The Psychology of Quality and More

A Toolbook for Quality Improvement and Problem Solving (contents)

Pareto Chart: How to do it

The Quality Toolbook > Pareto Chart > How to do it

When to use it | How to understand it | Example | How to do it | Practical variations

<-- Previous | Next -->

How to do it

1. Identify the group of items that are to be charted and consequently sorted. This should be a complete and single group, for example, 'Sales of product line 32'.

Break down the group into actual items to be measured. It can be useful to do this in several different ways plotting several charts, in order to find the most spiky chart and thus the most significant grouping. For example, product sales may be broken down by industry, geographic area and turnover, with a Pareto Chart being drawn for each.

If possible, aim not to have too many items in each group (about seven is a good maximum), or else bars will be too low to be of use. Thus, when breaking down by industry, you might use major groupings such as 'chemical' and 'medical' rather than finer groups such as 'acids' and 'polymers'. If a group is found to be significant, then it can be broken down further.
 

2. Identify the unit of measurement. All items on the chart must be measurable in this unit. Common units include frequency, cost and size.

Use a measure which best reflects the key objective. For example, if the aim is to reduce cost, then measure the total failure cost of each defect type rather than the number of defects. The figure below shows how using a different measurement unit can significantly change the Pareto ordering.
 

Fig. 1. The effect of weighting

3. Determine the sample size or period during which measurements will be made. This should be enough to ensure that the final chart will have bars of varying height, sufficient to compare them. A test of this is that increasing the number of measurements should not change the order or relative height of the bars (unless the process is unstable).

If the measurement unit is a simple frequency, aim for a total of 50 or more measurements.

In any repeat measurement, keep constant any variables which might distort the figures. This includes the sample size or measurement period, along with anything else which might affect the result, such as people, materials, etc. If this cannot be done, then recognize it and take it into account when interpreting the final chart.
 

4. Measure the items, for example using a Check Sheet, as in the diagram below. Ensure that people doing the measuring are able to do it correctly, training them as necessary.
    
5. Sort them into size order, putting the largest measure first. If there are more than about six items, and the smaller measures are significantly smaller than others, then these can be added together to form an 'others' group (placed at the end of the list, whatever its size).
    
6. Plot these in vertical bars, with the largest bar on the left, as below.

Fig. 2. Building the Pareto Chart

7. Interpret the results and act accordingly. If there is no clear highest bar, then it may be worth producing another Pareto Chart using a different set of measured items. If there are one or two bars which are clearly higher, then these may be carried forwards for further analysis and action (possibly with another Pareto Chart).

Common sense should always be used during interpretations, as the highest bars do not always represent the best action items. For example, in producing a report, 'analysis' may be a high-value activity, so the aim may be to increase this at the expense of other activities. There may also be other selection criteria to take into account, such as the time and cost of corrective actions.

<-- Previous | Next -->