iSixSigma

Setting Targets – looking at 3rd Quarter

Six Sigma – iSixSigma Forums Old Forums General Setting Targets – looking at 3rd Quarter

Viewing 12 posts - 1 through 12 (of 12 total)
  • Author
    Posts
  • #38294

    Bambace
    Participant

    I need to set cycle time targets around a process.  We currently look at the median, mean and span to measure our process.  I heard that if you cannot set targets based on historical data then you can you use your Q3 as a target.   Any truth to that?   I am struggling with setting a target.
     
    HELP

    0
    #114462

    Peppe
    Participant

    And also Q3, Q2, Q1 ..  was without target ? Yes, it can be an opportunity for you to set a target based on your last results, so that you can easy achieve it. But a simple question : how this company work ?  How it know what have to do and how do it ? What management do ?
    Anyway, you can agree your new target based on historical data or Qx average and set it higher enough, depending of effort/improvement  you plan, at least, in any case it is an improvement activity.
    Rgs, Peppe

    0
    #114489

    Marc Thys
    Participant

    Jessica
    Targets should be set based on the Critical Customer Requirements, not just based on historical performance!!!

    0
    #114532

    arvind pathak
    Participant

    Jessica,
    Do hypothesis testing and select a statistically significant target. Ensure your target does not fall within confidence interval(95%band).
    Taking Q3 as target is misleading.
    Arvind Pathak

    0
    #114533

    indresh
    Participant

    yes its fine if you are ok with 95% as target in case your historical mean is absent. ensure data is normal
    thanks
     

    0
    #114545

    Bambace
    Participant

    Thanks for the help.   My data is non normal (p-value > .05)    I still do not understand if its good to set a traget based on a Q or not.   I have historical data but do not know the next steps. Can anyone help?  I can email a sample run of the data if possible
    AHHHHHH

    0
    #114567

    mand
    Member

    Jessica, Per my understanding if P>.05, its a good enough indication that your data is Normal.

    0
    #114820

    Michael Schlueter
    Participant

    Hi Jessica,
    In my view it doesn’t matter much whether you base your targets on one or more quarters, when you focus on your ultimate goal.
    I suppose your ultimate goal is a stable cycle time process, probably with lower processing times (both in mean and span) than today.
    If so, your process is likely to be ‘out-of-tune’ anyway. So I suggest starting with reasonable targets, trying to stabilize your process within this range and to re-adjust targets towards your ultimate goal step by step in future.
    This way you intentionally drive your cycle time process through a series of (intentionally) ‘out-of-control’ and (achieved) ‘in-control’ stages. At each re-adjustment step you increase focus on the next major barriers present in your cycle time process at that time – and eliminate them step-by-step.
    Hope this helps. Kind regards, Michael Schlueter

    0
    #114824

    Dan Fong
    Participant

    Jessica,
    I supposed you are trying to reduce cycle time. With no benchmark or customer spec, I believe it is reasonable to use the 3 quartile point from your historical data as the initial target. That means your goal is to go from the median to 3Q. Make sure you set up control charts to monitor on-going cycle time and update the charts ie. control limits when you have collected more data.
    Dan Fong

    0
    #114841

    AlStats
    Participant

    Jessica, you should not set numerical arbitrary goals, better than that is to listen what your process is trying to tell you in regards with the voice of the customer. As Dr. Deming stated in his book “Out of the Crisis” page 20, making reference to Dr. Lloyd S. Nelson: “If you can improve productivity, or sales, or quality , or anything else, by (e.g.) 5 per cent next year without a rational plan for improvement, then why were you not doing it last year”. One simple and easy way to listen to your process is the use of process behavior charts (control charts), they can help you see if your process is predictable (only random common variation) or not (out of control or special variation is affecting your process). When the process is predictable you can use the mean of the process as a good target without forget that the process limits (control limits) are good indicators of what to expect in regards with variation.
    Saludos!
    AlStats
     

    0
    #114846

    Mike Carnell
    Participant

    Jessica,
    I am not sure why anyone would use a past quarter as a target. Unfortunately there are a lot of people who do that Yoda thing and just close their eyes and use the force to guide them which doesn’t make much sense when we are supposed to be driving people to use data to make decisions. I can’t imagine why anyone would want to use data to manage that data driven program – when we can simply attach our old management style to the “culture shift.”
    This is where we begin. Map your process – not sitting in a meeting room while experts do their best to impress you. Go follow stuff around. It doesn’t have anything to do with what people believe the process should be. Your cycle time is a product of what the process is. We calculate a thing called entitlement – the best you can do with the process you have. Time only the value added portions of the process and do not include queue times, transportation times, rework, etc. Basically if you were to eliminate all NVA from the process – if you set a target without knowing this you run the risk of setting a target below what can be achieved unless you make some serious technology changes. That can be a different issue.
    Now you have two numbers. Where you ran historically (mean & std. deviation) and the optimal the process (this should be a mean and std. deviation as well) will produce. Everything on the Process Map that is NVA is a target. Time them and see what your pareto tells you decide where you will target projects and what you believe the effect will be (some will have a target to shift a mean, some reduce vatriation, some both). Where you have a part of the process feeding into a rework process – get a GB or a BB and have them work a project to get rid of the rework. You can them roll up the target project savings time and propose a number that is the reduction for the entire process. Someone needs to stay on top of the whole thing and project manage it.
    Just my opinion.
    Good luck.

    0
    #114892

    KWalker
    Participant

    As others have said, your target should focus on what your customer expects, not an unrelated statistical measure of how it’s doing now. Do you have a quoted turnaround time, or a contractual agreement? Another thing to consider is that it’s very often the variation in your process the customer feels, not the mean or median. It’s the tail, when you are very late, that probably gets the most attention – focus on reducing the span of your process. Even if your median meets the target, they still can’t count on when you’re going to deliver, and that can cause them a lot of frustration. If you accomplish that, and get those sources of extreme variation under control, the central measure may well also shift. Back to setting a target – focus it on reducing that span, say from 5th to 95th percentile by some amount, or bringing 95th percentile in from 20 days late to 5 days late, etc. If you subdivide your process output, you may well find some subset that performs very well – over some short time period, one person, one group, etc – use that as your target, and try to make your worst as good as your best – that is what your process is capable of when it’s under control – you just need to get it there.

    0
Viewing 12 posts - 1 through 12 (of 12 total)

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