Skip to content
Quality Tools | Part 1 of 8

Which Quality Tool When: Start From the Question

The classic quality tools each answer one question. An overview of the tools, the order that works for most problems and a worked example from a coating line.

Every plant has an explanation for its defects. “It is the late shift.” “It is the material.” “It has always been like that on Mondays.” Few plants have a count. The classic quality tools exist to close that gap. They were collected for teams at the process, not for specialists, and all of them work with a pencil, a sheet of paper or a simple spreadsheet.

A tool is a question

The most common mistake is to start from the tool. A team learns the cause and effect diagram and from then on draws one for every problem. It works better the other way round: start from the question in front of you and pick the tool that answers it.

The question The tool
What happens, and how often? Check sheet
Does it differ by shift, position, product or day? Stratification
Which few items carry most of the problem? Pareto chart
How does the work really run? Flowchart
How is a measured value spread? Histogram
Does it change over time, and is that more than noise? Run chart, control chart
What could cause it? Cause and effect diagram
Do two measured variables move together? Scatter diagram

Four management tools join them when the material is words instead of numbers: the affinity diagram brings order into many statements, the tree diagram breaks a goal down into actions, the matrix diagram relates two lists, and the prioritization matrix ranks options.

The example in this series

All eight parts follow one case. A plant with 160 employees makes sheet metal enclosures and cabinet doors. Powder coating line 2 coats 1,200 parts per day on a conveyor with three hanging rows. Every part is inspected after curing.

In a baseline of 20 days, 24,000 parts were inspected and 1,920 had a defect. That is a defect rate of 8.0 percent and a first pass yield of 92.0 percent. Rework, which means sanding or stripping and coating again, cost EUR 28,800 in four weeks or EUR 345,600 per year.

Twelve weeks later the same check sheet showed 1,104 parts with a defect in 20 days, a rate of 4.6 percent. No new equipment was bought. Two findings made the difference, and both came out of simple sorting.

An order that works for most problems

The tools build on each other. This sequence fits most recurring problems on the shop floor and in the office:

  1. Draw and count. A flowchart of the real process, written definitions of the defects and a check sheet at the place where defects are found.
  2. Find the vital few. A Pareto chart by count and by cost. On line 2, three of seven defect types carried 77.5 percent of the defects.
  3. Find where it sits. Stratification, histogram and run chart. Thin coating sat on the bottom hanging row, inclusions on Mondays.
  4. Find and verify causes. A cause and effect diagram collects what the team suspects. A scatter diagram and a trial test the lever.
  5. Decide. Affinity diagram, tree diagram and prioritization matrix turn ideas into ranked actions with owners.
  6. Prove it. A control chart and a second Pareto chart on the same check sheet show the effect.

Not every problem needs every step. When the answer is clear after step three, stop analyzing and act.

What the tools are not

They are not a statistics course. Process capability, hypothesis tests and the project cycle belong to Six Sigma, and structured root cause methods to problem solving. The basic tools come before both. They make sure that the problem is defined, counted and located before anyone starts a project.

Where to start

Pick one process with a recurring problem. Define seven to ten defect types, put a check sheet at the place where the defects are found and count for four weeks. The next part shows how to build that check sheet and how to sort what it collects.

Newsletter

One Lean idea.
Every two weeks.