iSixSigma

Proving a reduction in support tickets per month

Six Sigma – iSixSigma › Forums › Old Forums › Software/IT › Proving a reduction in support tickets per month

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • August 9, 2007 at 1:26 pm #26674

    mcleod
    ★ 10 Years ★
    Member

    I have monthly data of support tickets to a help desk and am trying to provide a reduction after improvements were made to a process and therefore should reduce overall tickets on this specific issue.  The “before” mean was 228 tickets per month with an “after: mean of 150 tickets per month.
    Should I treat this as attribute data?  My first instinct is that this is counts data but I am challenged with using the right tools to prove a significant reduction.  The C control chart via Minitab seems to graphically show a significant reduction. 
    Because I have all tickets for each month would a hypothesis test even be necessary as no sampling was involved?

    0
    August 9, 2007 at 7:06 pm #64804

    Taps
    ★ 10 Years ★
    Member

    Scott,
         I’m working on a project exactly like yours. We’re trying to reduce the number of trouble tickets related to email problems for our client. What steps did you take to reduce the mean from 228 to 150? Do you have any suggestions?
    I’m rather new to Six Sigma, but I would treat this as attribute data since trouble tickets are something you count (counted data). Have you tried using a p chart?
     

    0
    August 9, 2007 at 8:46 pm #64805

    mcleod
    ★ 10 Years ★
    Member

    This was a level 1 project so the team actually used some basic Voice of the Customer (VOC) activities to identify specific customer pain points.  These used this information to find a way to better notify customers of processing failures with regularly scheduled updates.  The quantity of tickets were then observed to lower with the c control chart showing a reduction of the mean as well as the control limits of the process.  Again my question is more around is the c control chart indeed the right choice and is there an appropriate hypothesis test to show statistically significance when comparing the before & after scenarios with monthly ticket data like this.  This all assumes a form of inferential statistics like this is even needed given that we have the total quantity of tickets for each month – my inclination is that we do not need it here.

    0
    August 10, 2007 at 5:38 pm #64806

    mcleod
    ★ 10 Years ★
    Member

    Finally was able to dig up my copy of Understanding Variation/Managing Chaos by Donald Wheeler.  In this case, like most others, the XmR (Individual Values and a Moving Range) chart provides the best application as it is empirically based and the general case of the other types of charts such as p, np, c and u control charts.  The p and np can be used “only when the binomial probability model is appropriate” (including a need for the probability of success for each item to be constant).  The c and u can be used “only when the Poisson probability model is appropriate.”  While both may be able to provide added value I do not believe my data come from these specialized distributions.

    0
    September 12, 2007 at 5:07 am #64832

    sumant
    ★ 10 Years ★
    Member

    Scott,
    To check whether there is reduction in tickets per month, you can use Chi-square test to check the significance. As you are counting the tickets per month, it is discrete data and Chi-square test is best suitable for this kind of data.

    0
  • Author
    Posts
Viewing 5 posts - 1 through 5 (of 5 total)

The forum ‘Software/IT’ is closed to new topics and replies.