1. What decision or task gets better?
Name the task in one sentence and the person who does it today. If you cannot, the idea is not ready.
2. How will we know it worked?
Pick one or two measures before you start, such as time saved per item or error rate against a reviewed sample.
3. Where does the data come from?
Check that the data exists, is accessible and may be used for this purpose.
4. Who reviews the output?
Decide where a person checks results, and how many are checked.
5. What happens when it is wrong?
List the worst realistic error and who is affected. Design the review around it.
6. Who owns it afterward?
A system without an owner decays. Name one.
7. What is the exit?
Decide in advance what result would make you stop.




