Data Accuracy Not So Accurate

Six Sigma – iSixSigma Forums General Forums General Data Accuracy Not So Accurate

This topic contains 3 replies, has 2 voices, and was last updated by  Jim 2 years, 1 month ago.

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
  • #55821


    Hi, I would like to ask how can be an accuracy be more interpreted realistically. My case here is that we process jobs in a month for about 100,000 tasks with 15 errors to 12 errors a month. Our accuracy is always 99.99% or 99.98% which the client disagreed that 99.99 is not accurate if it makes a lot of errors every month. How can we interpret the accuracy to the client by not giving the 99.99%



    At a high level, since you don’t have a lot of detail in your questions…Your first step is to define an error from the voice of your client/customer.

    If your definition of an error doesn’t match theirs you need to recalibrate your measurement system to use their definition. I have worked on many projects where internally the errors/defects look minimal, however when you dig in further the client/customer is not getting what they need so the project team has to rework the measurement system to capture a true error. Almost always the process is performing worse than originally thought.

    I would suggest doing a search on VOC, (voice of customer) on this site or ‘Google’ it to help in how you should be measuring errors and defects to client/customer’s expectations if you haven’t already done so.


    Chris Seider

    It would seem to be easy to count the # of errors if it’s already known to be 10-15 per month. That’s the ultimate accuracy–assuming the system to catch the errors is correct.



    I’d agree with the VOC investigation – perhaps the issue isn’t data accuracy, but more the nature of the errors. I saw some data a while back that stated there were almost 10,000,000 flights originating in the US in a year. At the error rate you mentioned, that’s between 1200 & 1500 flights where something goes wrong; probably an unacceptable rate! The same could be true for medical testing. Sure 99.99% sounds great, but is it an acceptable rate for the customer?

Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.