Rank a short list by comparing just two neighbors at a time, swapping until nothing moves.
Ranking eight items at once is hard. You hold them all in your head, juggle, change your mind and end up with an order you don't quite trust. Swap Sort breaks the job into the easiest judgment there is: which of these two comes first? You only ever compare two neighbors, and you swap them if they're in the wrong order.
Lay the items out in a line, then work from top to bottom, comparing each pair of neighbors and swapping where needed. When you reach the bottom, go back to the top and do it again. When you can make a full pass without a single swap, the list is ranked.
Use it when you need a clear priority order for a short list, when ranking everything at once feels overwhelming, or when a group can't agree on what should come first.
Work down the line one pair at a time, then loop back to the top. A pass with no swaps means you're done.
Choose a part of the diagram or a tab to explore each step.
Of these two, which ranks higher on your criterion?
Is the lower item the stronger one? Then swap the pair.
What's the next pair down the line?
Did you make any swaps on this pass? If so, start again from the top.
Swap Sort can look slow on paper. In practice it's often the quickest route to an order people actually trust.
Comparing two things is far easier than placing one among ten. Your attention stays on a single pair, so each decision is quick and clear, and you're less swayed by whatever you happened to look at last.
Nobody can keep eight items and all their trade-offs in mind at once. Swap Sort never asks you to. The line of cards holds the ranking for you, and you only ever think about two items at a time.
A wrong call on one pass isn't permanent. Items keep meeting new neighbors, pass after pass, so a misplaced item gets another chance to rise or fall. The final order reflects many small judgments, not one big guess.
A group can gather around one line of cards and debate one pair at a time. Disagreements stay small and specific (“is A really more urgent than B?”) instead of turning into an argument about the whole list.
Agree on the question you'll ask of every pair, such as “Which will cut customer wait times more?” One criterion is ideal. If several matter, agree which one is in charge and use the others only to break ties.
Write each item on its own card. Shuffle the cards so the starting order carries no hidden bias, then lay them out in a single column. Five to twelve items work best.
Starting at the top, compare the first two cards. If the lower one ranks higher, swap them. Then compare cards two and three, then three and four, all the way to the bottom.
Go back to the top and run another pass. Keep going until you complete a full pass without a single swap. That's your ranked list.
Read the list from top to bottom. Does the order feel right? If two neighbors look wrong, compare them once more. Then draw a line under the items you'll act on.
Index cards or sticky notes, a wall or table to lay them out in a column, a marker and 15–30 minutes. With AI, use the prompt below.
Team: a six-person IT help desk at a mid-sized company. Criterion: “Which fix cuts average wait time more?” Items: seven process fixes, shuffled onto cards.
Form redesign, known-issues page, new ticket categories, password self-reset, backlog purge, triage rotation, remote-desktop tool.
New ticket categories sank all the way to the bottom in a single pass. Password self-reset moved up one place.
Password self-reset and the triage rotation each climbed one place. The backlog purge sank two.
Password self-reset reached the top. Triage rotation vs. form redesign took five minutes of debate, the longest call of the day.
One last swap put the triage rotation above the known-issues page. Pass 5 made no swaps, so the list was done.
1. Password self-reset. 2. Triage rotation. 3. Known-issues page. The team drew its cut-off line under number three.
Swap Sort is a long series of small, clear decisions, which suits AI well. It can run the passes, keep track of the order and argue each close pair from both sides. But the judgment in every comparison should come from you or your team, because only you know what really matters in your work.
Let AI keep the list and the swap count while you make every call.
Run a swap sort on my list. Show me one pair at a time, ask which ranks higher, and track the order.
When a pair is too close to call, hear both sides before you decide.
Make the best case for each: will the triage rotation or the form redesign cut wait times more?
Turn a vague goal into a question you can ask of every pair.
Turn our goal of 'better support' into one clear question we could ask when comparing any two fixes.
See how the order changes under a different criterion, such as cost.
Re-rank my sorted list by cost to implement. Which items move the most, and what does that tell me?
List your items, pick a mode, then copy the prompt into Claude, ChatGPT, Gemini or Copilot. In coach mode the AI keeps track of the list and shows you one pair at a time, while you make every call.
The “Open in” links put the prompt in the web address. If your challenge includes anything sensitive, copy and paste instead.
After the session, click one to copy it, then paste it into the same chat.