paretolisted

# Pareto A small subset of causes produces most of the outcome. Find it, spend about a fifth of the effort against it, and show the cut so the user can argue with it. Estimate roughly throughout. 80/20 is shorthand for "expect lumpiness," not a ratio to hit, and the two percentages needn't pair — a fifth of the effort might buy 95% of the outcome or 40%. A rough ranking that costs minutes beats a precise one that eats the budget it was meant to allocate. ## When this is the wrong tool Some work is only worth doing completely: security fixes, compliance changes, migrations, a specific reported bug. 80% of a data migration is a corrupted database. Say the request isn't a Pareto candidate, then do it in full or ask. Declining the framing means declining its shape too. Give a checklist to finish rather than a ranked list the reader can stop partway down — on completeness-critical work the skipped item is the one that matters. Skip the ceremony entirely on anything small enough that ranking it costs more than it saves. ## Stage 1: rank the causes Rank causes, not tasks. Work items sorted by payoff are just a to-do list; asking "what is generating most of this?" is what finds the one confused module behind eight scattered bug reports. Weight by cost or severity rather than raw counts, so a rare catastrophe isn't buried under a constant nuisance. Rank on effect size alone — what a fix would cost plays no part yet. Letting "that's expensive" suppress a cause here buries the