A Data-Quality Checklist for Reports

Bad numbers in a polished report are worse than no report. Here's a checklist you can partly automate.

Before you ship

• Do the totals actually sum?
• Is any single figure wildly out of line with its peers?
• Are percentages within a sane range?
• Does every "see Figure/Table/Appendix X" resolve?
• Is terminology consistent throughout?
• Do dates form a coherent timeline?

Automate the boring parts

Most of these are mechanical. Outlier checks, reference resolution, and date ordering are exactly what a signal scanner does in seconds — freeing you to focus on judgment calls.