gap is the whole of hidden singles, and it's the single most common thing an intermediate solver is missing.
The fix: switch questions. If you've been hunting squares, hunt numbers. Take the digit the board already has most of and find a box that's still missing it.
3. You went wrong twenty minutes ago
This is the one that isn't really being stuck, and it's the one nobody checks first because it means admitting the last twenty minutes might be scrap.
The test: you've got a square with no candidates at all. Every number is ruled out by something.
(This one needs case 1 done first — you can't spot a square with no candidates until you've written candidates in. That's the main reason case 1 is worth two minutes even when you're sure it isn't the problem.)
That is not a hard puzzle. That is a contradiction, and a valid Sudoku cannot contain one, so a digit somewhere on your board is wrong. Nothing you do downstream will help, and every further move builds on the mistake.
The fix — and this is the part worth learning, because it's faster than starting over. Don't scan the whole grid. Take the empty square with no candidates and look only at the squares that share a region with it — its row, its column, its box. There are twenty of them, not twenty-four: eight in the row, eight in the column and eight in the box, but four of the box's overlap with the row and column you've already counted. Your wrong digit is almost always one of those twenty, because they're what did the ruling out.
For each one, ask the question you'd ask if it were empty — where else could this digit have gone? A digit you placed correctly will have a reason you can reconstruct. A digit you guessed won't, and the moment you can't rebuild the reason, you've found it.
If nothing in the neighbourhood is wrong, widen to the peers of those squares. Still faster than a restart.
A word on the app's mistake counter, since it's relevant here and it's easy to misunderstand. With "Show mistakes" on — which is the default — the app highlights conflicts as you make them and counts them toward a three-mistake limit. That catches the digit that lands on top of another one immediately.
What no conflict check can catch, ours or anyone's, is the digit that's simply wrong but hasn't bumped into anything yet. It sits there looking fine for twenty minutes and then strands you. That's the one this section is about, and finding it is a manual job.
4. You need a technique you don't have
The test: your candidates are written down, you've tried both questions, nothing contradicts, and there is genuinely no square with one candidate and no number with one home.
Then the puzzle is fine and it's asking for something you haven't met. This is a real wall and it's the only kind of stuck where "look harder" is honestly useless.
The fix: learn the next technique. In order, and each one takes a few minutes:
- Naked singles — the square with one number left.
- Hidden singles — the number with one square left.
- Pointing pairs — the number cornered onto one line, cleared from the rest of it.
Those three finish almost every Easy puzzle in our app — 205 of the 207 that ship inside it — and 16 of the 57 Mediums. If you have all three and you're still walled, the next names are naked pairs, hidden pairs and X-wing, all of them in order, with what each one buys you, in Sudoku techniques, in the order worth learning them.
How to tell this apart from being stuck for reason 2, since they look identical from the inside: if you've written every candidate in one region and no square there has fewer than two, and no digit missing from that region has fewer than two homes, that region genuinely has nothing a single can find. Two or three regions like that in a row and you're in case 4, not case 2.
5. You've been at it too long
The test: how long have you been looking at the same three squares?
The fix: put it down. Ten minutes, or overnight.
This sounds like the advice you give when you have no advice, but it's mechanical rather than mystical. You've built up a set of assumptions about which squares are worth looking at, and they're wrong or you wouldn't be stuck. That set is sticky. Walking away is the cheapest way to drop it, and the square you couldn't see is very often the first thing you spot coming back.
Nothing in Sudoku goes off. The puzzle will be in the same state tomorrow, and so will your notes.
The order matters
Run them in order, because the cheap tests rule out the expensive ones.
Two minutes writing candidates into one region settles case 1 and usually case 3 at the same time — the contradiction shows up the moment you write it. Switching questions costs another minute. Only after both should you conclude you need a new technique, because "I need a harder technique" is a much more flattering diagnosis than "I never wrote anything down," and it's wrong far more often.
That's the whole method: figure out which stuck you're in before you spend effort on it. Most of them cost under a minute once you know.
This is the longer answer to why there's no hint button. A hint would have solved one square in case 4 and told you nothing about the other four.