Turn a fuzzy worry into one clear sentence that everyone agrees is worth solving.
A problem statement is a single, plain sentence that says who has a problem, what isn't working for them and why it matters. It sounds like paperwork. In practice it may be the most important sentence in a project, because every idea you generate later will aim at whatever it says.
You get there by drafting, not by declaring. Write several versions, test each one against a short set of checks, and rewrite until the sentence is specific enough to act on and open enough to allow many solutions. When everyone involved can read it and nod, you're ready to generate ideas.
The most common trap is a solution in disguise. “We need an app” isn't a problem; it's an answer. A good statement describes the gap and leaves the answer for later.
Use it when you're starting a project, when a team keeps talking past each other, when ideas feel scattered, or when someone has already fallen in love with a solution before anyone has agreed on the problem.
Test every draft against the seven checks. Each “no” or “maybe” tells you what to rewrite next.
Who exactly has this problem, and who feels it most?
What happens today, and what should happen instead?
What happens if nobody solves this?
Does the sentence describe a problem, or quietly prescribe an answer?
Is it big enough to matter and small enough to act on?
Could anyone involved say it from memory?
Do the people who live with the problem agree this is it?
Defining the problem can feel like a delay. It's actually where much of the leverage in a project lies, for a few simple reasons.
Ideas follow the question. If the statement points at the wrong problem, even brilliant ideas miss. A few minutes spent aiming can save weeks spent building the wrong thing.
People naturally leap to a solution, usually the first familiar one. Writing the problem with no solution inside it keeps the door open to options nobody has thought of yet.
In most groups, each person privately holds a slightly different version of the problem. Writing it down brings those differences into the open, where they can be discussed instead of quietly pulling the work in different directions.
Later, when you have twenty ideas, the statement is how you judge them. Does this idea close the gap for these people? If not, it's off target, however clever it is.
Gather the raw material: what people say about the situation in their own words, what you've seen happen, and any numbers you have. Note where each point came from. Don't tidy or interpret anything yet.
Each person writes their own one-sentence version of the problem without discussing it. Separate drafts are valuable because they reveal where people see the problem differently.
Where do the drafts agree? Where do they differ, and why? Disagreement found now is cheap. Disagreement discovered after you've built something is expensive.
Take the strongest draft and test it: who, gap, why it matters, no hidden solution, right size, one sentence, agreed. Rewrite whichever part fails, then test again.
If it feels hopeless, narrow it to one group, place or moment. If only one answer fits, widen it by asking what that answer was for. Chunking helps you move up or down a level.
Read the final sentence aloud and check that everyone can stand behind it. Then put it where everyone will see it during the idea session, so it keeps the work on target.
Sticky notes, a whiteboard, the seven checks and 20–30 minutes with the people who live with the problem. With AI, start with the prompt below.
Subject: a community health clinic's diabetes care. Goal: patients come back for their three-month check. First draft: “We need a better reminder system.”
“A better reminder system” is an answer. The team agreed the real problem was patients not returning, whatever the cause.
Front-desk records showed most missed visits came from patients working hourly shifts, not from patients in general.
Follow-ups are booked at checkout, three months ahead, but many shift schedules are posted only two weeks in advance.
A missed check means rising blood sugar can go unnoticed until it becomes an emergency-room visit.
Not “improve diabetes care” and not “rewrite the reminder text,” but getting these patients back for their check.
A nurse, a scheduler and two patients read draft 3. One patient added: “And rebooking means 20 minutes on hold.”
“Patients with diabetes who work hourly shifts often miss their three-month check, because the visit is fixed before their schedule is known and rebooking is a chore.” A reminder text would never have fixed that.
AI is good at the part people find awkward: asking “what exactly do you mean?” again and again without getting tired or polite. It can also produce many rival drafts in seconds. What it can't do is know which version is true in your world. That part is yours.
Let AI question you, one point at a time, until your draft is precise.
Ask me one question at a time about my problem until you could explain it to a stranger.
See many versions of the statement side by side, then pick and mix.
Write 8 one-sentence problem statements from my notes, each framing the problem a different way.
Catch answers hiding inside the problem before they narrow your thinking.
Does my statement hide a solution? If so, name it and rewrite the sentence without it.
Read the statement through other people's eyes before you commit to it.
Read my statement as a patient, a nurse and a manager. What would each say is missing?
Describe the situation in your own words; messy is fine. Copy the prompt into Claude, ChatGPT, Gemini or Copilot. In coach mode the AI asks one question at a time, drafts with you and tests each draft before you settle on one.
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.