iSixSigma

Control Charting Revenue

Six Sigma – iSixSigma Forums Old Forums General Control Charting Revenue

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #50719

    IBV
    Participant

    Hi,
    I need some help with control charting. I have a store where I am a supplier of widgets. The revenue is pretty consistent on the weekdays and then falls off on the weekends (b/c most of my customers are manufacturers.
    Is it possible for me to control chart my sales by dollars sold per day to see if it’s out of control at any point? My sales are pretty consistent week-day after week-day.
    What control chart should I use?
    I suspect that my data is ‘variable’ and would want an x-bar, R chart but I don’t understand the sample size determination. (constant sample size, 3-5 for x-bar, R | variable sample size, >=10 for x-bar, S chart). What’s my sample size? A week?
    How do I treat my weekend days (pull them out?) when revenue falls off? I don’t want them to trigger an out of control situation. Maybe I just leave them in?
    I have data going back a year.
    My daily sales look something like this (numbers modified to protect data):
    8/8     $21,093.128/7     $20,045.238/6     $21,100.008/5     $19,381.988/4     $19,321.018/3     $10,304.39 (sales fall off on sat/sun)8/2     $7,493.23 (sales fall off on sat/sun)8/1     $21,203.23
    Here’s what I want to do: make sure my weekday sales remain constant. If they get out of control (statistically), I want to know it immediately so I can figure out if one of my supplier relationships is broken or advertsiing isn’t working, etc. — so I can look for the root cause.
    Thanks for any help you can offer.
    IBV

    0
    #174742

    annon
    Participant

    IBV,
    You are dealing with $$, a continuous measure, so you can use XbarR, XbarS, or IMR charts (among others).  My two cents…
    I would use an IMR chart, taking a daily sample Mon-Fri (n=1).  I would go back and test the historical data for normality and ensure I preserved the time order.  If app. normality is not present, then subgroup your data (ie daily average v. daily sum, weekly vs. daily) and see if that moves you closer to normaliy.  
    Or, you could simply manage to the chart knowing you have a non-normal parent population and simply use extra caution when investigating special causes….CC is not an exact science…it is meant to be practically interpreted in some cases…..Good luck.
     
     

    0
    #174746

    Brandon
    Participant

    You’re applying far too technical tools to a simple situation. Plus, you’re analyzing the “Y”, not the “X’s”.
    You want to know changes in your customer’s buying habits then track that metric – not total sales.
    You’re not applying DMAIC, dude, you’re trying to force “tools” on something…anything. Go back to basics.

    0
    #174747

    annon
    Participant

    What what you suggest as an alternative method of achieving the stated goal?

    0
    #174748

    Brandon
    Participant

    Start at the beginning. Write a Problem Statement, an Objective Statement, identify X’s, determine critical X’s, track critical X’s…..
    This guy jumped right into analyzing Y’s. Follow DMAIC….it works. Jump into the middle, you have garbage.

    0
    #174751

    Darth
    Participant

    Annon,
    Shewhart just turned over in his grave. You don’t arbitrarily subgroup data to try and get a normal distribution. What happened to rational subgrouping and the idea that Shewhart developed his charts to be very robust to normality constraints. Bad advice.

    0
    #174754

    annon
    Participant

    Duly noted.  Thanks for the correction.

    0
    #174760

    xu
    Participant

    Brandon,
    You’re trying to force DMAIC on everyone. IBV isn’t saying there’s a problem yet, just that they want to measure their process. I don’t see why control charting IBV’s store’s revenue is a problem.
    You’re assuming that there’s a problem and that the X’s need to be measured.
    I don’t see why you can’t control chart revenue. When there’s a problem, then you look for the X’s. Wouldn’t you do the same for the number of cars coming off a line?
    Barry

    0
    #174763

    Brandon
    Participant

    You make some good points Barry; however, look at the last line of his post. That made me think he has some issues other than revenue he wants to track.
    Rather than watching revenue then saying “Oh, I have a problem, I’d better check some other stuff.” Why not track the X’s for revenue from the get-go?

    0
    #174765

    Michael Mead
    Participant

    Try using a week as a subgroup. If any one day is unusual, it will show up on your range chart.

    0
Viewing 10 posts - 1 through 10 (of 10 total)

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