What is Span
Six Sigma – iSixSigma › Forums › Old Forums › General › What is Span
- This topic has 20 replies, 14 voices, and was last updated 20 years, 2 months ago by
Leung.
-
AuthorPosts
-
May 20, 2002 at 5:12 am #29475
Dan HickeyParticipant@Dan-HickeyInclude @Dan-Hickey in your post and this person will
be notified via email.Does anyone have a nice clear and detailed explanation (graphs, powerpoint?) of SPAN that they use to explain the concept and how it works to their customers???
0May 20, 2002 at 6:12 pm #75617I don’t know how it works, and I’m afraid to find out what’s in it, but Spam has been around for ages. I’ve never seen a Powerpoint presentation on it. I just take it out of the can, lick the Spam jelly off and go to town. It makes a great SS training too, “The Uses for Spam”. Just make sure you don’t limit yourself to edible options. In a pinch, you can even throw it in a blender and use it for brake fluid.
0May 20, 2002 at 6:28 pm #75619
GabrielParticipant@GabrielInclude @Gabriel in your post and this person will
be notified via email.Farley: Dan is not asking about “SPAM”, but “SPAN”.
0May 20, 2002 at 8:56 pm #75623There was a thread discussing this topic earlier. Span can be defined as
span = (customer WANT date – actual RECIEVE date) for a product or service0May 21, 2002 at 5:06 pm #75635
PatrickParticipant@PatrickInclude @Patrick in your post and this person will
be notified via email.From what I’ve gathered over the last little while, span is a metric that is used to measure the ability to meet customer expectations. You could look at it as range or variance, if you wish. GE refers to span in their annual report ever since they have become more customer focussed. For example, if you are to deliver a product on day x, your performance might show a range (or span) of plus or minus 2 days. The precept is that your customer is not necessarily satisfied until your span is 0. You can find out more about it by visiting the GE website.
0May 23, 2002 at 3:09 pm #75691Another example of a company (GE) creating (complicating) a simple concept.
This is simply a measure of the lack of timeliness of a delivery.0May 23, 2002 at 5:50 pm #75701
billybobParticipant@billybobInclude @billybob in your post and this person will
be notified via email.Spam is the other funny meat..not to confused with my possum meat.
“Go Red Sox….its the big weekend series against the Yankees”
Billybob0May 23, 2002 at 6:16 pm #75704Yes, another metric…
Span merely tries to describe the variation in delivery. (just as standard deviation would)
We currently live and die by the % on-time delivery metric here at company X.
Does anyone know of other metrics to describe timeliness of delivery Should we use standard deviation? As a customer I would rather have my product delivered “early” than “late. However, we cannot assume that every customer wants everything early if they can get it. Some customers actually give us a “span” to ship.
Any thoughts?
0May 23, 2002 at 7:58 pm #75715
Carlos T.Participant@Carlos-T.Include @Carlos-T. in your post and this person will
be notified via email.Greg,
I think you nailed it on the head. The best way of determining the span is by asking the customer what is acceptable for “early” and “late.” By knowing the upper and lower specification limits you can then determine defects and opportunities and calculate the process capability based on continuous data.
CT0May 23, 2002 at 9:35 pm #75725
PatrickParticipant@PatrickInclude @Patrick in your post and this person will
be notified via email.I agree with you but we have to be careful. As a customer, there are costs associated with both early AND late deliveries. They are not the same. Early implies carrying unnecessary inventory and creating opportunities for losses and damages while late invariably implies impact to production…..
0May 24, 2002 at 7:24 pm #75775Span is a measure of range, generally around some target like due date. In some cases it’s the highest value minus the lowest value, or it could be the 95th percentile value minus the 5th percentile value, or some such calculation. It can also be expressed as a ratio of the two numbers.
For some customers, it doesn’t matter that you did something quickly, just that they received it when they expected to. In order to reduce variation around customer expectations, it is helpful to see how far off from the target you are on a continuous basis rather than how quickly you fulfilled the order.
In addition to Patrick’s comments about the “evils” of early delivery, if “early” happens frequently enough, it becomes the target in the eyes of the customer.0May 27, 2002 at 4:15 pm #75804
Mike CarnellParticipant@Mike-CarnellInclude @Mike-Carnell in your post and this person will
be notified via email.Patrick,
You made some very good points. Typically the people who don’t mind it coming in early are the ones that do not understand the cost associated with early arrival. It is a security blanket, just a inventory has always been. It makes the production and puchasing people feel all soft and warm inside to see it sitting there. They don’t pay for the material.
Late obviously does not work.
We did this type of thread once before. I didn’t understand then why we needed “span” since we already have range and Standard Deviation. Is it really just the “we are different” mentality. I have never had anyone explain what the advantage of “span” is over the regular measures of dispertion.0May 27, 2002 at 7:41 pm #75810
PatrickParticipant@PatrickInclude @Patrick in your post and this person will
be notified via email.Mike, I think you are right; it is a “we are different” mentality. Or perhaps it is a way to reach people who know little about (or are intimidated by) statistics…
In their 2001 annual report, GE refers to entire business units reducing their span by 70% (looks good to a layman but doesn’t really say much to me…).
Patrick0May 27, 2002 at 10:26 pm #75812
ktrainnParticipant@ktrainnInclude @ktrainn in your post and this person will
be notified via email.SPAN is a variation measurement used typically for non-normal distribution’s. First take your data (let’s say delivery data) and subtract each data point from the customers required date, giving you a delta to commit. Next take your data and calculate the 10%, median, and 90%. The SPAN is the difference between your 10% and 90% number. So if you deliver a product 10% of the time 13 days early or earlier (-13 days) from your customers required date and 10% of the time you deliver the product 14 days later or later than your customers required date (+14) then the total SPAN is 27 days (14 – (-13)). Companies target a SPAN of 2-4 days. Also depending on your sample size you can use 10/90 or 5/95 or 1/99 percentiles to measure SPAN. There is an improvement/analysis strategy used to uncover underlying processes that hold the best SPAN it is called segmentation and stratification.
Hope this answers your question.
ktrainn0May 28, 2002 at 1:34 pm #75831
Mike CarnellParticipant@Mike-CarnellInclude @Mike-Carnell in your post and this person will
be notified via email.ktrainn,
I am OK with how it gets calculated – basically it is a truncated range. What I don’t understand, I think Patrick answered it, is why we need to invent a thing called span. Just another iteration of the “we are different” mentality. Dr. Deming documented this mentality 40+ years ago in his Obstacles to improvement and it is alive and well today.
Thanks for the explanation.0May 28, 2002 at 4:11 pm #75840
ktrainnParticipant@ktrainnInclude @ktrainn in your post and this person will
be notified via email.Point taken. My question though, is what metric do you use to measure the variation in your delivery process? I think this is one of the reasons why SPAN and similar metrics were developed.
Also Stability Factor = (1st quartile) / (3rd quartile) from a BoxPlot.
Continuous Metric Improvement, maybe I am wrong though?
Ktrainn0May 28, 2002 at 4:53 pm #75843
Mike CarnellParticipant@Mike-CarnellInclude @Mike-Carnell in your post and this person will
be notified via email.ktrainn,
The bottom line is that if “Span” is working then we really need to leave it alone. It just seems so infantile to have people latch on to a metric that is really just a spinoff of range as long as it has their own name. Seems like a waste of time and effort.
One of the original advantages that Motorolan’s speak about with the advent of SS was the common language. It seems a little self defeating to bring in dialects such as “Materialese” because the materials/Purchasing function refuses to speak basic stats.
Thanks.0May 28, 2002 at 6:02 pm #75849Mike,I think Patrick has the answer. SPAN is simply an easier way (vs. Variance, or Std Deviation) to get everyone involved; particularly those with less formal academic backgrounds. As long as it promotes the right behavior – I don’t see anything wrong with it. The more we make it tangible for people the better off we’ll be. Ever try improving a delivery process without finding a way to get the warehouse team involved?PAB
0May 28, 2002 at 6:04 pm #75850
PatrickParticipant@PatrickInclude @Patrick in your post and this person will
be notified via email.ktrainn,
In answer to your question…As a customer, I don’t like the span metric because “if it ain’t here when I want it, it’s a defect”. So I really think that we could look at this from a discrete data perspective and assign a sigma value that way. It’s harsh but we could always look at the COPQ data afterward to illustrate the need to be “on target” versus “give or take a few” days. In the automobile industry, we measure deliveries in terms of minutes, and the penalties are severe….
Cheers,
Patrick0May 28, 2002 at 6:51 pm #75853Mike,
SPAN is simply an easier way (vs. Variance, or Std Deviation) to get everyone involved; particularly those with less formal academic backgrounds. As long as it promotes the right behavior – I don’t see anything wrong with it. The more we make it tangible for people the better off we’ll be. Ever try improving a delivery process without finding a way to get the warehouse team involved?
PAB0May 29, 2002 at 12:58 pm #75875I seems to me that it is important to use the metrics that people are comfortable with (as long as there is some rationale behind them). My job is to encourage the use of the metrics to continuously improve.
With time, you can usually help a few people move to more statistically sophisticated techniques. But please realize that, for most people, standard deviation is a very statistically sophisticated concept.
My 2 cents.0 -
AuthorPosts
The forum ‘General’ is closed to new topics and replies.