Make your process’ performance visible. You may know how it performs ‘on average’, but how good (and bad) does it get? Plot on a graph the frequency of different results to see the natural variation inherent in any process. Is the spread of results larger than customers really want? Is the process skewed?
2. Cause & Effect (‘Ishikawa’) Diagrams
Draw a horizontal arrow from left to right, pointing at your problem’s description, with four arrows labelled ‘Man’, ‘Machine’, ‘Materials’ and ‘Methods’ pointing diagonally onto the first to make a herringbone. Brainstorm causes of the problem, and attach these to the relevant ‘M’ arrow. Select the most likely ones for further investigation.
3. Scatter Diagrams
If a process’ input and outputs are variables (can be change on a sliding scale, rather than discrete steps), plot one against the other to see if they’re related. Can you change one input and get a predictable (-ish) change in the output?
4. Check Sheets
If you have a range of products / services, and a range of problems, draw a table with the products on the top, and problems down the side. Tally up each problem’s frequency for each product to identify the dominant factor.
5. Pareto Diagrams
Count the frequency of each problem type, and plot them on a bar chart, in order from worst to best. Tackle the worst ones first - typically 20% of the problems will be causing 80% of the pain.
Having difficulty understanding what’s going on? Draw a picture of the process’ flow to see how each element is related, and how external factors can influence things.
7. Run Charts
Plot performance over time (hourly, daily, or weekly results), and look for patterns, cycles, and trends. Take corrective action before problems occur.
Next section of this article: What To Do Next