Continuous improvement
Seven basic quality tools: which one to use
Choose among the seven basic quality tools using the question and data you have. Follow a short route to worked examples and explanations.
The seven basic quality tools help teams collect, display and investigate process information. They serve different questions, so using all seven is not a measure of a good investigation.
ASQ's reference lists cause-and-effect diagrams, check sheets, control charts, histograms, Pareto charts and scatter diagrams. The seventh item varies across lists: stratification, a flowchart or a run chart may appear.
Choose the view the investigation needs, then collect the records that can support it.
Match the question to the tool
| Need | Tool | Next explanation |
|---|---|---|
| Record repeat observations consistently | Check sheet | Check sheet vs checklist |
| See which categories account for a stated measure | Pareto chart | Counts, delay or cost |
| Inspect the distribution of numerical values | Histogram | Histogram vs control chart |
| Follow behaviour in sequence with model-based limits | Control chart | Reading process signals |
| Examine paired numerical observations | Scatter diagram | Reading a relationship |
| Organise possible explanations | Cause-and-effect diagram | Fishbone example |
| Separate relevant groups or follow the work sequence | Stratification, flowchart or run chart | Mapping the process |
A tool can reveal a pattern without establishing its cause. Preserve that distinction when the findings move into the action plan.
Follow the order-inspection evidence
The check-sheet Note uses Minitab's published teaching example: 50 orders, six defective orders and seven defects. Those counts could be grouped by defect category because the collection record retains the categories.
A Pareto of those category counts would rank frequency. It would not rank cost unless the dataset also contained credible costs. A time chart would need observation order or dated subgroups, which the category totals alone do not supply.
The available record determines which view the analyst can build honestly.
Keep the process groups available
Stratification separates observations into useful groups, such as equipment, product or shift, when those fields are relevant and recorded. It can reveal a difference concealed in an overall average.
Choose those groups around the process question. Splitting a dataset repeatedly until one category looks unusual increases the chance of finding a misleading pattern. Treat an exploratory result as something to check with further evidence.
Move from the plot to a testable explanation
A fishbone diagram collects possible causes. A scatter plot can examine a relationship. Neither confirms a corrective action solely by displaying it.
Use the root-cause guide to connect the observations with checks that distinguish among explanations. If the next decision concerns a change, the PDCA example shows how to write a prediction and review a bounded trial.
At the next improvement meeting, state the question first. Then select one view that the current records can support, and identify the missing record needed for the following decision.