ORDREL / LEARN · ENGLISH
Pareto Analysis of Production Losses: A Worked Example
A Pareto analysis sorts comparable categories from largest to smallest and shows their cumulative share of the total. It helps choose where to investigate; it does not prove causes or guarantee an 80/20 pattern.
Choose one measurement basis
Decide whether the question concerns event frequency, lost minutes, rejected units or cost. Do not mix those units in one series. Use one process boundary and period, with mutually exclusive categories and a documented rule for concurrent events.
Counts and rates answer different questions. A high-volume product can have more defects but a lower defect fraction. If exposure differs, inspect denominators before choosing an improvement priority. Severe low-frequency risks also need a separate assessment rather than automatic dismissal by rank.
Rank a fictional stop log
Fictional manufacturing teaching example. These figures are not ORDREL customer results.
| Category | Minutes | Share | Cumulative share |
|---|---|---|---|
| Changeover adjustment | 45 | 37.5% | 37.5% |
| Material interruption | 30 | 25% | 62.5% |
| Equipment fault | 20 | 16.6667% | 79.1667% |
| Inspection hold | 15 | 12.5% | 91.6667% |
| Other documented stops | 10 | 8.3333% | 100% |
Total = 45 + 30 + 20 + 15 + 10 = 120 minutes. Each share is category minutes ÷ 120 × 100. Cumulative share is the running sum in descending order. The first three categories account for 95 ÷ 120 = 79.1667%, so an 80% threshold is first reached by including the fourth category.
Keep full precision for the cumulative calculation. Adding rounded labels can give 99.99% or 100.01%; this is a display artifact, not an extra lost minute. The fifth category still needs a useful definition, not a bucket for missing records.
Turn the ranking into a question
The largest category suggests where a detailed observation may be valuable. “Changeover adjustment” still covers several possible mechanisms. Break it down using consistent subcategories, inspect representative events and connect hypotheses to evidence.
A smaller loss may be cheaper or safer to recover, or may constrain the customer requirement more directly. Combine the ranking with consequence, feasibility and the process constraint. Do not claim the chart identifies the root cause of the loss.
Reproduce the chart in ORDREL
Open Pareto analysis and use minutes as the common unit. Enter the five labels and values 45, 30, 20, 15 and 10. Review descending order, a total of 120 and the cumulative percentages above. Check how any threshold highlight is defined before describing a “vital few.”
The OEE tool gives equipment loss context, while Fishbone supports the subsequent investigation. Keep any report or export free of customer identifiers when sharing it.
Common mistakes and follow-up
- Do not enter a negative loss to offset another category; correct the source accounting.
- An event counted in two categories distorts both the total and the ranking.
- Changing category definitions between periods can look like improvement when nothing changed.
- A better-looking Pareto may result from more “other” entries or less complete recording.
- After a trial, compare the absolute loss and exposure as well as the share: a category can gain percentage share while its minutes fall.
Use the same measurement basis after the change. Record the collection coverage and any excluded time. A useful follow-up question is “Did the loss per comparable production opportunity decline?” rather than “Did the colored bar become smaller?”
Sources and method notes
Sources support the method definitions. Examples, interpretation and tool instructions are ORDREL teaching material, not an endorsement by the referenced organizations. See methods and corrections.