Joined: 15 May 2007 Posts: 774 Location: Baton Rouge, LA
Posted: Sun Sep 09, 2007 11:09 am Post subject: Is There A Little Of Achille In All Of Us?
For any following this series, in part two of the Achille series, the final question was:
Realizing Achille’s position and with a knowledge of common and special causes and SPC, how would you advise Achille to proceed?
Achille needs a more enlightened understanding of management. His is a very common mistake and one often taught in business classes everywhere. A misunderstanding of data usage fosters short-term thinking and “knee-jerk” reactions to meaningless variation.
Numerical data can be very helpful, even if it is historical in nature. It can also be very misleading if improperly interpreted. In order for data to be helpful it must be seen in a context. Simply looking at a group of numbers and trying to draw information from it will be frustrating. It can also consume a great deal of resources while only making things worse.
In order to make data more useful it must be logically grouped. One of Achille’s mistakes is not choosing a logical grouping for the data he was trying to review. For instance, each month has a different number of working days. Figures based on such a group will naturally vary because of this difference. This makes the data less appropriate for comparison, one group to another.
Choosing a grouping that will be more meaningful requires several considerations.
Is it logical. A month is not a logical group for data based on daily sales, is we wish to compare one to another.
Does it occur often enough to give timely information. Grouping information by the year may be logical, but it will take a great amount of time to spot patterns. By the time the message is seen it may be too late to react.
Can information be accurately obtained without a great deal of effort. Hourly information may be useful, but may be impractical to obtain.
Does the group “fit” the information. For instance, with measuring sales, how long do sales take to complete. If most sales are completed in the same day they are started, daily measures may make sense. If jobs normally take three days, three-day subgroups might work better, and so on.
Achille decided most of his jobs spend two days in the shop and he can easily obtain daily sales data. Taking his past sales, he formed two-day subgroups. Now when a holiday occurs, the data is unaffected, he simply combines the day before and the day after the holiday. Weekends and the number of days in the month are also no longer a factor.
These two-day subgroups are then plotted on a run chart and the mean calculated. The mean is $7843.00 per two days. The upper and lower control limits might also be calculated. The square root of $7843 is about $280. Three times $280 is $840 so the upper limit is $8683.00 or $7843 [mean] plus $840 [3 times the square root of the mean.] The lower control limit is $7003.00.
This is an expanded view to make the variation more clear. The upper and lower limits are not shown.
Achille has an extremely stable system for producing $7843.00 every two days or $3921.56 per day, as previously mentioned. This means there are likely no special causes in the system. The daily variation is due to common causes and is basically meaningless. A trend might be demonstrated in any number of ways. For instance seven or more points all on one side of the mean might signal a trend. Also six or more points increasing or decreasing with no change in direction.
With this simple tool, Achille can now more accurately spot trends. He can judge effectiveness of changes to the system, IE advertising, sales approaches, etc. He may also predict future sales, unless the process changes. Achille is basically looking at the same numbers, just in a more meaningful method.
As one example, this method can be used to keep management from asking technicians to explain every little variance in their flat rate or productivity numbers.
Joined: 15 May 2007 Posts: 774 Location: Baton Rouge, LA
Posted: Mon Sep 10, 2007 2:00 pm Post subject:
Hi Bud,
Bud wrote:
As one example, this method can be used to keep management from asking technicians to explain every little variance in their flat rate or productivity numbers.
That's a great example of separating the signal from the noise. What a waste of time!
I have found it best to teach techs to track their own numbers. This removes a lot of the pressure to manipulate the figures. Rather than a threat, it can be seen for what I think it should be, a tool for improvement. This is greatly aided if pay is also not dependent on the numbers.
A tech knows his mean production. He employs a new method and can see if it is effective or not. The aim is to always increase production, while increasing quality. What better way than at the source of the production?
Another major key is removing fear. As long as people are rated based on production, there will be faked figures, excuses and little improvement, in my experience. Thanks Bud, I appreciate your reply.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
FAQ
Search
Memberlist
Usergroups
Profile
Log in to check your private messages
Log in
