Is a not-very-well-kept record better than no record at all?

[Ingeniero1]

We have a situation at one of our product lines where the product output is occasionally tracked for number of rejects or defective parts. Unfortunately, this is not done all the time. There may be weeks when the daily output is closely monitored and the results recorded. Unfortunately, there also are weeks when the output, although still closely monitored, is recorded perhaps for only one or two days. Either way, no rejects ever make it past inspection, so bad product getting out is not an issue.

Other than to attempt to have the results recorded daily, which I honestly do not know at this time how successful I would be, I have two options:

A. Make no mention of these records.
B. State that these results are recorded on a random basis that is sufficient to provide evidence to indicate whether the process is in control.

Your opinions, please?

Thanks!
Alex

[Marc]

By "... not-very-well-kept..." do you mean there are entries missing?

[BadgerMan]

Either way, no rejects ever make it past inspection, so bad product getting out is not an issue.

You seem to be implying that there is a subsequent sorting activity (100% inspection) to screen out the defective product?

sufficient to provide evidence to indicate whether the process is in control.

I guess I would question if the process is truly in control if you are needing to screen out defectives from the lot.

Regardless whether you are recording the results of a monitoring activity (in-process inspection) or the results of a final inspection activity, I think you need to adequately record the results of the inspection activity (8.2.4?). Just my humble opinion based on my interpretation of your comments.

[Govind]

Originally Posted by Ingeniero1

We have a situation at one of our product lines where the product output is occasionally tracked for number of rejects or defective parts. Unfortunately, this is not done all the time. There may be weeks when the daily output is closely monitored and the results recorded. Unfortunately, there also are weeks when the output, although still closely monitored, is recorded perhaps for only one or two days. Either way, no rejects ever make it past inspection, so bad product getting out is not an issue.

Other than to attempt to have the results recorded daily, which I honestly do not know at this time how successful I would be, I have two options:

A. Make no mention of these records.
B. State that these results are recorded on a random basis that is sufficient to provide evidence to indicate whether the process is in control.

Your opinions, please?

Thanks!
Alex

Alex,
Your options A & B may help you to save from a finding during the ISO9001 auditing, but you got to get to the root of this situation.

If the monitoring activity is not done regularly, what is the impact it does to the next process steps?

You may be increasing the processing cost, cycle time by letting a defective product travel until final inspection, which could have been detected at the earlier stages of the process.

If this monitoring at this step is not required, you may as well remove it, since it is a non value added activity.

Another point,
Your option B:
"State that these results are recorded on a random basis that is sufficient to provide evidence to indicate whether the process is in control."

In my opinion this statement should be rephrased as:
State that these results are recorded in a periodic interval from samples selected on a random basis that is sufficient to provide evidence to indicate whether the process is in control.

Going by Control chart theory, merely selecting on a random basis will not indicate that process is in control.

Regards,
Govind.

[DDaenen1]

I wouldn't go for A nor for B. If it is audit related, i would just let it be recorded as a non-conformity and use that as leverage towards the management to get the recording installed 100%. I've used this approach several times in the past, and yes, in some cases i got a hard time from the management how i could let this happen but i swallowed that easily if i got the problem fixed permanently as result of the whole situation.

[Ingeniero1]

Perhpas some clarification is in order - my oversight:

The last step in the realization of our products (control instrumentation) is calibration.

Occasionally, a defective instrument makes it to the calibration step, and cannot be calibrated. When this happens, the unit is sent back to one of the previous operations for repair or replacement of the suspected faulty component. This procedure is followed religiously, and no instrument leaves the calibration shop without proper calibration – no if’s, and's, or but’s. After calibration, they go to a final inspection where everything else is checked.

By the way, a calibration record (card and a copy) always are generated when a unit is calibrated. These are properly kept and there is no issue with these.

The problem is that we do not always keep track of the number of units that have to be sent back for repair or parts replacement, which records also entail a brief description of the problem. As I said earlier, some weeks we may have a daily record of how many had to be sent back and the diagnosis, typically zero to two per day, but other weeks we may have no record of how well, or badly, things were.

These records tell us whether ‘too many’ units are being sent to calibration that shouldn’t, and why. We do not suspect that this is the case, as anytime that we do keep records, they show that the number is very small – as would be expected. Yet, there are voids or time lapses when no records are kept for perhaps two or three days. Could these be a problem?

Alex

[Wes Bucey]

Originally Posted by Ingeniero1

Perhpas some clarification is in order - my oversight:

The last step in the realization of our products (control instrumentation) is calibration.

Occasionally, a defective instrument makes it to the calibration step, and cannot be calibrated. When this happens, the unit is sent back to one of the previous operations for repair or replacement of the suspected faulty component. This procedure is followed religiously, and no instrument leaves the calibration shop without proper calibration – no if’s, and's, or but’s. After calibration, they go to a final inspection where everything else is checked.

By the way, a calibration record (card and a copy) always are generated when a unit is calibrated. These are properly kept and there is no issue with these.

The problem is that we do not always keep track of the number of units that have to be sent back for repair or parts replacement, which records also entail a brief description of the problem. As I said earlier, some weeks we may have a daily record of how many had to be sent back and the diagnosis, typically zero to two per day, but other weeks we may have no record of how well, or badly, things were.

These records tell us whether ‘too many’ units are being sent to calibration that shouldn’t, and why. We do not suspect that this is the case, as anytime that we do keep records, they show that the number is very small – as would be expected. Yet, there are voids or time lapses when no records are kept for perhaps two or three days. Could these be a problem?

Alex

As I see it - the key term in your query is "problem," not "record-keeping."

In terms of a Business Management System, the "problem" is simply lack of information to determine a "root cause" or "common cause" of nonconformities which prevent the product from being calibrated prior to going out the door to a customer. Without such information, it is nearly impossible to implement corrective action or preventive action (in my opinion, "rework" means a failure in design, components, or process - it is NOT corrective action.)

If it were MY SHOP, I would concern myself with what is best for the operation, cure it and then, only then, concern myself with whether my record-keeping was a problem from a viewpoint of auditing a Standard.

There is no question on the table of

• whether nonconforming product can reach the customer,
  only a question of
• whether design, or components, or assembly processes, or some combination are the factor in nonconforming devices reaching the calibration station.

If anything, the "problem" is that you don't know and can't know without adequate records.

[BadgerMan]

Good points Wes.

As others have said or alluded to, it is part of a process that you are missing.

If you don’t generate complete records regarding the nature of nonconformities and the subsequent repair actions (8.3), how can you adequately analyze such data (8.4), determine, the root cause(s), and take preventive or corrective actions (8.5)?