how to calculate % improvement
Six Sigma – iSixSigma › Forums › Old Forums › General › how to calculate % improvement
 This topic has 8 replies, 6 voices, and was last updated 16 years, 9 months ago by Brit.

AuthorPosts

December 21, 2005 at 10:36 am #41791
I am facing the difficulty in calculating the % improvement of a process. Initially, the process is having the mean Xbar1 and standard error of mean S1. Then, I have improve it to Xbar2 and S2.
Now, how can I calculate the % improvement of the process? Just using the mean (Xbar2 – Xbar1)/Xbar1? In this case, I am not take care of the spread of the mean.
The second method is, if I have the 95% confidence interval for both case, can I use the lower bound of the improved process minus the upper bound of the initial process? in that case, what is my error? still 5%?
The third method is I test the mean using 2 sample t test (95% confidence), the lower bound of the mean difference would be the minimum process improvement within 95% confidence level. However, how can I calculate the % improvement? should I divide it by the upper bound or lower bound of the mean of the initial process?
Please advise.0December 21, 2005 at 10:52 am #131446You find out the process capability of process (Before and after improvement). The difference will give you the improvement you have made. If your project was to reduce the shift in your process (Zst Zlt), find out the Z shift (before and after improvement). If the Shift has reduced then you have improved your process.
0December 21, 2005 at 1:28 pm #131454Siva,
How do you attach a $ value to the improvement ?
Vinod0December 21, 2005 at 3:56 pm #131456There are different kinds of improvements. The $ saved or % improvement depends on what the improvement was. If you saved time on processing a part then you should be able to calculate the total number of minutes saved in a year. Each minute should have a dollar amount allocated to it.
You can also have an improvement that doesn’t necessarily save money, but makes the work “ergonomically” better for the employee. This type of improvement is difficult to calculate, but should benefit the employee in the long run.
Jeremy0December 21, 2005 at 4:55 pm #131457I suggest avoiding % as a metric. The main problem is, and we’ve seen it, that a percentage is some numerator divided by a denominator (N/D) and if the denominator changes, it will generate a false sense of improvement.
Example, % hours spent in rework = (rework hours )/(total hours)
If the total hours are decreased due to some other reason than rework, you’ll see the % increase indicating that your improvement had no effect. Similarly if total hours are increased and there was no change in rework, you’ll have a false sense of improvement.
Keep your units real. Not only do you avoid % issues, but it is easier to convert to $.
Instead of trying to normalize something, select a metric that is already normal.0December 21, 2005 at 4:55 pm #131458Jeremy,
In my case the goal is to reduce the defect rate by 10%.Unfortunately there is no micro measurement system in place to really come up with an accurate savings amount
I have calculated the process sigma level and it is 1.8.Once I improve the process how do I attach a $ amount with respect to the reduction in defect rate?
Could you help me out here
Vinod
0December 21, 2005 at 5:01 pm #131459Try some of these ideas:
Cost/Year = (Labor Cost/Year) + (Material Cost/Year)(Labor Cost/Year) = ($ Rate/hr)*(number processed/year)*(hr/number processed)
(Material Cost/Year) = (number processed/year)*(Material Costs/number processed)
(Savings is the delta between baseline and improved)
Or the approach from The Goal
Profit = Throughput Sales – Inventory – Operational ExpensesThroughput Sales – Rate at which the system generates money through sales (not production)
Inventory – all the money the system has invested in purchasing things it intends to sell (if you avoid cost/unit mentality, you’ll think of more things and avoid the pitfalls of production accounting)
Operational Expenses – all the money the system spends in order to turn inventory into throughput (see above note)
Hope this helps.0December 22, 2005 at 2:13 am #131481Siva,
I got a few questions:
1. Sometimes the value of improvement alone is not enough to conclude anythings. For example I have improved the cycle time by 100 hours, but if the original process take 10000 hours, I only improved by 1%. However is my origianal process take 200 hours, I have improved it by 50%. Thus I can say that the second improvement is better compare to the first improvement, am I right?
2. If we only comparing the mean of the improvement, doesn’t take care of the spread (standard error of mean). there may be a high possibility that the initial and improved process are overlapping. For example, if I got one process initially the cycle time is 100 hours (95% confidence level is 80 to 120) compare to improved cycle time mean is 50 hours (95% confidence level is 40 to 60), then can I said it is at least reduced by 20 hours within 95% confidence level? The second case is the initial process cycle time is 100 hours (95% confidence level is 70 to 130) and the improved process cycle time is 50 hours (95% confidence level is 25 to 75), then the improved process seems may be worst than the initial process in some period of time. Thus the improvement for the second case is not as good as the first one, am I right?0December 23, 2005 at 4:25 pm #131565A good way to handle this is via cost of quality. In terms of your problem, there are a few costs associated with the defects.
1. Cost of waste (materials/supplies/energy) of producing an unusable product.
2. Cost of rework if you replace the defective product (if you know how much it costs to produce 1 units, then multiply that ny the cost of replacing x defective units.
3. Cost of inspection – if you improve a process enough so that it does not produce defects, then you reduce the need for inspection (labor/materials/etc.)
4. Potential cost Reduction. Cost of losing a customer when/if they receive a bad product – if they do not buy from you anymore because you cannot meet their specs, then what is the potential cost to the organization.
By the way – when talking to management, it is often better to talk in terms of revenue generated rather than cost avoidance, if you can do that. Not always possible, but when it is, use revenue. 1, 2, & 3 above can be termed in revenue generated. 4 would be a cost avoidance.0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.