Quality Management Depth (17) | Designing Quality Performance Metrics: Avoiding "Buck-Passing" Indicators

1. Introduction: A Quality Manager Overwhelmed by Metrics

A certain automotive parts company, with an annual output value of 680 million yuan and approximately 900 employees, implemented a "comprehensive quality assessment" at the beginning of 2025. The metrics of customer complaint rate, internal scrap rate, and first-time inspection pass rate were broken down by quarter and assigned to various departments, with the Quality Department having an overall weight of 40%. The quality manager's personal performance was 60% determined by these three metrics.

In the first quarter after the implementation, the quality manager's performance coefficient was 0.72, resulting in a loss of over 2,000 yuan compared to the previous quarter. The reason for the deduction, as stated on the HR assessment form, was a 0.3 percentage point increase in the customer complaint rate. Tracing back these three new complaints, they all stemmed from a single design change—without notifying the process department for verification or the quality department for inspection tool confirmation, production proceeded directly according to the new drawings, leading to dimensional discrepancies that reached the customer. The sign-off section on the change order was empty, with no signatures from the Quality Department.

In the second quarter, another incident occurred. To reduce the internal scrap rate from 1.8% to 1.2%, the workshop applied for a concession release for a batch of parts with minor appearance defects, which the Quality Department signed off on, citing "high historical acceptance by the customer." Three months later, this batch required a mass rework at the customer's site, incurring direct costs of 470,000 yuan and a quality alert from the customer.

In the third quarter, the quality manager did something that surprised the boss: he submitted a revision application, requesting to reduce his department's weight from 40% to 25% and increase the weights for R&D, process, and procurement. The boss's first reaction was, "You are shirking responsibility." The quality manager responded, "I am not shirking responsibility; I am placing it where it belongs. The current calculation only penalizes the department that discovers the problem, not the department that creates it."

2. Judgment Framework: Three Common Misjudgments

Misjudgment One: Directly Assigning Result Metrics to the Weakest Responsible Party. Customer complaint rate, scrap rate, and pass rate are all "results"—the product of multiple stages. Assigning them to the Quality Department, which is at the end of the chain and has the least influence, is essentially making the "pursuable person" take the place of the "responsible person." This design will lead to two predictable outcomes: the Quality Department starts managing for metrics, focusing on data interpretation and release signatures rather than the manufacturing process; and R&D, process, and procurement remain unbound by assessments, so areas like change management, error-proofing design, and incoming quality control never receive the necessary investment. The quickest way to test whether a quality metric is a "buck-passing" metric is to ask: does the root cause of the metric's deterioration primarily occur within the department's control? If the answer is no, it is a buck-passing metric.

Misjudgment Two: Absence of Process Metrics Leads to Uncontrolled Result Metrics. Companies that only assess results will inevitably see two things happen simultaneously: short-term numbers look good, but long-term numbers are worse. Result metrics have a lag effect; process improvements often take two to three quarters to show results, while short-term tactics to reduce results (concession releases, reduced sampling, selective batches) have immediate effects. Therefore, result metrics must be balanced with process metrics, and process metrics must clearly define "what counts as done": 100% completion of change order sign-offs, monthly verification of error-proofing devices, complete first article inspection records, and sampling of incoming key characteristics according to a stricter plan. Result metrics guide direction, while process metrics guide the path. A half-missing assessment system forces people to take shortcuts.

Misjudgment Three: Unbounded Joint Responsibility Degenerates into "Everyone's Responsibility Equals No One's Responsibility." Many companies, after learning from their mistakes, swing to the other extreme—every department shares all metrics, and the entire company is penalized for customer complaints. This design appears fair but actually dilutes responsibility: the truly responsible departments are averaged out because "everyone shares the responsibility," and no one feels the pain. The key to designing joint responsibility is not "everyone's responsibility," but three clear points: who is the primary responsible party, who is the supporting party, what is the upper limit of the supporting party's penalty, and what is the objective evidence for the penalty. Joint responsibility without an upper limit ultimately becomes an emotional account for the entire company rather than a management account.

These three misjudgments share the same fundamental issue: the assessment scheme is a mechanism for allocating responsibilities and rights, not a numbers game. It determines "who in the organization feels the pain that aligns with the root cause of the problem." When aligned, the assessment automatically drives improvement; when misaligned, the assessment automatically leads to data beautification and responsibility shirking.

3. Implementation Actions: Five Practical Steps

Action One: Map Each Quality Metric to Its Responsible Party. Responsible Party: Led by the quality manager, confirmed by HR and heads of various business departments. Method: Break down each quality metric in the current assessment form, clearly stating "when the metric deteriorates, the root cause usually occurs in which stage" and "which department is responsible for that stage," creating a mapping table that lists primary, supporting, and exempt roles. Criteria: There should be no metric in the table where the primary responsible department is the Quality Department but the root cause primarily occurs in another department; each metric should have at least one non-quality department's supporting responsible person sign off.

Action Two: Balance Result Metrics with Process Metrics. Responsible Party: Designed by the quality manager, approved by the boss. Method: The weight of result metrics for the same assessment object should not exceed 30%, with the remainder covered by process metrics; process metrics should be verifiable from existing records, with one metric corresponding to one evidence source. Criteria: Each result metric should have at least two process metrics that can be verified within 10 minutes—metrics that cannot be verified should not be included in the assessment form.

Action Three: Define Boundaries and Upper Limits for Joint Responsibility. Responsible Party: Proposed by the quality manager, implemented by HR. Method: Set an absolute upper limit for the penalty of supporting responsibility (e.g., no more than 5% of the department's total performance score) and require that penalties be accompanied by objective evidence (change order number, review record, inspection data), not based on subjective impressions. Criteria: Each item of supporting responsibility penalty in the assessment form can be traced to specific evidence; any penalty without evidence in the current period should be revoked.

Action Four: Define Metrics Before Assessment, and Announce Any Changes One Assessment Cycle in Advance. Responsible Party: Quality manager and data management department. Method: Write the statistical scope, data source system, data collection time, and exclusion rules for each metric into a written document and include it in the system document management; if the scope needs to be adjusted, it must be announced and the historical baseline revised before the start of the new assessment cycle. Criteria: The current period data for any metric should be independently calculated by two people with consistent results; there should be no mid-period changes to the scope. The direct cause of the "scrap rate" metric going out of control at the company was that concession releases were not excluded from the scrap rate or listed as a process metric, leaving a gap that was naturally filled by human nature.

Action Five: Connect Assessment Results to Improvement Mechanisms, Not Just Paychecks. Responsible Party: Quality manager and heads of various departments. Method: Clearly state that "not meeting the target does not necessarily result in a penalty, but it will trigger a root cause analysis and improvement commitment," with the closure of improvement items counted in the next cycle; for departments that fail to meet the target for two consecutive cycles due to the same root cause, escalate to management review. Criteria: Each unmet metric should have a corresponding improvement item number, responsible person, and completion date; the closure rate of improvement items should be included in the department's quarterly evaluation.

4. Case Development: What Happened to the Revision Application

The quality manager did not just make demands; he first conducted a data review, reclassifying customer complaints over the past four quarters by root cause stage: 38% were due to change management, 21% to incoming materials, 24% to missing process parameters and error-proofing, and only 17% were due to inspection oversights by the Quality Department. This classification brought all controversies to the table at once—it did not target any department but focused on "where the problems come from."

The boss understood the chart and agreed to the proposed adjustments: the Quality Department's weight was reduced to 25%, with R&D, process, and procurement taking on 30%, 30%, and 15% respectively for quality-related metrics. The three process metrics—"change order sign-off completion rate," "monthly verification completion rate of error-proofing devices," and "execution rate of stricter sampling for key incoming materials"—were also added to the assessment form, each with a cross-departmental responsible person.

The costs were real. First, the heads of R&D and process had clear resistance in the first two quarters, with the R&D director stating at a management meeting that "quality metrics should not be in my assessment form." It took the boss two meetings to set the tone and push through the changes, during which the quality manager faced significant lateral pressure. Second, the data collection for process metrics was manual for the first three months, with one quality engineer spending about 6 hours per month until the MES reports were configured for automation. Third, in the first quarter after the revision, the company's overall customer complaint rate did not immediately decrease; instead, it appeared to increase due to more detailed data classification. This led to the quality manager being questioned in the second quarter, but he held his ground with the root cause classification chart.

The results became evident in the fourth quarter: complaints due to unassessed changes in dimensions dropped from 9 per quarter to 3, incoming material complaints dropped from 5 to 2, and the number of concession releases decreased from 14 batches per quarter to 4, each with a written customer confirmation. The annual COPQ loss decreased by about 1.8 million yuan. The quality manager later said bluntly in an internal review, "After the assessment was corrected, I no longer had to chase people for improvements; they came to me to discuss how to improve."

5. Self-Check List

• In the current assessment form, are there any metrics where the root cause primarily occurs in another department but the Quality Department is the primary responsible party? How many?
• For each result-based quality metric, are there at least two process metrics that can be verified within 10 minutes?
• Is there a clear upper limit for the penalty of supporting responsibility, and can each penalty be traced to objective evidence?
• Are the statistical scopes of metrics documented, independently verifiable by two people, and are there any mid-period changes to the scope during the assessment cycle?
• Does each unmet metric necessarily trigger an improvement item with a responsible person and completion date, and is the closure rate of improvement items included in the department's evaluation?

The pain of metrics should align with the root cause of the problem.

Knowledge code: 13.3.3

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.