That Day, The Customer's Acceptance Report Solidified the Air in the Meeting Room
I remember a few years ago, we shipped a crucial batch of products for a major European manufacturer. That batch went green light all the way from production to FQC (Final Quality Control), making us think it was a sure thing and we could finally get some good sleep. However, on the day the customer came for on-site acceptance, the QA Leader suddenly rushed into my office, pale-faced, saying, "We're doomed! The customer's FQC inspection results are different from ours; they say our outgoing standard is too lenient!" My heart sank then; I thought, how could that be? Our design and process all conformed to their specifications. But when the customer's report came out, their judgment of "appearance defects" in the OQC (Outgoing Quality Control) stage was miles apart from ours. In the meeting room, everyone's gaze was frozen.
Where's the Problem? Are Your FQC and the Customer's OQC the Same Standard?
To put it simply, often it's not that your product is poorly made, but that your definition of "good" differs from the customer's. Your company's FQC (Final Quality Control) might be set based on internal process capabilities and empirical values, deeming a Cpk of 1.08 quite good already. However, the customer's OQC (Outgoing Quality Control) might directly benchmark against the end application, requiring at least a Cpk of 1.33 to pass. Such a discrepancy can, at best, lead to returns and financial losses, and at worst, result in losing the customer entirely. This is no laughing matter.
So the key is that your FQC/OQC must be "aligned" with the customer's acceptance standards. This is not just about numerical agreement; more importantly, it's about a consistent understanding of "defects." You might think a scratch less than 0.5mm in length is acceptable, but the customer might consider any scratch unacceptable. This difference in perception is the root of all problems.
How to Actually Do It? Data Alignment is Key
Frankly, to avoid this frustrating, repetitive situation, the most effective method is "data alignment."
- Thorough Early Communication: In the early stages of product development, all acceptance standards must be finalized with the customer. This means not just providing them with a specification sheet, but presenting specific acceptance methods and judgment criteria for each critical inspection item, such as appearance, electrical properties, and mechanical characteristics. For example, if the customer's requirement for DPMO (Defects Per Million Opportunities) is 6210, then your FQC target value cannot be set at just 10000.
- Standardization of Samples: For items with more subjective judgments, such as appearance defects, it is essential to create "limit samples." That is, ask the customer to personally approve or provide specific samples, indicating which defects are acceptable and which are not. This is far more effective than describing it in words a thousand times. If you have a set of "customer-approved" NG (Non-Good/Rejected) samples, your inspection personnel will not be talking past each other with the customer.
- Regular Cross-Verification: Once cooperation is stable, samples should also be regularly sent to the customer for secondary verification. If their inspection results differ from yours, immediately stop, identify the discrepancies, and correct them. This is somewhat like calibration, ensuring both sides' measuring tools are the same length.
Therefore, the key is that you must internalize the customer's "acceptance standards" into your "outgoing standards."
The Most Common Pitfall: Reliance on Experience Can Be Your Downfall
The biggest pitfall I've encountered is our engineers relying too much on "experience values" (empirical data/past experience). Thinking, "This customer always inspected it this way before, so it should be the same this time?" or "This defect used to be released, so it should be fine this time." The result usually ends in trouble.
One time, we changed our production line supervisor, who insisted that our past outgoing specifications for a certain electrical parameter were too strict, and loosened the standard by 0.5% in internal FQC. Consequently, the customer's automated testing machine directly flagged it as NG, leading to a surge in returns. We then discovered that the customer's acceptance standards were hard-coded into the machine's program and would not change based on our company's empirical values. This kind of situation, where internal "presumption" leads to detachment from customer standards, is truly common.
To put it plainly, no matter how rich your experience, it cannot outweigh the customer's acceptance standards written down in black and white.
One Thing You Can Do Today
Find your product's "customer acceptance standards," compare them against your FQC/OQC standards, and check for any blind spots.