Calculating Customer Returns ppm

A

Andrews

We are calculating customer returns ppm as follows:

Customer returns ppm = (No. of parts returned by customer due to defects / No. of parts supplied) x 1 million

Does anyone feel this is wrong? If so why?

FYI, we are measuring and monitoring this monthly.

If someone uses someone other measurable for quality performance , please let me know.
 
Last edited by a moderator:
A

Al Dyer

I would ask if there is a time frame included?

If you sent a batch of 500 and they were returned, why divide them by the total number of parts supplied to the customer over the course of a year or so?

If the equation was total parts returned over total parts shipped over the life of the product, maybe.

I would also ask:

Does keeping this measurement lead to continuous improvement or is it cost effective?

Does the customer require the measurement?

Does the customer submit the PPM to you on their basis and are you then double dipping and using resources not required?

Looking forward to your response!!!

:bigwave:
 
S

Sam

Andrews, DCX uses this approach to rate their supplierrs.
No. of defectives items (to date)/No. items shipped (to Date) X 1,000,000. Anything over 50 PPM causes points to be deducted from your rating. If you do not furnish large quantities of parts your rating can diminish quickly.
This is similar to "zero defects", but with PPM you either achieve the requirement or you lose points.
IMO, it is a meaninglesss measurement for Mfg. purposes.
 
M

Mobobo

Andrews, it is an obvious mistake to count like that, if there is no delivery this month, but return parts delivered previous months, how to count for this situation? Now, I am confused about this also. IMO, we can get the PPM from our customer as our returned PPM
 
Last edited by a moderator:

Jen Kirley

Quality and Auditing Expert
Leader
Admin
Mobobo said:
Andrews, it is an obvious mistake to count like that, if there is no delivery this month, but return parts delivered previous months, how to count for this situation? Now, I am confused about this also. IMO, we can get the PPM from our customer as our returned PPM
This is the same concept that educators are pressing for changes in measuring success over time in improving test scores in students.

1. If you are measuring PPM failures in non-identical products, the measurement alone can't accurately reflect that changes in PPM differed through better process control--it could have been other reasons.

2. If you don't track and make the measurements over a sufficiently long term in the same group of parts (and students) but track in a short time period with different groups, same as #1 you can't reliably be sure the numbers reflect control. This means you would need to count defects by lot, not by time.

3. Cutoff periods for failure should be considered when making these counts. At what point do most failures in these parts' lifecycles occur? Is there a curve of failure, including while in use? And if so, is there definition of how interaction with other parts, or end use by customer contributes to failure?

I hope you can see why I do not like simple metrics like PPM failure rate. It's too easy to miss important data, and it's too easy to assume the method is accurate when in fact it can be woefully misapplied by the other party.

If you can show a curve of percentage of failures over time, IMHO you can show better value because expenses of part failure after placed in service also affect your buyer. If you can show the curve changing to reflect reductions of failure in all phases with emphasis on later phases, you can communicate system progress and not just a good batch.

This of course relies on buyers to understand and agree with the method. When the group agrees on a simple concept but one member speaks in favor of an improved, but somewhat more complex concept, it could be rejected and that one member can feel pretty lonely.
 
A

Andrews

Jennifer Kirley said:
This is the same concept that educators are pressing for changes in measuring success over time in improving test scores in students.

1. If you are measuring PPM failures in non-identical products, the measurement alone can't accurately reflect that changes in PPM differed through better process control--it could have been other reasons.
2. If you don't track and make the measurements over a sufficiently long term in the same group of parts (and students) but track in a short time period with different groups, same as #1 you can't reliably be sure the numbers reflect control. This means you would need to count defects by lot, not by time.

3. Cutoff periods for failure should be considered when making these counts. At what point do most failures in these parts' lifecycles occur? Is there a curve of failure, including while in use? And if so, is there definition of how interaction with other parts, or end use by customer contributes to failure?

I hope you can see why I do not like simple metrics like PPM failure rate. It's too easy to miss important data, and it's too easy to assume the method is accurate when in fact it can be woefully misapplied by the other party.

If you can show a curve of percentage of failures over time, IMHO you can show better value because expenses of part failure after placed in service also affect your buyer. If you can show the curve changing to reflect reductions of failure in all phases with emphasis on later phases, you can communicate system progress and not just a good batch.

This of course relies on buyers to understand and agree with the method. When the group agrees on a simple concept but one member speaks in favor of an improved, but somewhat more complex concept, it could be rejected and that one member can feel pretty lonely.

Sorry for the delayed response.

1. We are manufacturing around 300 different parts that fall into 35 different categories or families. Collecting data for each part will be tedious and will not fetch much MHO.Please correct me if I am wrong.

2. As far as we are concerned, usually rejections from customers are identified during their receiving inspection and rarely from the assembly line.Both are being accounted together in our ppm rating.

3. Main reason for tracking this measurable is to assess our performance in meeting customers quality requirement and to take actions immediately.
 
C

csterling

Hi to everyone...

I have being reading all the post and comments regarding the calculation of ppm's.

I'm interested in this subject not only because I consider my self as a "Quality oriented guy" but also because our customer are asking if we are measuring our ppm's.

For now, the question I have is:

1- what is the most used PPM calculation?
a) Customer's PPM (rejected-returned material / total amount shipped)?
b) Internal PPM (total scrapped / total produced)?

2- Is there a list of acceptable PPM levels by industrie? I'm interested in the medical devices levels.

Thanks in advanced for all the information
 

mjoakin

Involved In Discussions
Hi to everyone...

I have being reading all the post and comments regarding the calculation of ppm's.

I'm interested in this subject not only because I consider my self as a "Quality oriented guy" but also because our customer are asking if we are measuring our ppm's.

For now, the question I have is:

1- what is the most used PPM calculation?
a) Customer's PPM (rejected-returned material / total amount shipped)?
b) Internal PPM (total scrapped / total produced)?

2- Is there a list of acceptable PPM levels by industrie? I'm interested in the medical devices levels.

Thanks in advanced for all the information


I see there was no answer to this, and I have one more questions. I need to calculate my internal PPM, but, I have (lets say) 4 different products, ie:

Product A - 1000 units, 2 rejections
Product B - 2500 units, 35 rejections
Product C - 1750 units, 5 rejections
Product D - 1970 units, 0 rejections

Product B will have the greater PPM, but I want to calculate my internal PPM.

If I sum all rejections and divide it by produced units, the company PPM will be affected due to Product B, my internal PPM will be 5,817 ppm, which is unacceptable.

Now, if I have a 5th product, product E, where the units (due to they are small parts or easy parts) are 45,000 with 0 rejections, this will decrease my PPM to 804 ppm.

How can I get an accurate calculation? I don't think is right when you have a problematic product and an easy one, and such variations of produced/rejected units affects the final PPM that way.
 

Bev D

Heretical Statistician
Leader
Super Moderator
your calculations are correct - it is the intent or use of the data that you are struggling with. this is why summary data is so misleading. it usually doesn't answer the actual question being asked.


what will your organization do with this data? once we understand that we can perform the correct analysis to answer that question.
 
G

gstewart

Like financial indicators, PPM is an INDICATOR.
Like financial indicators, it is often misinterpreted as a result.
( Example Return on assets, sell the building and rent it back and its vastly improved ... but seriously!!! ).
In an ideal world an indicator can be used to find and examine potential problems. But in this world we seem to think the indicator itself is more important than the underlying real world facts.

If there is a single logistics error and a wrong package is sent to a customer ,with say a couple of thousand product, the ppm will go through the roof. Does that mean the manufacturing quality is really bad? No. It doesn't even mean the logistics system is bad, it could be the only error in 10 years.
What is the consequence in reality? Usually the cost of return freight. ( though potentially shortages could occur ).

If the overall plant ppm is low but the product supplied to one customer is high, then that customer's satisfaction wont get better because of the other products you don't supply him.

Then there may be other customers who don't bother returning failed parts unless they are high cost items ( not typically Auto customers though ).

I find an itemised ppm of known failure types to be of use during annual management review, so we can track improvement of low incident failures.
 
Top Bottom