iSixSigma

MSA in transactional process

Six Sigma – iSixSigma Forums Old Forums General MSA in transactional process

Viewing 21 posts - 1 through 21 (of 21 total)
  • Author
    Posts
  • #40223

    Srinivasan
    Member

    Hello Experts,
    Need your advice on how to do MSA in transactional environment. I am working on a project to reduce the TAT to process a file. The problem I face is I cannot do a gauge R&R because once a file is processed it moves out of our queue and is blocked for any further activities. I work in a financial services firm which provides services for third party customers. This makes my work even tougher as my customers are still ignorant about six sigma benefits and I cannot ask them to give me some sample file to be reworked. Indeed they want to see the benefit of this project before talking about six sigma to them!
    Can I go ahead without doing a gauge R&R if I am sure about my operational definitions and the data collected?
    How else can I go R&R check???
    Appreciate immediate help!
    Uma
     
     

    0
    #124200

    “Ken”
    Participant

    Uma,
    I attempted an answer to this question recently on this forum.  I believe the gentlemen that received my answer has worked to use my suggestions.  Perhaps this link to the last Transactional MSA thread might help get you moving in the right direction, but not sure:
    https://www.isixsigma.com/forum/showmessage.asp?messageID=74732
    You can also search this forum for additional past threads on this subject.  Let me know how this info works for you, and if you need additional clarification.
    Best of Luck,
    Ken

    0
    #124204

    Srinivasan
    Member

    Ken,
    Thanks for the inputs. Let me try doing this and get back to you. But this would help me for tracking the end to end cycle time to process a transaction. If am looking for TATof certain sub activities that is involved, and the system does not capture these details, am not sure how to go about it then.
    Uma

    0
    #124207

    “Ken”
    Participant

    Uma,
    I’m not familiar with the acronym “TAT”, please define.  If transaction time is not presently tracked, or you can not acquire these data, then it looks like you may need to launch a data collection campaign.  Most importantly, how can you assess the accuracy and precision of something you don’t measure or cannot acquire measures for?
    Some things to think about:
    Do you have a process flow for the system you want to improve?  Can you include the nominal times at each step in the system based on your observations into this flow illustration?  If so, then you could sit back and ask two three things:

    Are there redundancies or wasted efforts in the process?
    Can the wasted efforts be removed?
    Where does the work spend most of it’s time?
    Give this post remarks some thoughts, and yell back if questions.
    Regards,
    Ken

    0
    #124223

    annon
    Participant

    Ken,
    When you mentioned the following in the earlier thread:
    …I suggest going out onto the floor, and select a random sample of at least 15 in bookings, and 15 out bookings. In fact, 20 bookings on each end is best. You can simplify this activity by selecting about 20 bookings sets randomly across the entire process flow. If a log exist, then compare the logged values to the database values by taking the differences. After computing all 40 differences…

    Could you define “in bookings” and “out bookings”? 
    Do you arrive at the “40 differences” by comparing the 20 inbookings and 20 outbookings to their sister measures in the database:  20 inbook differentials + 20 outbook differentials = 40
    Thank you for your patience.

    0
    #124228

    “Ken”
    Participant

    annon,
    I would use the definition for “in-booking” as being the transaction that occurs when moving a quantity of something, could be a lot, or a batch of product, into an area.
    An “out-booking” would be a similar movement transaction, but that movement is out of the area.
    In pairs, for a given unit of product moving through the area the workstation or area cycle time could be computed as {out-booking time – in-booking time}.  Various companies and production systems use different terminology for these terms.
    Hope this helps to clarify my previous explanation.
    Ken

    0
    #124230

    annon
    Participant

    As always, greatly appreciated.  Thank you.

    0
    #124243

    “Ken”
    Participant

    My pleasure!
    Good luck,
    Ken

    0
    #124248

    Peppe
    Participant

    Uma, without data is quite difficult to do an MSA, anyway you can try to measure the average of TAT = total qty files processed/time. You can do it per people, shift, workstation, as is nore convenient for your scope.
    I
    Rgs, Peppe
     

    0
    #124251

    Srinivasan
    Member

    Ken,
    TAT – Turn around time  / Transaction time.
    Let me put it this way, The system captures the intime and outtime of every unit that was processed. the process includes multiple steps which may or maynot be similar for all units. The nature of these units differ from one to another. Even if I could group them, there would be about 20 plus sub groups.
    The process uses two different applications to process a transaction. There are non-value added activities in the process. Considerable amount of time goes in duplicating the work because there are 2 applications. (which is why I wanted to look at the time taken for sub activities in the process)
    Though is there is a defined process map, its purely at the convinience of the analysts as to how they want to sequence their steps. An individuals way of work may not be easier for other.
    I have started data collection activity for transaction time (& sub activities) with detailed operational definitions. But how would I do MSA? Is it necessary?
    Thanks for the help.
    Uma
     
     

    0
    #124276

    “Ken”
    Participant

    Uma,
    The system sounds like it has way too much complexity, and far too little discipline.  You mentioned that the system captures in-time and out-time for every unit that is processed.  How are these movements actually captured and recorded?  That is the essence of doing an MSA.  If there is error in capturing the movement transactions, then the estimates of TAT will have considerable error.  An MSA seeks to do two primary things: 1) capture the process used to conduct the movement transactions, and 2) measure the error in logging the movement transactions.
    Hope this is getting you closer to understanding.
    Ken

    0
    #124279

    Srinivasan
    Member

    Ken,
    When a person opens a particular file for processing, time he opened is captured under his ID. Similarly, when he comes out of the file once he is done, the time is captured under his ID again. So, there will not be an error in the way the time is recorded. But process (sequence of sub activities) could vary if the same file is worked by another operator, hence the TAT will also vary.
    I am not in a position to say whose work has more non-value adding activities and because of which the TAT increases. Also to notice that if TAT increases, then productivity goes down.
    If I understand you right, you mentioned “1) capture the process used to conduct the movement transactions” is what is a problem. Cos if I have to understand the movements and the time taken for the same, I need to capture data sub-activity wise. Or will something like a “time and motion” study help?
    Thanks again.
    Uma

    0
    #124286

    “Ken”
    Participant

    Uma,
    You are simply interested in knowing the error associated with starting the work, and ending the work.  It sounds like the start of the work is automated, opening an electronic file.  But, how is the end of work signaled?  If this is an automated operation, then there is little to no measurement error in computing the processing cycle time.  You MSA should consists of two simple process flows showing the front end initiation transaction and end transaction that points out where the actual measure is taken.  That’s it!
    Ken

    0
    #124290

    Srinivasan
    Member

    Ken,
    The end of the process is also captures electronically when the operator closes the file.
    So u say I need not bother about the sub activities while doing a MSA?? Well, I am glad if I need not… But what about the reproducibility factor???
    May be its a stupid question. When I get into analyze phase and I need to know how it took x minutes for a guy to process a file , I need to look into the sub activities. In this case, I only know my measurement is correct end-end but not for the individual activities. Can I still go ahead by collecting data manually for the same.
    Looks like my life is much simpler now.. Thanks :-)
    Uma
     

    0
    #124293

    “Ken”
    Participant

    Uma,
    Slow down just a bit…  To capture the total cycle time for completing the work requires the start time on the clock, and the end time on the clock, regardless of what goes on in between those times.  Does that sound reasonable to you?  Let’s not just wash our hands, and say we don’t need to do an MSA.  I instruct all of folks to insure they understand the way these two time endpoint measures are taken, and know why there is little value in performing an MSA.  In fact, during a management review of the Measure phase I ask that they explain clearly to the management team why they did not do the MSA analytics, not simply that it didn’t make sense.  I have them show the starting and ending transaction process flows and point out why there is no need for the MSA.  Does this requirement make sense to you?
    Ken

    0
    #124307

    hari
    Participant

    Uma,
    I agree with Ken. Infact today morning I have asked my GB to do the MSA for the Transaction processing. Our TAT is in days and Mesurement system is difference between Received Date(Manual Entry) and Processed date(System Entry). In he Measure phase we will be doing the MSA on Y and in this case it is TAT. But in this case we have to do the MSA on the starting point and ending point. I have asked my GB to collect the data based on Rational sub grouping principle and check whether Received Date is accurate or not against the Standard Date(We have a step in the process, the moment the document reach us we record it). Do the Accuracy check on the Processing date. Here, checking accuracy is more critical than the R&R. But still you can do Repeatability and reproducibility and with my experienec, typically R&R will not be an issue.
    I hope you also do the MSA(Accuracy, R&R) for your case and explain the benefit and application of six sigma to your customers.

    0
    #124338

    Srinivasan
    Member

    Ken,
    I get your point. The start time and end time is captured by the system and that data is available at the end of everyday.(The process cycle time is 24 hours max. so the files that come in at the beginning of the day goes out by EOD, The spill over becomes a backlog but its highly unusual to have backlogs in this process).  How ever, there will not be any measurement system error.
    Thanks a ton.
    Uma
     

    0
    #124855

    Laura
    Participant

    Good approach Ken,
    Only issue I see with the approach is you IMR charts can only be used for rationally sampled data not randomly sampled data. So a simple Excel chart will have to used unless you are taking a rational sample.

    0
    #124871

    “Ken”
    Participant

    Laura,
    How do you select a rational sample?
    Ken

    0
    #125029

    Zoeb
    Member

    Hi uma,
    If you conducting MSA for TAT then you should have a strong opertional defination of the start time and end time. And the procduer to collect should be defined clearly so that if you when you take trials the data what yu are capture for MSA will be be accurate and the system will give rigth type of figures
     
    regards
    Zoeb
     

    0
    #125255

    howe
    Participant

    Uma, I think this should be very similar to conducting a gage R&R under destructive testing. You would have to pick out batches of processed items (perhaps consecutive items or one day’s work – you need some rational sub-group of the data) with the assumption that items in each batch are identical. Then conduct your gage study using these batches. So operator #1 would run the process on day 1 & day 2 and you would check repeatability within operator #1. Operator two would run the process on day 3 & 4, and again you can check for repeatability for this operator. Now, with data from two operators you can check for reproducibility between them.

    0
Viewing 21 posts - 1 through 21 (of 21 total)

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