I for one, don't accept a "normal" or "standard" defect rate. Clearly this supplier doesn't care about the quality they provide.

Anyone remember the number of plane crashes if the air traffic controller was only 99.9% right? wasn't it about 526 plane crashes per day? Something like that.

I think it was about 1,000 pcs of mail got lost every day if the post office only delivered 99.9% of the mail.

Expect 100% good quality every time.

Very good! thanks for the input. You may add one to the list of "what if" I heard at Quality seminar by Phil Crosby (many years ago at that time promoting "0" defects)i.e. The defect rate of 99.9% for a pediatrician regarding dropping baby's post delivery. The problem is with "who cares" obviously there is no question about some one carring about droping infants - apparently my vendor does not see his hardware as improtant. Me thinks tis time to drop this vendor.

Hey better yet, ask your supplier if it OK for you to pay only 97% of his invoice? I mean what's 3% between friends. right?

Here's some inputs: Start out with Acceptance Samping, initially starting out with ANSI Z1.4 for attribute sampling (go / no-go). Then with good tracking and decreace in defects, switch to ANSI Z1.9 for variable sampling so that you can get more precision on just how good or bad the products are.

But at the same time, the vendors themselves must be held to good and standardized sampling practices and also to their own corrective actions based on what their sampling rates tell them. The vendors need to eventually get into statistical process control (SPC) and other statistical tools to understand their processes and the effects on product quality. They need to keep records of their samples and the defect rates they find. You should also be reviewing how they do their sampling (checking for any possible autocorrelation of samples), what their corrective actions are, and how they are notifying you of bad defects and also how they are making improvements. They need to be doing a lot of the defect control, because whatever you have to do is really a verification of what they were supposed to be doing all along.

Otherwise you will be left in the dark to "trust" their verbal words to you.

Catherine, ASQ CQE, MSQA

Thanks for information and input Catherine. I have set and measured to an AQL of 0.65 level II. A target I extrapolated back from my AOQL. The supplier is simply attempting to point out that the AQL I selected is too ambitious for this industry. He is aware I am representing a high end (niche) market that expects a higher value. The supplier does a terrific job of fire fighting each assignable cause typically through a process change or product improvement but is reluctant to institute any long term plans for continuous improvement targeting a goal I would find acceptable. E.g. the supplier states 3% is the defect rate they have been willing to live with simply because that is what it is (In their perspective). I had it in mind before I started partnering with this supplier to establish that there were other suppliers in the same industry that have experienced better. In lieu of going through the trouble of a survey I though I would just take a shot at his forum for some specific data regarding "normal defect rates" on items I mentioned in my original note. I have no problem with the input so far an appreciate each response.

Thanks

Tom

Thanks for information and input Catherine. I have set and measured to an AQL of 0.65 level II. A target I extrapolated back from my AOQL. The supplier is simply attempting to point out that the AQL I selected is too ambitious for this industry. He is aware I am representing a high end (niche) market that expects a higher value. The supplier does a terrific job of fire fighting each assignable cause typically through a process change or product improvement but is reluctant to institute any long term plans for continuous improvement targeting a goal I would find acceptable. E.g. the supplier states 3% is the defect rate they have been willing to live with simply because that is what it is (In their perspective). I had it in mind before I started partnering with this supplier to establish that there were other suppliers in the same industry that have experienced better. In lieu of going through the trouble of a survey I though I would just take a shot at his forum for some specific data regarding "normal defect rates" on items I mentioned in my original note. I have no problem with the input so far an appreciate each response.

Thanks

Tom

This comes across as a vendor who is used to customers just accepting what is given to them. Anyone in modern quality, such as ISO, has to work at not just fire-fighting, but also preventive actions and root cause analysis (RCA) and other investigative tools.

You should create "penalty points" in the contract / orders that you would use to deduct portions of payment for each percentage or half-percent that they are over the target.

Excellent advice. I will bring this idea to our Purchasing Agents. I suppose the penalty clause should be put on the PO boiler plate. In my environment For Hardware suppliers there is no other written contract other than the P.O.

Thanks

"100% on Time and 0 defects is what I get from ordinary suppliers."

-What my customers tell me.

WHat I work to.

probably what you work to.

WHy should he get a break at 3%?

milo

Can't really say what is the accept rate in this business. All my prior experience was in Electronics, Telephony, and Aerospace/military. This is my first time in commercial/consumer products in the Quality role. So far I have the support of the company owners - albeit there has been no increase in the cost of the hardware.

I have stated my expectations to the supplier of less than 0.5% and will continue to hold them to it. Most of this hardware is made in China - so the quality level I'm looking for will require the supplier to sort at his depot and live with the long turn around regarding C.A.

So you and your boss think that its ok to not meet the terms of the agreement .5% of the time?

You should open a church. I'll join and pick the commandments I want to "honor."

Seriously, your .5 % is too generous for variables data- heres why.

A process operating under statistical control, within +/- 3 standard deviations is represented by the area under the normal curve. lets take the case where the process is just capable of meeting the requirements.

That area, for +/- 3 sigma is 99.72%. so .28% is expected outside the +/- 3 sigma area. You are giving them almost double the permission to err of a statistically controlled process by using .5% instead of .28%. Thats a great way to spoil children, and create welfare entitlement thinking!

Add some meat to your diet.

Contracts spell out requirements. signatures indicate acceptance.

Any wiggle room makes the contract useless.

Either they can deliver using in control processes, or someone in purchasing is so enamored of cheap price that they are looking the other way.

In the steel business, we call those people "WHORES."

milo

Not saying 0.5% is great - Just 500% better then what the supplier was proposing. Also, I'm looking at attribute as opposed to variables.

By the by - since your in the steel business what do you think the stack up tolerance would be on a item that contains 8 pieces of SS punched out and assembled? Each piece should be about 0.056 " thick. I have no control over the assembly being made from the same metal sheet. (were talking China and 2nd tier distribution)

Attribute is more difficult- Its go / no go. So you have to clearly have standards of Pass/ fail. Undisputable standards. Unambiguous.

I worked with a company that made food slicers. bought blades from china. Stainless. slivers on the finished blades. That would have come off in use and ended up in someones salad....

Eventually we found a US company that was competitive and had NO quality issues.

Subsequent analysis showed some samples from Chinese supplier were not even proper grade.

Can't help you on the stack up analysis. Just on what the tolerance ought to be for material given the agency standard. BUT, My 2 cents is that whatever those stackups are, the standard deviation of a sample of thirty of them ought to be one sixth or less of the total variation in the sample of thirty. Otherwise, its a crappy process.

For what its worth.

milo