A useful data question names something you need to understand and makes clear what an answer would help you do. “Analyse our business” offers neither. It leaves a person, or an AI tool, to guess what matters.
Begin with the situation. What has become difficult, expensive or uncertain? Then name the decision, the period and the evidence you expect to use. You may discover that the question needs records you do not yet keep. That is a useful finding in itself.
A small-business example
A shop owner asks, “Which products are doing badly?” This could mean low sales, poor margins, frequent returns or stock that takes up too much shelf space.
A sharper version is: “Which products have had no sales in the past twelve weeks, although they were available for sale throughout that period?” That question gives the stock team a specific list to investigate. It still does not prove that those products should be discontinued.
Try this
Write one question you would like your business records to answer. Underline every vague word: good, better, expensive, popular, recent. Replace each with an observable condition. Finish this sentence: “If I knew the answer, I could decide whether to…”
Use Define the decision to test that last sentence, Set the boundary to specify the period, and Data inventory to see whether the evidence exists.