iSixSigma

Transactional Accuracy – What is the right metric?

Six Sigma – iSixSigma Forums Old Forums General Transactional Accuracy – What is the right metric?

Viewing 6 posts - 1 through 6 (of 6 total)
  • Author
    Posts
  • #52155

    Anoop
    Participant

    Hi – I have initiated a project to improve the Accuracy of a process which is already performing at around 99.5%. Since any error would have a financial impact, there is a requirement from all stakeholders to improve it to a near zero defect stage. There are several sub processes and the process undergoes a 100% Quality Check as a stage 2 activity. My plan is to look at First Pass Accuaracy , ie, look at all errors reported by the Stage 2 QC than the one corrected at QC stage and the final accuracy reported to customer. Now, my question is what is the right project metric for me – should it be the first pass error count or the First Pass Accuracy % and  should I be looking at the data daily, weekly or monthly data for analysis. Given the already high performance of the process, I suspect that the data is most likely going to be non-normal. I did a quick analysis basis monthly data and the data was non normal. Still I did a quick correlation check between errors and volumes – If I use error count the correlation is significant and I use accuracy % the correlation is non significant. Given that close to zero defect is what I am aiming at, I was thinking maybe error count might be helpful for analysis rather than accuracy %. Please let me know your thoughts. Thank you.

    0
    #183092

    GB
    Participant

    You need to pin your Stakeholders down to what “near-zero” means…Sounds like a BS metric…sloganistic.
    FPY is a starting point…Make sure it is true FPY and not “test/tweak until pass”.
    In addition to FPY, you’ll want to break it up and start measuring RTY to get a better picture of sub-process variation.

    0
    #183095

    Anoop
    Participant

    Thank you for your suggestion. Everyone would really like to see zero errors coming out of the QC stage atleast given the financial impact of the error. Could you comment/share your thoughts on Error Count Vs Accuracy % and also on the frequency of the metric (daily/weekly/monthly) for purposes of analysis.
    Many thanks..

    0
    #183113

    Severino
    Participant

    Would it be feasible for you to control chart the time or number of transactions between errors?  Since your management hasn’t given you a clearly defined goal, a control chart might.  If you can move your process average above the UCL you could state that you made a statistically significant improvement to the process.  The advantage of working with a control chart is that the underlying distribution is irrelevant in many cases so you need not be concerned about non-normal data.

    0
    #183124

    Kluttz
    Member

    This is probably going to be spectaularly unhelpful and beside the point, but I’ve never let that stop me before;
    When you’re getting in the range of a 99.5% yield, I think you’re better off re-visiting your defect definition and OBF’s than you are focusing on the remaining 0.5%.  If your out-of-box failure rate is > 0.5%, who really cares what your in-house yield is?  If your OBF% is greater than 0.5, you’d getting a bigger bang for your buck by doing a RCFA on those and rolling poka-yoke into your production (or potentially shipping) process to prevent those defects from occuring. 
     

    0
    #186830

    Sharma
    Member

    Hi,
    You have received very good responses, just to add to those valuable suggetion, I would like to tell you that classify your defects in some defined categories & assign waitage according to there occurrences and then define a factor of impact of these defects. Now if you can take a “sum product of  ( No. of defect X wait age of category X factor of impact)” you will get a number, track this number. This number should tend to zero. If you are dealing with same component or nature of work is same then you can use that as denominator.
    Do let me know if this helps you or not.
    Regards,
    Yogesh
    [email protected]

    0
Viewing 6 posts - 1 through 6 (of 6 total)

The forum ‘General’ is closed to new topics and replies.