Most analysis failures aren’t math errors… they’re hygiene lapses. Build a boring, repeatable checklist: (1) Reconcile counts: do totals match inputs after filters and joins? (2) Confirm definitions: do variables mean what the decision requires, and are time windows aligned? (3) Pre-write the one-sentence takeaway before you compute: then see if the numbers actually support that sentence. (4) Show uncertainty: intervals, ranges, scenario bounds. (5) Reproduce from raw: scripted steps, fixed seeds, and clear file names so the result rebuilds on a clean machine. Now add two more for production work: (6) Visual integrity: scales that don’t mislead, consistent axes, no cherry-picked zooms. (7) Peer check: someone else walks the pipeline and tries to break it. Process beats cleverness… because process catches cleverness when it goes astray.
