True Causes

Incident investigation: why the same failure keeps coming back

A post-mortem is usually one analyst reasoning on the group's behalf, and it usually names the most visible failure: the outage, the returned part, the picking error. Those are consequences. The causes sit further back, spread across the people who saw different parts of it.

True Causes turns the review into the group's own analysis. Everyone proposes causes anonymously, so nobody writes around a manager. The group then answers one question at a time, does this cause significantly influence that one, and a link is kept only when three quarters agree. A split is never quietly dropped: it goes to another group, and if that splits too, it is published as contested.

Several teams or sites can run the same question separately and merge their maps, which shows which causes they found independently.

Why are deliveries late?
A finished map, from a walkthrough run with stand-ins.
We measure errors perorder, so a bigger order…New staff copy bad habitsfrom old handsPeak-hour load exceedsplanned capacityThe best pickers are movedto whatever is late, so…The handheld takes fourseconds to confirm, so…Supervisors are measured ondispatch time and nothing…The forecast is made bypeople who never see a peakTargets reward speed overaccuracy⟳Fixes are patched locallyand never shared⟳A short pick is recorded ascomplete so the line keeps…⟳Small errors are hidden toavoid blame⟳Two products with almostthe same code sit on the…⟳Overtime is the normal waywe finish a week, not the…⟳Picks are wrong oftenenough that customers now…⟳Nobody checks a palletbefore it is wrappedWe spend more time fixingerrors than we would…Open it, tap any claim →
Start a dialogue, freeTry the five-minute walkthroughWhy it works