Turn the problem around, backwards or upside down, and see what the flipped version shows you.
When you've looked at a problem the same way for too long, every idea starts to look like the last one. Reversal shakes that loose. You write the problem as a sentence and turn it around: flip the goal, the direction, the order, the roles or the point of view. Then you ask what the upside-down version shows you.
Most reversed statements sound silly at first. “Instead of the app reminding users, users remind the app” isn't a plan. But it points somewhere new, and when you bring it back to reality it often becomes a practical idea you'd never have reached by pushing forward.
It's quick, it needs no training, and it works on almost anything: a goal, a process, a product or a rule.
Use it when your ideas all sound alike, when a team keeps pushing harder on the same approach, or when you want a quick, playful way to reframe a problem before an idea session.
State the problem, turn it around, then mine the upside-down version for an idea you can actually use.
Choose a part of the diagram or a tab to explore each step.
What are you trying to do, in one plain sentence?
What's the opposite, backwards or inside-out version of this?
What does the reversed version show you, and what practical idea does it suggest?
Reversal is quick and a little playful, but it works against some stubborn habits of thought.
Once we've framed a problem, we keep pushing in the same direction, only harder. A reversal doesn't ask you to try harder. It points you somewhere else entirely, so new options appear with very little effort.
Unlike a random stimulus, a reversal is always connected to your problem. That makes the leap feel strange but still relevant, so it's easier to find a path back to something useful.
Flipping a statement shows what you took for granted. When “users must link a bank account” becomes “users don't need to”, you suddenly notice that the assumption was never tested.
Each reversal takes seconds and costs nothing. Because reversed statements are obviously not proposals, people feel free to play with them, and nobody has to defend a silly idea.
Write the problem as one plain sentence with a clear verb: “How do we get users to respond to reminders?” Underline the parts you could flip: who acts, the action, the goal, the order.
Write three to six reversals. Flip who acts, flip the goal, start at the end, swap the roles, view it from the other side. Don't stop at the first one, and don't worry if they sound absurd.
Take one reversal at a time and ask: What's interesting here? What would have to be true for this to work? Where does something like it already happen?
Turn each insight into a practical idea that serves the original goal. Partial flips count: you don't have to reverse everything to use what the reversal showed you.
Shortlist the most promising ideas with a quick NUF Test, then pick one small, cheap test you could run within two weeks.
A sheet of paper with the problem written at the top, a pen and 15–30 minutes. With AI, the prompt below generates and explores reversals with you.
Subject: a seven-screen setup in a budgeting app, where most people quit at “link your bank”. Goal: more people reach their first budget. Constraint: no extra engineers this quarter.
Start at the end: show a sample budget first, and ask for details only when the user wants to make it their own.
Instead of users setting up the app, the app sets itself up: it guesses categories from three quick questions and lets users correct it.
Instead of “How do we get people through setup?”, ask “How could they skip it?” A no-setup mode that works from typed-in numbers.
From the user's side, setup isn't “onboarding”. It's handing a stranger the keys to their bank account. Lead with trust, not features.
“Users must link a bank to get value.” What if they didn't? Let the first month run on manual entries, and offer the bank link later.
Five reversals gave 14 ideas in 30 minutes. Three made the shortlist after a quick NUF Test.
Reversal needs two skills: producing lots of flips quickly, and spotting the useful idea inside a strange one. AI is excellent at the first and decent at the second. It doesn't know which flips fit your users, budget or brand. Let it produce and explore the reversals, and you decide which ones to bring back to reality.
Get many reversals at once, sorted by what they flip.
Give me 10 reversals of 'How do we get users to finish setup?' Label each by what it flips.
Find places where the flipped version already works.
Where does 'the app sets itself up for the user' already happen, in any industry? Say how sure you are.
Turn an absurd flip into a practical idea.
Take the flip 'users don't need to link a bank' and give me 3 practical ideas we could ship.
Reverse just one part instead of the whole problem.
Keep everything about our setup the same except the order. Give me 3 ways to reverse just that.
Write your problem in a sentence, pick a mode, then copy the prompt into Claude, ChatGPT, Gemini or Copilot. In coach mode the AI helps you flip the problem in several ways, explores each flip with you and helps you bring the best ones back to reality.
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.