Skip to content

    Expert services · 8D with the statistics to find where the cause is.

    Advanced Problem Solving

    The 8D is closed and the defect is back?

    On-site and remote. The investigation plan is agreed before anyone collects data. Scoped per project.

    What we deliver

    • 8D facilitation and review
    • Design of experiments (DOE)
    • Hypothesis tests, ANOVA and regression
    • Claim dispute analysis between tiers
    • Measurement system analysis (MSA)
    • Data collection plans
    • MANOVA where the question and the data justify it

    How it runs

    Test the cause before you choose the action.

    Start with the question

    What exactly is wrong, where it shows up and where it does not. Which explanations are on the table, and which decision the answer has to support. Containment stays in place while we look.

    qualitehInvestigation planisis notwhatwherewhenhow muchexplanationsmaterial lotmachine settingmethodmeasurementdecision to support

    Check the gauge before the numbers

    When appraisers or gauges disagree, every later argument is about noise. So the measurement system is checked first, with an MSA where it is needed.

    qualitehMeasurement checkappraiser Aappraiser Bappraiser Cpartsmeasurement spreadpart-to-part spread

    Plan the trials with the people who run the line

    Factors, levels and samples by production date, agreed with production before anything changes. A designed experiment tests several factors together instead of one at a time.

    qualitehExperiment planlowhightemperaturepressurehighhighmaterial lottemperaturepressurematerial lothighlow

    Let the data separate the explanations

    Comparisons, ANOVA and regression test which factor carries the variation and which does not, and how sure we can be. In a claim dispute, this is where the data tests whether the cause sits where the claim says, or somewhere else.

    qualitehAnalysistemperaturepressurematerial lotinteractionlot Alot Btemperatureshare of variationmaterial lotmachineshiftresidual

    Fix it, then prove the fix

    The corrective action goes in, and its effect is checked on production data, not on a trial batch. The 8D closes on evidence, and the analysis, the open questions and the owners stay with your team.

    qualitehValidationlower limitupper limitbeforeafterproduction datacorrective action

    A test shows which explanation the data supports, and how sure we can be. Who is responsible is still for the parties to decide.

    Typical use cases

    • The 8D was closed and the defect came back.
    • Your customer or your sub-supplier disputes a claim, and both sides have an opinion but no data.
    • Several factors could explain the variation, and trying them one at a time costs production time.
    • People argue about the numbers before anyone has checked the gauges.
    • The corrective action is in place and nobody can show that it worked.

    Exit criteria

    We agree the criteria and supporting evidence before work starts. Your team reviews progress at the agreed milestones; customer approval remains with the customer.

    • The root cause is stated with the data behind it and reviewed with the people who own the process.
    • The corrective action is in place and shown to work on production data, not only on a trial batch.
    • The control plan and PFMEA are updated where the cause was found.
    • For a claim dispute: a written analysis both parties can read, with the method and the data described.

    Frequently asked questions

    Do you replace our 8D process?

    No. We facilitate and review it. Your team owns it, and the plant's quality manager stays the approver.

    When is a designed experiment worth it?

    When several factors could explain the variation and testing them one at a time would cost too much production time. We start with a screening design, then look closer at the factors that matter.

    Can you support a claim dispute with our customer or our sub-supplier?

    Yes. Where we can, we agree the question, the measurement and the data with both parties first. The analysis is written so both sides can read it, and the decision stays with the parties.

    Can statistics prove the supplier is responsible?

    Statistics test technical explanations within the limits of the study. Responsibility and the commercial settlement stay decisions for the parties.

    What happens if the data cannot answer the question?

    We say so. We agree further measurements or trials, and an inconclusive result is recorded as inconclusive.

    Which methods do you use?

    Standard statistical methods: comparisons, ANOVA, regression, measurement system analysis and designed experiments.

    How long does it take?

    Most projects run for two months or longer. The question, the data available and the trials the plant can run set the duration. We agree the investigation plan and its review points before data is collected.

    Related resources

    Talk first

    Discuss your service requirement

    Tell us the issue, location and deadline. We can then agree the responsibilities, reporting and scope of support.

    Discuss Advanced Problem Solving

    Your Privacy Matters

    We use cookies to analyze site usage and improve your experience. Learn more