First-Off Part Passed—Why Does Batch Production Still Drift?

First-off Part Passed—Why Does Batch Production Still Drift?

At 8:00 AM, after the lathe tooling was changed, the first-off part was produced and all inspection results were qualified. The team leader signed off and released the batch. By 11:00 AM, during a patrol inspection, the critical dimension was found to be approaching the upper limit. The operator slightly adjusted the parameter, and by the afternoon, the dimension gradually shifted in the opposite direction. By the time quality finished re-screening the batch, production claimed the first-off part was qualified, quality stated the process had already shown a trend, and equipment said the machine had not triggered any alarms. No one was wrong, but the batch of in-process products became stuck.

This type of issue often leads to misjudgment by the factory owner: if the first-off part is qualified, it seems to indicate that the process, equipment, and personnel are all fine. However, the first-off inspection only proves that a single sample at a specific moment meets the requirements, not that the process remains stable for the following several hours. For factory owners, what truly needs to be managed is how the dimensions, temperature, tooling, materials, and operational adjustments are continuously visible after the first-off inspection, forming a chain of facts from change detection, order acceptance, handling, to closure.

First-Off Passed, Answering Only 'Can We Start?'

The value of first-off inspection is to confirm that the product is ready to start production after changeover, tool change, material change, or parameter setting. It cannot replace process monitoring during batch production. Equipment performance changes as it heats up, tools wear down, material batches change, and measurement methods may also deviate. Any of these changes can cause the process to drift gradually without triggering any obvious alarms.

On the shop floor, the signing off of the first-off inspection is often treated as a point of responsibility transfer. Once quality signs off, production assumes it can proceed without concern. Once production starts, quality assumes that subsequent checks should be handled by the operators. When operators detect a trend, stopping the machine may affect output, and actively reporting may lead to being questioned about why the first-off inspection didn't detect it. The safest approach is often to make minor adjustments and observe first. Taking ownership too early can lead to more errors and greater responsibility; making fewer mistakes is safer. Without clear rules for escalation, transfer, and closure, changes are absorbed by individual experience until the product exceeds specifications, at which point the issue becomes an organizational problem.

The Real Danger Isn't Out-of-Tolerance, But Continuous Small Adjustments

A single defective product is easy to notice, but continuous minor adjustments are often overlooked. Each time the operator makes a small change, it appears normal on its own; only when the sequence of time is connected does the process movement become apparent.

Therefore, at least three types of facts must be recorded after the first-off inspection. First, the change in key quality characteristics over time, not just whether it is pass or fail. Second, the time, reason, original value, and new value of each adjustment. Third, the product range before and after the adjustment. Without these three types of facts, the next shift can only see the last parameter and not understand how the process arrived at this point.

Control limits should not be equated with product specifications. Specifications answer whether a product is acceptable, while process monitoring answers whether the system is stable. Even if dimensions remain within specifications, if they continuously move in the same direction and repeatedly approach the boundary, the shop floor should investigate tools, fixtures, temperature, material, and measurement methods, rather than waiting for the first defective part to appear.

Replace 'Release' with Four-Stage Guarding

The first step is first-off confirmation: clearly define the product, equipment, program, tooling, material batch, and measurement method. The second step is short-cycle confirmation: observe the first few pieces or a small segment of production after startup to check if the process quickly stabilizes. The third step is trend monitoring: set inspection frequency based on risk, retain continuous data and adjustment records. The fourth step is shift handover confirmation: trends that have not been resolved, adjustments made, and pending verification actions must be handed over to the next shift.

These four steps do not mean every piece must be sent to the quality department. Operators can self-inspect within their authorized scope, shift leaders are responsible for confirming trends and shutdown boundaries, quality is responsible for measurement rules and abnormal judgment, and process and equipment handle reasons beyond shop-floor authority. The key is to provide both authority and responsibility: if operators are required to be responsible for trends, they must be told when they can stop, which parameters can be adjusted, and who will verify the adjustments after changes.

Factory Owner Randomly Checks One Drift Chain Weekly

Owners do not need to review all dimension data; they can randomly check one record per week that is close to the boundary or has been adjusted multiple times, to check five things: how long after the first-off was the first re-measurement, when the trend appeared, who made the adjustment, which products were affected, and what was used to confirm stability.

If the answer can only be 'the master adjusted it based on experience' or 'everything later turned out okay,' it indicates the factory is still managing results, not the process. Medium-sized growing factories are most likely to get stuck here: they have first-off, self-inspection, and patrol inspection tables, but they have not connected continuous changes with responsibility. Small factories can start with one critical dimension and one bottleneck machine; large mature factories should prevent having too much system data but dispersing adjustment reasons across equipment, quality, and process modules.

A first-off part that passes is worth releasing, but not worth relaxing. Truly reliable batch production is not about getting the first piece right, but about the shop floor being able to see, catch, and close the issue before the product deviates from specifications when the process starts to change.

I am Huang Chaoxing, inventor of the TreeMES stack-light MES system. Follow me for in-depth insights into the management logic behind frontline stories in manufacturing.