Should your Pareto chart rank defects by count or by cost?

Rank defects by count when the decision is about how often customers or staff encounter them. Rank by a consistently defined cost when the decision concerns recorded loss. When these questions compete, show both rankings from the same incident population and explain which decision each supports.

A Pareto chart does not automatically mean “80% of the problem comes from 20% of causes.” ASQ describes ranking categories using measures such as frequency, cost or time and calculating their shares of the chosen total. The useful result is the observed concentration in your data, not a target pattern imposed on it. [S1]

This fictional example covers 60 distinct dispatch-failure incidents recorded during one month at a UK distributor. Each incident belongs to one primary category under a written classification rule. The amounts are invented direct resource-cost estimates in pounds, on the same basis for every category. No tax, currency conversion, lost-sales estimate or supplier recovery is included.

Real records need an incident identifier and a cost reference. If one event has several defects, choose whether the chart counts events, defect occurrences or affected orders. Do not call all three “incidents.” For this example, mutually exclusive incident categories allow the count total to reconcile to 60.

The two most frequent categories account for 52 ÷ 60 = 86.7% of incidents but £500 ÷ £2,180 = 22.9% of estimated cost. The two largest cost categories account for £1,680 ÷ £2,180 = 77.1% of estimated cost, from only 8 ÷ 60 = 13.3% of incidents. Percentages are rounded to one decimal place.

Neither view is wrong. A team reducing the volume of routine corrections might start with labels. A team investigating the largest recorded loss would look first at transit damage. “Largest recorded loss” still does not mean “largest saving available”: the cause, recurrence and cost of changing the process remain unknown.

Suppose the £600 per damage incident is only an early estimate. At £200 each, transit damage becomes £400 and the total becomes £1,380. Wrong items then rank first at £480. The original cost ranking depends on an assumption worth checking before committing the improvement team.

The break-even point is £480 ÷ 2 = £240 per damage incident: above that figure, damage outranks wrong items on recorded cost; below it, wrong items rank first. This comparison is a sensitivity test, not a forecast or an instruction to value every incident equally.

Keep uncosted incidents visible. Missing cost is unknown, not £0. Also review consequences that this cost measure does not capture. A low recorded amount cannot justify dismissing a serious customer, safety or compliance issue; that issue needs its own competent assessment outside this ordinary dispatch example.

When you redraw the chart, retain the period, population, category rule and cost basis. A changing product mix, price or reporting practice can move the bars without a process improvement. Start with one reconciled log, publish the two rankings together and state the next evidence-gathering decision. The chart has then done useful work before anyone claims a saving.

QSE Advisory is published by QSE Advisory (office@qseadvisory.com). Research, drafting and the recorded editorial checking pass were AI assisted. Every worked scenario is fictional.

Prepared with AI-assisted research and drafting. Reviewer and Assessor: Laraib Naveed Ahmad.

This article is general information. Verify current requirements and seek qualified advice for your specific circumstances.

Spotted an issue? Contact our editorial desk.