1. Is the scope explicit?
State the reporting period, refresh date, included entities and exclusions. An annual target cannot be compared directly to actuals covering only nine months. Two neighbouring numbers may measure different things.
Add a note when the source is incomplete. An explained limitation helps the reader make a decision; apparent precision hides the problem.
2. Are data types consistent?
A date stored as text can break monthly grouping. An imported amount with an unexpected separator may remain text. Check a sample as well as blanks and extreme values.
Identifiers are not always numbers. A customer code starting with zero must preserve that zero when it is part of the key. Choose types based on business meaning.
3. Has a join multiplied rows?
Check reference-key uniqueness before joining sources. If a product appears twice in the reference table, a merge may multiply sales rows. Compare row counts and totals before and after the join.
Do not remove duplicates without understanding the cause. Two similar rows may represent two legitimate events. The key and data grain determine whether they are duplicates.
4. Are ratios calculated correctly?
An overall margin is not necessarily the average of percentages. Compare total margin with the corresponding total revenue. Define what should happen when the denominator is zero.
Example: a sale of 100 with a margin of 20 and a sale of 900 with a margin of 90 produce a margin of 110 on revenue of 1,000, or 11%. The simple average of 20% and 10% is 15%, which answers a different question.
5. Can someone reproduce the calculation?
Keep the sources, document transformations and separate inputs from formulas. Test refresh with a new file. Ask whether a colleague could reproduce the result without asking about undocumented operations.
A checklist does not replace business knowledge. It makes checks visible and helps build reports that can be explained, reproduced and improved.