You’re working hard on something, but it never quite finishes. It keeps almost being done - one more tweak, one more pass - and the end never actually arrives.
That’s the trap. Work rarely fails because people stop trying. It fails because nobody defined what “done” actually means. Without a clear finish line, the work keeps growing and never truly finishes.
This tool makes you draw the finish line before you start - and makes it testable. You write down what the work must produce to be done, as specific criteria you can answer yes or no. Then you can compare the finished work with the line and know, without arguing, whether you’re there.
Imagine this: a team sets out to “improve the onboarding flow”. Vague - it could never end. So they define done as testable criteria: new users reach the first value in under three minutes (yes/no), the sign-up form has four fields or fewer (yes/no), the drop-off between steps two and three falls below fifteen per cent (yes/no). Now the work has an end, you can prove you’ve reached - not a feeling that it’s “better”.
First, list what must exist for the work to be finished. Then make every item testable by asking, “how would someone prove this?” If the answer isn’t yes or no, rewrite it until it is. Finally, lock the definition before work begins. Done means every criterion answers yes.
This is the owner’s call, at kickoff. The person accountable for finishing the work defines done - usually solo, before the work begins.
Solo ✓ About 15–20 minutes at the start of a piece of work - before you begin, not partway through.
The time required is indicative and will depend on how complex the work is.
Read the short definition guide first - about 6–8 minutes. It is a downloadable PDF on this page.
Bring one piece of work you’re about to start - a deliverable, a project, a task with a real end you’re accountable for.
Use it whenever:
Not the right tool if: the work is finished and you just need to prove it’s visible to whoever needs it - that’s Done ≡ Show It, the quick check at the end. This tool defines done at the start; that one confirms it’s shown at the finish. (For ordering the path to done, use Execution Sequencing System.)
You’ve done it well when: you can hand the finished work and the definition to someone else, and they can tell you yes or no on every criterion without needing you to explain.
An undefined “done” is a finish line that keeps moving because everyone fills in the gap differently and nothing can be held against it. Writing done as concrete criteria fixes the line in place; forcing each criterion to be answerable yes or no is what stops it drifting again - a criterion you can’t test is one you can quietly renegotiate. The test is the mechanism: every real completion criterion must be able to fail before it can pass.
One piece of work, at the start. Something you’re about to begin and are accountable for finishing. Define done before you start, not partway through.
The proof question - the heart of the tool. For every criterion: “how would someone prove this is true?” If there’s no yes/no answer, it isn’t a criterion yet - everything else exists to serve this question.
The definition sheet (supplied). Where you write each done-criterion and its yes/no test. It forces testability - a criterion with no test doesn’t count.
Honesty about vagueness. Don’t accept vague words like “works well” or “high quality”. If you can’t test it, it isn’t part of “done”.
Output / Takeaways
You leave with:
Important rules
Planning to facilitate this activity? Access the resources below to get started.