I need a way to measure PPM that will provide a useful tool for benchmarking



Im looking for replies to help our organization define the best way to measure PPM that will provide a useful tool for benchmarking, trend, and management review purposes. Please provide your answer to the following question:

1,000,000 parts are shipped during the period and assume all situations are shipped during the period.

Situation 1: 20,000 are returned because upon incoming inspection by the customer it was determined that 50% of the parts were out of specification due to a undersized diameter (cavity 1 and 2 of a 4 cavity mold). All 20,000 parts are returned cavity 1 and 2 are sorted and scrapped, the remaining 10,000 are re-shipped to the customer.

Situation 2: 20,000 parts are returned from the customer because two different parts were mixed together. All parts are returned, sorted in-house, and then re-shipped to the customer.

Situation 3: 20,000 parts are returned from the customer because the wrong label was on the parts. Upon return, the correct label was affixed and the parts were re-shipped to the customer.

Situation 4: 20,000 parts are shipped to the customer, found to exhibit non-conforming parting line flash but deviated and used as is by the customer. Corrective action is requested.

Situation 5: 20,000 parts are shipped to the customer and found to me manufactured with degregated material. Parts were scrapped at the customer and credit was issued to their account.


Additionally, if anyone knows where reference material may be available, please provide.

Thanks to everyone taking the time to review and respond!

Al Dyer

I see your point as I replied in your email. If this is taken as an ongoing situation and not 5 individual scenarios I believe the total parts shipped as 1,050,000 and the total parts rejected as 100,000 which I believe is a PPM of 95,230.

Any help?


Your PPM would be 20,000. PPM does not consider sort and rework activities. Maybe not fair but hopefully consistant.

I base this comment on working with the Automotive industry which is not always known for consistancy.


Al Dyer
Mngt. Rep.
JAE Tech Inc.
[email protected]

[This message has been edited by Al Dyer (edited 17 October 2000).]

John C

With 70k plus rejects out of 1M shipped, and given the pattern of the fallout, you can stop worrying about ppm alltogether for a while. What you need now is to stop counting and start watching, so as to sort out your process problems. Even when you have done this, keep in mind the old saying; 'You can't fatten a pig by weighing it'.
rgds, John C


Al, you said,
Originally posted by Al Dyer:
Your PPM would be 20,000. PPM does not consider sort and rework activities.

By rework, do you mean GSHENK's situation number 5, where the customer gave a deviation and used 'as is'?

I don't think that situation 3 should be considered rework. The label may be an integral part of the product.

For instance, in my line, a product without a UL label is automatically rejected, by our customer, local inspectors, etc. Without the label the liability of the product shoots though the roof, and many insurance companies won't pay claims on fires caused by unlabelled products.

We can't just send a label out to be affixed without also sending a UL inspector. The part has to be returned as a reject.

Secondly, situation 1 resulted in 10,000 parts scrapped and situation 5 resulted in 20,000 parts scrapped, for a total of 30,000 out of a lot of 1,000,000. Wouldn't the ppm, ignoring situation 3, be 30,000?

(Haven't had to deal with ppm myself, so put up with my questions.)



You shipped one million parts to your customer, he returned 100,000 defective parts. From your customer's viewpoint this is 100,000 PPM defective. Which is probably not acceptable.
Whatever you do internally to color the results will not change the view of the customer.
In my dealings with DaimlerChrysler, the cutoff point for an acceptable supplier (quality) is 50PPM.

Al Dyer


This is a difficult situation to follow considering the variables.

1:Through all 5 situations how many total parts have been shipped to the customer.

2: If I read the situation correctly there were 100,000 parts shipped with some type of nonconformance.

3: In situations 4 and 5 are the 40,000 parts shipped on top of the original 1,000,000 shipped? By re-shipping the nonconforming parts we are adding to the original shipment of 1,000,000.

No matter what type of nonconformance or deviation, the 100,000 nonconforming parts are added to the overall PPM.

Maybe GSCHENK could clear it up. We need to know the total number of product shipped (from shippers?) versus the total number of nonconformances reported by the customer (C/A Reports).

I do agree that if PPM's are where they are there is a systemic problem that needs to be addressed.



For the information of everyone, the situations I presented are examples in an attempt to define a protocol for measuring.

The total parts shipped during the period is 1,000,000 plus the re-shipment of 10,000 from situation 1, 20,000 from situation 2, and 20,000 from shipment 3.

Thanks for the info thus far!


In response to an E-mail to GSCHENK:

Yes, I would still consider situation #4, the parts were identified by the customer as nonconforming. Even in situation #1 20000 returned 10000 defective, from the viewpoint of the customer it was 20000 defective.
Yes, I guessed that you were providing examples of different situations that occur from time to time, that's why I did not elaborate on the defect rate.
I do not know of any examples that provide direction on the use of PPM. I do know that if it is not used properly it can create havoc, both internally and with your customer.
Daimlerchrysler rates their supplier's using the PPM criteria; Max 50PPM for an acceptable rating. As an example we recently hit 83PPM, 2 parts defective out of 24000. The only way to improve is to ship additional quantities with no defects. We will not make it with only two months left in the year.
Improvement of PPM is based entirely on quantity. If you ship small quantities it will hurt you, if you ship large quantities it will either help you or hurt you depending on the quantity of defective product.
IMO, be careful when setting rules for PPM, get buy-in from others in your company. Establish goals to show improvement,i.e, if your present PPM is 2700PPM set a goal 1350 PPM to be met within a year. Lower the goal on successive years.
Hope this helps.

barb butrym

Quite Involved in Discussions
the nice thing about PPM (or maybe the bad thing?) is that you set the rules...loud and clear.... before you begin to measure. Any other attempt skews the data. You need to identify all opportunities and how to score/handle them up front,,,and yes the process should be under some sort of control before you start..data/scenarios like the above shouldn't happen by the time you have identified the opportunities and how to measure them...to be that savy and have those figures would be a waste.

John C

My response above was a bit hasty. GSHENK has explained that the figures given were only sample figures, not actual ones. I should have seen that and would apologise exept for the fact that I still come to the same conclusion and my answer is virtually the same;

1m parts in total were shipped (various parts)
Shipments include;
20k returned because 10k are total rejects.
20k returned because parts were mixed.
20k returned for wrong label
20k rejected by customer but given a deviation and used.
20k scrapped at customer premises.

One million parts were delivered. If this included some reworked rejects shipped again, it makes no difference to the figure.
The actual ppm depends on your perspective; To the customer’s purchasing and receiving inspection, it is 100k ppm. A defective part is still a defect even if it gets a deviation. Also, an in-spec part which you can’t use either because it is wrongly identified, unreliably identified (mixed batch) or returned to supplier as part of a reject batch, is still an unusable (defective) part. The only practical way of looking at it is that of 1m parts delivered, only 900k ppm were useable, which means that 100 ppm were defective in some way.
To each of your own company’s sections, the relevant number is different, eg; it’s 40k dppm to your packaging and labelling group and 40 dppm to your manf. line, but only 20k dppm to your purchasing department. So, to tell any one group that the dppm is 100k is meaningless.
And, if it is meaningless, then you should not do it, because it takes up resources and produces no useful response, only another question.

I could go on and explain what the 100k dppm means for each section, eg; how the 40k dppm in the packaging and labelling section breaks down to 20k dppm for the labelling group, and then I could start to work on the specifics of the labelling department process, using the figures, along with other data, to identify by pareto where we should begin taking action. I could use my tracebility codes to identify the lines, operations, shifts and operators that caused the failures and I could then initiate corrective actions to make sure that everyone did their jobs right in future. - And so on, thoughout the organisation.

But, why start with customer’s reject figures? Why should the customer know more about our process than we do? How is this possible? Is it possible? Is it OK?

No matter how I look at the dppm fall-out at the customer’s premises, I come to the conclusion that I have to improve control of my own process. This conclusion is inevitably followed up, if we look honestly at things based on our experience, by the realisation that, next time, the bomb will strike somewhere else - it will probably be scratches caused in removing the waste plastic, or colour defects due to too-sharp corners in the mould design.

Chasing old problems is a futile process. Collecting data, wholesale, leads to missleading data analysed by people with incomplete information. The way to maximise the quality of your data and the informed level of the people is to send in teams to collect data on a specific, real-time problem.
Regards the shipped lots; pray and forget about them.
And the returned lots; take agressive damage limiting action and try to find ways of re-assuring your customer, but don’t let this deviate you from the real work of getting control of your process.

This cuts across conventional wisdom and is hard to take by people taught that we must collect data to pareto our response, etc. But I am not just presenting a counter theory. I have seen that this does not work and so has everyone reading this; The constant discussion about the nature of the fails, the contradicting conclusions, the meetings and resolutions and arguments that lead only to misguided action and further rejects. Above all, the universal complaint; “We are so busy firefighting that we don’t have time to make any real improvement” (or too busy processing data?).

You have to direct this energy into your own process. With the best will in the world, I can’t see that to use customer returns to provide a useful tool for benchmarking, trend and management review purposes can be more than misdirected effort. Your own line performance data; live, simple and reviewed in real time by people who know the process, along with their direct observations on the need for improvement in committment, training, communication, compliance with documentation, adequacy of documentation, application, equipment, planning, etc, will produce small but relevant improvements in process that can give you the results you want. Then you’ll get no surprises. Or, if you do get suprises, you’ve not done a good job, so go back and do it again, and again, as you should be doing anyway.

But you'll say; 'we can't ignore ppms because our customer uses them to rate us'. But the rule still holds - let him do the counting, while you try to get the best possible out of the process by looking at what is actually happening. As Sam says; You have to be very careful how you use the ppm and, unfortunately for the supplier, the customer may not apply a lot of sense - you've just got to live with it. I had a supplier once, who shipped in hundreds of kits every day. The acceptable level was 500 dpm (not that we could do that but our supplier's were supposed to) Day after day, zero ppm, measured on a daily basis - no problem. Suddenly, one day, a kit with an info sheet missing; A THOUSAND DPM! - maybe even 1500! A truck with 5000 computers on it, $10M, held up for a mangy piece of paper and all hell to pay. Yet the guy's ongoing ppm was next to miraculous and just about exceeded what was possible in the circumstances. Yet that day, he was the devil and that was all they remembered until the next time the bomb fell. It's a tough life. But counting it is still not going to fix it.
Rgds, John C

[This message has been edited by John C (edited 19 October 2000).]

[This message has been edited by John C (edited 19 October 2000).]
Top Bottom