IATF16949 Practical Application (31) | Improvement Priority and Project Selection: 9.1.3.1 Trend Benchmarking Quadrants, Four-Dimensional Scoring Matrix, and Three Financial Benefit Confirmation Criteria

1. Scenario: 47 Improvement Items Listed, Only 12 Numbers Delivered at Year-End

Every first quarter, the management representative of a certain automotive parts company organizes a "collection of improvement opportunities." The quality department sends out notifications, and each department fills out forms. The final list contains 47 items: 18 nonconformities from internal audits, 9 customer complaints, 12 rationalization suggestions, 5 items left over from system version changes, and 3 items verbally mentioned by leadership.

The list is submitted, and no one objects during the meeting. However, after the meeting, no one claims responsibility. Three months later, only 12 items are actually being worked on, and 7 of these 12 are due to recent customer complaints that had to be addressed. At the year-end management review, the finance department flips through the improvement ledger and asks, "These 12 projects claim to have created 3.86 million in benefits, but I can't find a single item in the accounts. Where is the money?"

The silence in the meeting room reveals not a lack of execution, but three more fundamental issues that have never been addressed:

First, improvement opportunities have not been prioritized. The 47 items are listed in departmental order, not by impact. Which item to tackle first depends on which department pushes the hardest or which leader mentions it in the meeting. IATF 16949's 9.1.3.1 focuses precisely on this—prioritization.

Second, prioritization is based on feelings, not data. Statements like "this customer is important" or "this project is easy to push" cannot be questioned in the review meeting and cannot be verified a year later.

Third, the financial benefit criteria were not defined at the project initiation stage. Without locking the baseline at the start, it's impossible to calculate the difference at the end. Without finance's involvement at the initiation stage, they won't recognize the benefits at the completion stage.

These three issues can be fixed with a scoring card, a four-quadrant table, and a three-criteria rule. The following sections will cover "what to prioritize—how to prioritize—who calculates—how to run."

2. Solving "What to Prioritize": Eight Sources, Each Item Must Have a Data Origin

Improvement opportunities are not valuable unless they come with data. Any item that cannot be traced to a data origin is returned and not entered into the scoring pool—this rule can eliminate over 30% of the items on the list.

In the IATF 16949 system, there are eight fixed sources of data that can generate improvement opportunities. The first part of the ledger is the following table, with each source assigned a unique responsible person and update frequency to avoid "whoever thinks of it can propose it."

This table has three uses, and missing any one of them will render it ineffective.

Use 1: Each improvement opportunity must have a "data origin number." For example, "Customer A's delivery item on the scorecard was deducted 3 points in August 2026 (number 1-2026-08-A)," not "delivery performance is poor." Having a number allows tracing back to the original record, questioning in the review meeting, and verifying the accuracy of the judgment a year later.

Use 2: Items from the same source should be merged, not split. If 18 NCs from internal audits are listed individually, the result will be 18 small projects that cannot all be completed. The correct approach is to categorize them by root cause—18 items might all point to "control plan not updated after change," making it one project, not 18.

Use 3: Items from sources 1 and 2 must be forced into the scoring pool and cannot be screened out due to "lack of resources this year." IATF 16949 is clear: in identifying correction or improvement needs, deductions on the customer scorecard should be prioritized. Excluding them from the pool is equivalent to ignoring the clause requirements.

3. Solving "How to Prioritize": 9.1.3.1 Requires Trend Benchmarking Against Target Progress, Not a Popularity Chart

The original intent of 9.1.3.1 can be summed up in one sentence: compare the trends of quality and operational performance with target progress and derive actions to prioritize improvement measures aimed at "enhancing customer satisfaction."

This sentence contains two easily overlooked words.

One is trend. It's not about "what the PPM is this month," but "which direction it has been moving over the past 6 months." A target achievement rate of 100% but a trend of continuous decline for three months is a completely different situation from a 92% achievement rate but a trend of continuous improvement for three months. The former is like warm water, while the latter is like climbing a hill.

The other is customer satisfaction. The ultimate focus is on the customer, not on who internally finds it easier. This means that when two projects have similar cost-benefit ratios, the one directly related to the customer is prioritized—this is a rule given by the clause, not a negotiation result between departments.

Based on these two words, a four-quadrant table can be drawn. The x-axis is the target achievement rate (with 100% of the annual target as the boundary), and the y-axis is the trend slope (calculated using linear regression of the past 6 months' data, with a positive slope indicating improvement).

Determining the trend slope does not require complex tools. Take the monthly data for the past 6 months, fit a straight line using the least squares method, and divide the slope by the target value of the indicator to get the "average monthly improvement rate." Empirical threshold: an average monthly improvement rate within ±0.5% is considered stable, above +0.5% is considered improving, and below −0.5% is considered deteriorating. For indicators with high volatility (such as the number of customer complaints), use the median of the past 6 months to compare the medians of the first and last three months, avoiding bias from a single month's peak.

The true value of this table is to prevent a common mistake by managers: using "achievement" to mask "deterioration." Quadrant III is the area auditors like to question the most—why is there no mention in the management review when the indicator is clearly declining? If there's no answer, it's a double nonconformity in 9.1.3 and 9.3.

4. Scoring Matrix: Four Dimensions, 1/3/5 Scoring Anchor Points, and a Veto Clause

The four-quadrant table addresses "what to manage," while the scoring matrix addresses "which to manage first." Only four dimensions are needed—more makes it too complex, fewer makes it too biased.

Dimensions and Weights:

1/3/5 Scoring Anchor Points Table (scoring is only allowed in 1, 3, 5—no 2 or 4 points allowed, as middle scores are a cover for lack of judgment):

Total Score Calculation: Score = (C×30 + R×25 + E×25 + F×20) ÷ 5, with a maximum score of 100.

Priority Bands:

Veto Clause (must be initiated regardless of score, start within 30 days): Involves safety characteristics or regulatory compliance; the customer has issued a red card or formal corrective action request; a major nonconformity with the same root cause has occurred for the second time. These items do not participate in scoring discussions and are directly triggered—because their cost cannot be expressed by a score.

An Example Calculation. After merging and categorizing 47 opportunities, a company is left with 21 items, each scored as follows (three examples):

• Project A: Customer A's scorecard delivery item has been deducted points for four consecutive months, with a delivery punctuality rate of 91% and a target of 98%. C=5, R=3, E=3, F=5 → (150+75+75+100) ÷ 5 = 80 points, immediate initiation.
• Project B: The energy consumption of the injection molding workshop is 12% higher than similar lines, with an annualized savings potential of 420,000 RMB. C=1, R=1, E=3, F=3 → (30+25+75+60) ÷ 5 = 38 points, do not initiate—note that it is not initiated not because it is not worth doing, but because it does not affect the customer or reduce risk. In a resource-constrained environment, it is a correct abandonment. If resources are more available the following year, it can be reconsidered.
• Project C: Missing poka-yoke at a safety characteristic workstation, with a history of 2 units leaking. C=5, R=5, E=1, F=3 → (150+125+25+60) ÷ 5 = 72 points, falls into the "queue" band—but it triggers a veto, so it is actually treated as an immediate initiation.

The third example best illustrates the necessity of the veto clause: using financial and feasibility metrics to measure safety risks inherently underestimates them. The clause does not allow such underestimation.

5. Why Finance Does Not Recognize the Benefits: Three Benefit Criteria, Three Accounting Rules

The biggest internal friction in improvement projects occurs at the completion stage: the project team claims 3.86 million in benefits, but finance says it's not in the accounts. This is not about lying, but about not categorizing the types of benefits.

These three rules must be written into the project initiation documents, not discussed at the completion stage:

Rule 1: Baseline is locked at initiation and cannot be adjusted later. The baseline is the average of the actual values from the three months prior to initiation, with the data source system and data collection criteria noted. The most common dispute is "which month to use as the baseline"—the project team wants to use the worst month, while finance wants to use the average. Set the rule first, and the dispute is resolved.

Rule 2: Only recognize actual measurements, not projections. "Theoretical savings of 1.2 million RMB per year" has no weight in the completion report. The actual data for three consecutive months after completion must be reviewed, and three interference factors—production volume fluctuations, product structure changes, and raw material price changes—must be excluded. If exclusion is not possible, use a conservative value.

Rule 3: Soft benefits cannot be converted into money. Once "increased customer satisfaction equivalent to 300,000 RMB" is allowed, the entire ledger loses credibility, and finance will not participate in the review meeting the following year. Soft benefits should remain in the KPI column; their value lies in supporting the next year's budget negotiations, not in beautifying the current year's benefit figures.

To support these rules, four benefit confirmation points are set, and missing any one will result in "rebound upon completion":

The T+12 column is the easiest to cut but the most critical to keep. A significant proportion of improvement projects show a rebound in indicators six months after completion—measures depend on someone's supervision, and once that person is reassigned, the improvement disappears. The data in this column is the only basis for judging whether "the project is a real improvement or a false one."

6. Quota Combination: Prevent Firefighting Projects from Dominating the Year

Even with a completely correct prioritization method, a systematic bias can still occur: firefighting projects always come first. They are urgent, have clear time pressures, and are driven by customer pressure, while capability-building projects (knowledge accumulation, poka-yoke design, process simplification) can always be "deferred to the next quarter."

The result is that after three years, the company is still dealing with the same issues, just with different batch numbers.

To combat this bias, use quotas rather than relying on self-discipline. Annual improvement projects are divided into three categories, each with upper and lower limits:

Quotas are set based on the number of projects and the number of person-days invested, both metrics must be considered. Focusing only on the number of projects can lead to a false balance where "10 small projects account for 70% of the quantity but only 10% of the effort."

Quotas are not hard prohibitions but triggers for questioning in the review meeting: if firefighting projects exceed 40% in any quarter, the management review must answer why these fires are breaking out and which front-line defense has failed. This question is often more valuable than doing a few more firefighting projects.

7. How to Run: 30-Minute Monthly Review Meetings and 20-Field Ledger

The mechanism must be operational, and the meetings must be short enough for everyone to attend.

Monthly Improvement Review Meeting, 30 minutes, four fixed segments:

Three rules: green light projects are not reported; items without a data origin do not enter the agenda; the financial BP must be present, and the meeting is postponed if they are absent.

Project Scoring Card (20 fields, one A4 page):

Field 16 is the watershed of this mechanism. If the project owner field is predominantly filled with the quality department, it indicates that the company still views quality as "the quality department's responsibility." The holistic quality approach is: delivery items on the customer scorecard are assigned to planning or logistics, incoming material issues to SQE and procurement, personnel capability to HR, design changes to R&D, and benefit confirmation to finance. The role of the quality department here is to facilitate and verify, not to do everything for everyone.

8. Key Implementation Points and Common Mistakes

Key Implementation Points:

1. Run the first month focusing on two tasks—establishing data origins for the eight sources and merging existing improvement lists by root cause. After these tasks, the list usually shrinks to one-third of its original size.
2. The first round of scoring should not aim for accuracy but for "the same yardstick." If the same project is scored by three people and the score difference is within 10 points, it is considered usable.
3. The veto clause must be established and written into the procedure document at the first review meeting; otherwise, the first safety project will be blocked by "insufficient score."
4. The financial BP must be present at the first review meeting; waiting until completion to ask for their signature will not work.
5. Clear the ledger annually: projects in the observation pool for two consecutive years should either be downgraded and abandoned or have a reason for why they are worth keeping.

Common Mistakes:

1. Treating the improvement list as a report card. A longer list appears more active, but it is actually an unsorted wish list. 9.1.3.1 requires a prioritized list, not a quantity.
2. Using departmental balance instead of priority. Allocating two project slots to each department seems fair but spreads resources across items with a tenfold difference in importance.
3. Focusing only on achievement rates, not trends. Indicators that meet targets but continuously deteriorate are overlooked, which is the most common evidence of management review failure in audits.
4. Having more than six scoring dimensions. Too many dimensions turn weights into bargaining tools; four dimensions with fixed weights allow for year-to-year comparisons.
5. Allowing scores of 2 and 4. Middle scores push the responsibility of judgment back to the scorer, leading everyone to score 3, making the matrix ineffective.
6. Discussing benefit criteria only at completion. Not discussing the baseline at initiation will inevitably lead to disputes at completion; this is not about finance being strict but about a missing process step.
7. Converting soft benefits into money. Once converted, the ledger's credibility is zero, and all benefit figures will be questioned.
8. Eliminating T+12 tracking. Celebrating completion without follow-up, and no one knowing about the rebound six months later, turns improvement into a one-time performance.
9. Not limiting firefighting project quotas. 90% of the year's resources are used to handle issues that have already occurred, leaving no resources for front-end prevention, creating a closed-loop vicious cycle.
10. Assigning all improvement projects to the quality department. The owner field is uniformly filled with the quality department, indicating that improvement is still "the quality department's responsibility," and the cross-functional improvement mechanism required by the clause has not been truly established.

The issue of improvement priority is fundamentally about answering a management question: when resources are always insufficient, in what order do you abandon them? The answer provided by 9.1.3.1 is—first, look at the gap between trends and target progress, then prioritize customer satisfaction. Turning this into a scoring table, a non-negotiable red line, and a finance-signed benefit form, improvement becomes a decision chain that can be traced, questioned, and reviewed, rather than an annual repetition of statements.

Improvement is not about listing items, but about prioritizing and defining criteria

Knowledge code: 2.1.2

Version: v20261009

Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously enhance their quality capabilities.