Straight Fact · desk research and BespokeData methodology · 18 Nov 2026
A Pareto chart does not prove the cause
2 min read · Free article
A ranked chart can tell you which recorded categories are most common without telling you why the problem occurs. That is still useful: it directs attention and makes the counting rule visible. But frequency is not causality, and a label may describe a symptom or the place a problem was discovered. Treat the leading category as a hypothesis to investigate rather than a verified root cause or an automatic savings opportunity.
ASQ describes Pareto charts as tools for organising categories by frequency or magnitude [A2]. Our practical recommendation is to preserve the category definition, population and period beside the chart. Decide whether records can have multiple categories. If they can, counts may exceed the number of incidents and the denominator must be explicit. A familiar cumulative line does not correct inconsistent coding or missing records.
Consider 100 hypothetical late orders with one provisional primary category each: awaiting stock 50, approval queue 30, address clarification 15 and other 5. The first two categories account for 80 of 100 recorded cases, or 80%. This is arithmetic about the labels. It does not mean that changing stock and approval processes will prevent 80 delays. Some orders may already have been late before entering those stages, and one underlying issue may appear under several labels.
Inspect event sequences and a documented sample of cases. Ask what alternative explanations fit: order mix, supplier disruption, policy, system errors or inconsistent categorisation. Keep contrary examples in the evidence register. If stock is unavailable, find out whether the record was wrong, demand was unusual or replenishment was late. Each mechanism suggests a different intervention, and the category total alone cannot choose between them.
A practical action is to create an opportunity row with hypothesis, supporting evidence, contradictory evidence, proposed intervention, owner and test measure. For approval waiting, a limited routing pilot might be evaluated against both delay and rework. Keep baseline definitions unchanged and record concurrent staffing or system changes. A before-and-after reduction may be encouraging but remains confounded unless the comparison design supports a stronger conclusion.
These invented counts are not a customer finding or proof of a universal 80/20 rule. Small samples, inconsistent labels and absent timestamps limit what can be concluded. Read the full operational assessment guide for transparent baseline calculations, a benefit scenario distinguished from ROI and a worksheet that turns a ranked observation into a testable, owned next action.
Dated method & source register
Read and reviewed 18 Nov 2026. Desk research and original BespokeData recommendations and hypothetical examples; no primary study, independent expert review or Atora generation.
- [A2] ASQ · Pareto chart
Frequency and prioritisation reference; counts alone cannot prove a cause.