QualityDefects reach the customer and we find out late.
Control charts that say whether the process changed, and a control plan that holds the fix.
How it works
Quality, on the real screen
The question is never "is this number bad", it is "did the process actually change". Control charts answer that with limits computed from the data rather than typed in, and capability studies say whether the process can meet the specification at all. 28 of the 32 analyses compute their figures in a deterministic engine, so the numbers do not move if the model is having a bad day.
Lives at /dataforge once you are signed in.
In the box
What is in the box, and what you leave with
What is in the box
- Control charts: I-MR, X-bar R, EWMA, CUSUM, attribute
- Process capability, including non-normal
- Sigma metrics and rolled throughput yield
- FMEA and control plan templates
- Free Cpk calculator
What you leave with
- A chart with limits you can defend
- A capability figure and what it means
- An FMEA and a control plan
- A reaction plan for when it drifts
Further in
The fix, held by a control plan with a reaction plan
A saved DocuForge control plan inside an enterprise project, with each process step, its control method and its reaction plan.
The agents
The agents that do this work
Product screen, demo data
The method
How the work is done
DMAIC
Define, measure, analyze, improve, control, with a gate between each.
- Project workspace
- Five tollgates
- Charter template
- Capability study
- Control plan
Root cause analysis
Get past the symptom to the thing that is actually causing it.
- Fishbone
- Five whys
- Pareto
- Multi-vari study
- Regression
Other problems
The other five
- Value & ROIWe cannot show what the improvement work returned.
- Continuous ImprovementIdeas come in and nothing happens to them.
- Process OptimizationThe process is slow and nobody agrees on why.
- LeanThere is waste everywhere and no system for removing it.
- Operational ExcellenceImprovement happens in pockets and never becomes how we work.
Quality
See it on your own data
Thirty minutes. Bring a spreadsheet, and leave with the analysis and the documents it produced.