Rank problems on one declared basis

Rank problems on one declared basis

A Pareto display ranks categories using one declared basis. Sort the bars from largest to smallest, then calculate cumulative share against the same total. Decide whether the current question concerns frequency, money, time or another burden before drawing the chart. Mixing costs for one category with counts for another produces an invalid ranking. The cumulative line helps show concentration, but it does not promise an eighty–twenty pattern or justify ignoring rare severe events. Use the ranking to select an investigation, then return to the underlying observations and risk context before deciding what to change.

Follow the method

1. Counts and cumulative contribution
2. Nonconformities (count)
3. Category
4. Cumulative %

Choose a priority without confusing counts, time and causes

Fictional case: Fern Finishing records 50 nonconformities in one inspection window: 24 scratches, 16 burrs, eight missing labels and two other faults. These exact values remain in the core chart. Team leader Rosa wants to choose a starting investigation. A colleague proposes adding estimated repair dollars to the occurrence counts and calling the combined bars an impact score.

Follow the method

1. Scratch
2. Burr
3. Missing label
4. Other

The largest count is not automatically the largest loss

Construct and interpret a ranking on one declared basis, then distinguish a frequency priority from a severity or causal decision.

Fictional case: Fern Finishing records 50 nonconformities in one inspection window: 24 scratches, 16 burrs, eight missing labels and two other faults. These exact values remain in the core chart. Team leader Rosa wants to choose a starting investigation. A colleague proposes adding estimated repair dollars to the occurrence counts and calling the combined bars an impact score.

Role: Team leader and quality analyst

Normal condition

A ranking uses one meaningful unit, a defined observation window and mutually understood categories. The underlying event list remains available for investigation.

The gap

The draft mixes frequency and money and labels the first two categories “80% of our causes”. The record contains occurrences, not verified causes.

• The categories describe faults and may involve different mechanisms.
• A rare severe event may deserve a separate immediate response regardless of rank.

1. Freeze the question and basis

   Rosa writes “Which categories contribute most recorded nonconformity occurrences in this window?” She confirms that all four categories use the same recording rules and the same window.

   Why: The ranking becomes reproducible only when its scope and unit are fixed. A category total cannot be compared with another that covers a different period or population.

   Evidence: The title says occurrence count and names the window.

2. Order and reconcile the bars

   She sorts counts in descending order: 24, 16, 8, 2. The sum is 50. She checks that every logged event belongs to one reported category or is explicitly handled under the collection rule.

   Why: Sorting changes display order, not the underlying evidence. A missing category would distort every later cumulative percentage.

   Evidence: The category total reconciles to fifty recorded nonconformities.

3. Calculate the cumulative line

   Cumulative counts are 24, 40, 48 and 50. Dividing each by 50 gives 48%, 80%, 96% and 100%. The line uses its own percentage axis.

   Why: The first two bars happen to reach 80% in this dataset. That is an observation, not a rule that the plotting process must force.

   Evidence: The native chart retains count bars and a labelled cumulative-percentage axis.

4. Compare a different decision separately

   For recovery effort, she creates a separate minutes calculation: 48, 80, 80 and 60 minutes. Total estimated recovery is 268 minutes; burrs and missing labels tie for the largest time burden.

   Why: Multiplying count by an explicitly illustrative per-event duration changes the question. It shows why a frequency ranking need not be the correct priority for every objective.

   Evidence: A second record states minutes, assumptions and its different order; no mixed-unit score is created.

5. Choose investigation without inventing cause

   Rosa selects scratches for a frequency-focused observation study and flags any severe exceptions through the existing response process. She keeps the rival time-based priority visible for the resource decision.

   Why: A category name does not explain why the fault occurs. The next work must locate conditions, test explanations and verify the consequence of an action.

   Evidence: The decision record says priority for investigation, with cause still unverified.

A rare event is severe

One of the two “other” events involves a potentially significant product risk.

Use the applicable immediate escalation and containment process while retaining the descriptive frequency ranking. Do not wait for its bar to become large.

Frequency is one decision dimension. A Pareto chart does not authorize ignoring a severe low-frequency condition.

The action log records the risk response independently of the improvement-priority list.

Run a practice session

Materials

• Count cards
• Blank cumulative table
• Separate minutes worksheet

1. Name the question · 4 minutes

   Which unit is being ranked?

2. Build the two records · 8 minutes

   Why must minutes not be added to counts?

3. Complete the changed week · 10 minutes

   Where does the cumulative line finish and why?

4. Debrief the severe exception · 5 minutes

   Which decision is outside this frequency chart?

Debrief

• Ask a learner to defend a time-focused priority without altering the count chart.
• Require a plain-language statement of the uncertainty behind per-event durations.

Sort paper category cards, calculate cumulative totals, then open the answer.

Transfer into the work

Owner: Quality analyst and improvement sponsor

Record: Ranked source-linked chart plus bounded investigation decision

Review: At each comparable reporting window

Evidence: Same basis, complete categories and actual investigation evidence

Revisit collection definitions or choose a different basis when the decision question changes.