My habit when I'm stuck: jump around

Borders first, as usual

No surprise here. I flip the pieces, pull out everything with a straight edge, and build the frame. It tells me how big the picture is and gives the rest of the puzzle somewhere to land.

Pick out the obvious pieces and set them aside

Next I go through the pile looking for anything that clearly belongs somewhere: a bit of a face, a bright colour, a line of text, part of a sign. I don't try to place them yet. I just set them to the side in little groups.

Later, those groups connect into small chunks almost on their own, and then the chunks find their spot inside the border.

Don't get stuck on one area, jump around

This is the big one for me. When an area stops giving, I don't sit there forcing it. I move to a completely different part of the picture and work there for a while.

It sounds like you'd lose focus, but the opposite happens. Every piece you place somewhere else is one less piece in the pile. When you come back to the hard area, there are fewer candidates, and your eyes are fresh. Pieces I stared at for ten minutes often click right away.

Connect the chunks later

By jumping around you end up with several half-built islands across the table. That's fine. As they grow, they start meeting each other, and the gaps between them get small enough to fill by shape.

Try the habit on your own picture

The idea is to switch to a part that offers a different kind of clue, not simply another handful of the same plain pieces. If the sky has stalled, look for lettering, a window frame or a strong colour boundary. Build a small group whose details you can recognise on the box. That gives you somewhere useful to put your attention while the difficult area waits.

You don't have to copy a strict order. On a conventional rectangular puzzle, the border is a helpful starting point; an unusual outline may call for a different approach. If a clear detail in the middle is ready to assemble before the last edge turns up, work on it. An unfinished border doesn't mean the whole session has to stop.

Keep the islands movable

Use the reference picture to leave a small section near its likely place, with room to adjust it. Check its position against nearby landmarks rather than expecting your first estimate to be exact. A roof and a tree may eventually connect through a narrow strip you have not built yet.

Move sections gently on a supported surface. Avoid lifting a large unglued island by one edge, since joins can come apart. If the table is crowded, keep loose groups in shallow trays and reserve some working space for sections. Jumping between areas is less helpful if every switch buries the pieces you were using.

Return with a question, not the same random search

When you come back to the difficult area, see what has changed. There may be fewer candidates, another neighbour beside the gap or a longer printed line to follow. Ask which connectors the gap requires, which way the detail runs and whether a boundary piece belongs nearby. A specific question gives you a reason to reject or try a candidate.

If no piece fits, check the surrounding joins before repeating the whole pile. A plausible but wrong piece can leave a gap that is impossible to fill. Look for an interrupted line, an unusual seam or a piece that doesn't sit flat. Correcting that assumption is useful progress even if you have to take a section apart.

Leave yourself an easy restart

Before taking a break, leave the current group visible and keep it separate from unrelated pieces. You can make a short note such as ‘check the roof-to-sky boundary next’ if several areas are in progress. The next sitting then begins with a manageable choice instead of a complete re-sort.

Some sessions end with a connected island; others end with a better-organised pile. Neither needs to become a test of speed. This habit is about keeping the physical puzzle enjoyable when one patch stops offering clues, and leaving yourself free to return to it later.

And if one piece still won't go

Sometimes a single piece beats me even after all that. That's what I made PuzzlePal for: snap the piece and it suggests the row and column on the box picture where it probably belongs. Then I go back to doing it myself.