Does every Nonconformance Require a Corrective Action?

[KenS]

As part of the implementation of a Quality System in a multi million dollar company that has none, I set up a database on the Intranet to record in process discrepancies. The database is accessable by all employees and is configured to accept input from any point in the manufacturing process, from material receipt to product shipment. The operator has the ability to check a box on the input form to automatically generate a Corrective Action for review/distribution by QA. An example of an entry may be a jumper missing when trouble shooting an electrical rack. The technician will want a CAR if this is something he/she has seen before. Periodically I use Excel to review the data and identify additional entries needing CARs. In addition, this review is used to identify areas requiring training of personnel. As a result we have generated over fifty CARs and SCARs in the few months the system has been in operation.

The problem is the amount of entries, currently 497. The VP Quality (Corporate) questioned why all entries do not have a corrective action. I explained that the system was not only a Corrective Action system, but was gathering data for continuous improvement in many areas. I thought we had cleared this, but he questioned it again today, as "an ISO requirement", the result of a procedure I wrote for using the database.

My question - where does it say every discrepancy must have a written/documented corrective action?

[Randy Stewart]

It doesn't Ken. It is up to you/your system to determine when it is appropriate. If you documented it in your procedure that all will have corrective action then there may be a problem. I would question the system if the amount of entries haven't generated any CARs.
IMO I would be looking at how the entries are addressed by QA etc. When they are reviewed what criteria generates the CA, is the established criteria consistently used and understood, etc. If CA isn't required is there any feedback to the operator?
It would seem you have buy in from the operators, and that isn't always easy.

[Claes Gefvenberg]

First of all, that seems like a good way to get a grip on things. I'd say you have a very potent tool there. It's obvious that you have your sights set for continual improvement.

My question - where does it say every discrepancy must have a written/documented corrective action?

Interesting... Let's reason a bit:

So let's say a missing jumper is detected. I don't think I'm far off in guessing that some of these things are quite common, and are detected in the normal inspection in your process? That would mean that the situation is under control and just maybe it doesn't warrant a CAR, but rather continual improvement as you point out yourself. If you suddely detect a dramatic increase in some area... Now that's a different story: That should initiate a CAR, and that CAR could adress a number of entries...

"requiring training of personnel" That could be a very suitable corrective action, too...

The definition of a Corrective action is that it shall eliminate the cause of a detected nonconformity, right? And the definition of a nonconformity is that a requirements is not fulfilled. Then we come to your procedure. What does it say?

/Claes

[KenS]

I'm partially venting with this topic, and partially looking for input. My greatest fear is that the system will be overloaded with CARs which will defeat the value of the responses, or the techs will not see a clear result and will back off. I'm trying to find a balance and ammunition to sell it. 497 CARs in 4 months is to my mind ridiculous.

Initial operator buy in was not a problem, they were tired of complaining about the same things over and over. Once they saw one problem addressed and corrected they bought in big time.

The system collects a wide range of information on each discrepancy (drop down lists make it easy). The preliminary procedure lists criteria for a CAR, including things like time lost, repetition, input by the tech and, during the review, cost similarity to other instances.

As an example of the area of disagreement consider an electrical rack with over 1000 connecting wires. One is missing. The tech spent less than 15 minutes finding and repairing it. No damage was caused by the missing wire. On nine other tools this was not a problem. I do not consider this a candidate for a CAR, and it does not meet the criteria in the preliminary procedure. This is actually the first of several discrepancies brought to my attention by the powers above, and the first we disagreed on.

Incidentally, the procedure addresses discrepancies that are similar in nature (i.e. different jumpers missing in each of several racks) which get identified during review and addressed with a kind of global CAR, workmanship oriented.

[Claes Gefvenberg]

I'm partially venting with this topic, and partially looking for input. My greatest fear is that the system will be overloaded with CARs which will defeat the value of the responses, or the techs will not see a clear result and will back off. I'm trying to find a balance and ammunition to sell it. 497 CARs in 4 months is to my mind ridiculous.

Yes it is. I think you have come up with a winner tool, as long as you can keep your number of CAR's on a reasonable level.

Good luck, and do let us know how it's going.

/Claes