Control Plan Looks Good, but the Shop Floor Still Follows Old Habits? —— Five Steps to Drive Layered Process Audit (LPA) with the Control Plan
A plastic injection molding workshop produces an automotive interior component. The control plan (CP) is well-written: a critical dimension is marked as a special characteristic by the customer, and the control method is clearly stated as "sampling inspection of 1 piece every 2 hours using a vernier caliper, recorded in the 'Process Inspection Record Form'." The reaction plan also specifies "immediate shutdown and trace back to the last known good point if the dimension is out of tolerance." However, during a customer audit, the reality on the shop floor was revealed: the inspector manages two machines alone and hasn't touched the caliper in four hours, yet the record form shows entries every two hours with neat handwriting. When the auditor randomly asked three employees, "What do you do if this dimension is out of tolerance?" they gave three different answers.
A well-written CP and its actual execution on the shop floor are separated by a significant gap. This gap cannot be bridged by additional training or re-emphasizing the requirements; it needs a fixed, top-to-bottom audit rhythm—Layered Process Audit (LPA).
1. Three Breakpoints in CP Implementation
Document Breakpoint: After the CP is finalized, it is filed away, while the work instructions, inspection specifications, and record forms are not updated simultaneously. This results in "three documents with three sets of requirements," leading the shop floor to choose the easiest set to follow.
Cognitive Breakpoint: The language in the CP is designed for auditors, not for operators—sampling every two hours on a production line with a 40-second cycle time is impractical, so the shop floor "relaxes" the requirements while the documents remain unchanged.
Response Breakpoint: The reaction plan only lists the action names without specifying who should do what and within what time frame. After an alarm is triggered, there is no response or record, making the control measures ineffective.
The common issue with these three breakpoints is not that the CP is poorly written, but that no one consistently verifies whether the shop floor is actually following it. LPA aims to address this exact problem.
2. Five Steps to Implementation
Step One: Extract Audit Items from the CP, Not Re-compile a List. Each row in the CP already contains four elements that can be audited: characteristic, control method, frequency and sample size, and record and reaction plan. Convert these elements directly into on-site questions, such as "What measuring tool is used for this dimension, how often is it measured, where are the results recorded, and who stops the machine if it is out of tolerance?" For each process, select three to five items, focusing primarily on special characteristics. The audit list should be derived from the CP to avoid discrepancies between two sets of standards.
Step Two: Define Layers and Frequencies. Operators perform self-inspections at the start of each shift and during the process, auditing their own workstations. Team leaders conduct at least one audit per week, covering their own workstations and the connection with the previous process. Workshop supervisors audit once a month to ensure that cross-process flows and resources are adequately supported. Quality and process engineers audit quarterly to verify the consistency between the CP, work instructions, and inspection specifications. The higher the level, the less the focus is on "whether something was missed," and the more on "whether the system supports the shop floor in following the requirements."
Step Three: Three Observations and Three Questions On-Site. Observe the actions: ask the operator to demonstrate the measurement and recording process and check if the actions align with the documents. Review the records: ensure that the records are up-to-date and the data matches the actual measurements. Trace the process: randomly select a work-in-progress item and verify if the records from the first article inspection, process inspection, and release inspection form a complete evidence chain. Ask three questions: how is this characteristic controlled, under what conditions must the machine be stopped, and who to contact after stopping the machine. If an operator cannot answer these questions, the issue lies in the documents and training, not the individual.
Step Four: Nonconformity Grading and Response. Grade A: missing control measures for special characteristics, inaccurate records, and ineffective poka-yoke—immediate containment and reporting to the highest management on the same day. Grade B: unmet frequency requirements, missing records—correction and traceability of products within 24 hours. Grade C: unclear labels, outdated versions—incorporate into weekly improvements. Each grade should specify who does what within how many hours and to whom it should be escalated. The authority to stop the line should at least be given to the team leader.
Step Five: Close the Loop and Feed Back to the CP. Summarize nonconformities quarterly and perform a Pareto analysis: if the same issue recurs more than three times, determine whether it is due to "non-compliance by personnel" or "unfeasibility of the CP." If the latter, revise the CP and synchronize the work instructions, inspection specifications, and record forms through the formal change process. Any changes requiring customer approval must be submitted. After three months, the recurrence of nonconformities in this process was zero.
There is a non-negotiable boundary: whether the frequency can be relaxed is determined by data and customer approval, not by the shop floor. Before approval, the only correct approach is to follow the requirements or adjust resources, not to falsify records to make it appear as if the requirements are being followed.
3. An Example
The handling approach of the aforementioned plastic injection molding factory can be replicated: when LPA discovers that "records show measurements every two hours, but the shop floor hasn't measured in four hours," the first step is containment—trace and perform enhanced sampling inspections of the products from both shifts on the same day to confirm no out-of-tolerance dimensions. The second step is to determine the root cause, which in this case is insufficient inspection personnel and the potential for online automatic measurement of the dimension. The third step is a short-term measure—switch to online automatic measurement with first article confirmation, and have the equipment automatically retain the inspection records. The fourth step is a medium-term measure—calculate the Cpk of the dimension using data accumulated over three months, which is 1.55, and submit a control plan change request to the customer to officially adjust the frequency to one piece every four hours plus online monitoring. After approval, update the work instructions, inspection specifications, and record forms. Three months later, the recurrence of nonconformities in this process was zero.
Here is a boundary that cannot be compromised: whether the frequency can be relaxed is determined by data and customer approval, not by the shop floor. Before approval, the only correct approach is to follow the requirements or adjust resources, not to falsify records to make it appear as if the requirements are being followed.
4. Three Misconceptions
Misconception One: Treating LPA as a Record-Checking Exercise. Only checking "whether the documents exist" and not "whether the actions are performed and the data is accurate" can mask the most dangerous phenomenon of untruthful records.
Misconception Two: Assigning LPA Only to the Quality Department. The main executors of the CP are manufacturing and engineering. Without the involvement of team leaders and supervisors, LPA degenerates into another round of sampling inspections, and the shop floor will treat audits as surprise checks to be handled superficially.
Misconception Three: Only Issuing Nonconformities Without Revising the CP. Repeated issues either indicate a disciplinary problem or that the CP itself is unfeasible. The former requires personnel management, while the latter necessitates revising the documents and following the change process. Otherwise, the more nonconformities LPA issues, the more numb the shop floor becomes.
5. In a Nutshell
The value of a control plan lies not in how complete it is written, but in whether it is proven to be followed every day on the shop floor. Layered Process Audit (LPA) is the fixed rhythm that pins the paper-based control plan to the shop floor.
The CP is validated, not written—use a layered, fixed rhythm to turn paper-based requirements into on-site actions.
Knowledge code: 8.3.2
Version: v20260927
Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously improve their quality capabilities.