Inspection Stations Keep Increasing, but Customer Complaints Don’t Decrease — A Five-Step Method for Calculating the Economic Viability of Inspection Points

A hardware factory has set up five inspection stations after seven processes, with six inspectors, and averages twelve customer complaints per month. The quality manager's judgment is that the inspections are "not detailed enough," so another verification step was added before shipment. After three months, the number of complaints did not decrease, and the additional inspection labor cost reached 18,000 yuan. The real issue is not whether the inspections are detailed enough, but whether the inspection points are correctly positioned: three out of five stations intercept processes with extremely low defect rates and minimal downstream losses; yet there is no inspection station after the only irreversible welding process.

The number of inspection points and their locations can be calculated. The core principle is: The cost savings from preventing defects from flowing downstream must exceed the cost of setting up the inspection point. Below, we follow a five-step method, with a practical example included.

1. Calculate Four Key Numbers

The decision to set up an inspection point depends on four numbers; missing any one of them turns the decision into a guess.

• c: Cost per unit inspection. The inspector's salary is calculated based on effective working hours, plus the depreciation and maintenance costs of measuring tools. For a monthly salary of 7000 yuan, 21.75 working days, and 8 hours per day, the labor cost is approximately 0.67 yuan per minute; inspecting one piece for 45 seconds costs about 0.5 yuan, plus 0.1 yuan for tool depreciation, so c ≈ 0.6 yuan per piece. If inspection becomes a bottleneck, the production line's waiting time and rhythm loss must also be considered, often doubling c—this is the easiest item to underestimate.
• p: Defect rate at this point. Use the actual defect rate of the process over the past three months, not the overall factory yield rate. If there is no data, conduct a short-term statistical analysis by sampling 200 pieces from the process, which is more reliable than estimating based on impressions.
• F: Cost of downstream flow. The cost per unit of handling defects once they reach the customer, including return shipping, rework or replacement, complaint handling labor, claims and penalties, expedited shipping, and the prorated cost of lost orders. For automotive parts, export parts, and safety-critical components, this cost is often in the hundreds of yuan.
• Fi: Internal handling cost. The cost of rework or scrap when defects are intercepted at this process or downstream, typically around one-tenth of F.
• d: Detection rate. Inspection is not 100% effective. Estimate d using a blind sample challenge test: introduce 50 known defective pieces and see how many are intercepted by the inspector. The detection rate for manual visual inspection is usually between 0.7 and 0.9, while automatic inspection tools can achieve 0.98 or higher. Without d data, all subsequent calculations will be inaccurate.

2. Criterion: One Inequality

Without an inspection point, the expected loss per unit is p × F; with an inspection point, it is c + p × (1−d) × F + p × d × Fi. The inspection point is only worthwhile if the former is greater than the latter, which simplifies to:

p × d × (F − Fi) > c

Let the ratio of F to c be k (how many times the downstream flow cost is the inspection cost), and ignore Fi to get a very simple threshold: The defect rate p must be higher than 1/(d × k) for the inspection point to be worthwhile. The threshold calculated for d = 90% is as follows:

This table explains a common phenomenon: setting up a full inspection station at the end for low-value standard parts often results in a net loss; whereas for a 500-yuan actuator, a defect rate of just over 0.2% makes the end inspection point profitable. To decide whether to inspect, first look at k, then at p, without debating "quality awareness."

3. Position: Three Types of Points Must Be Set

After calculating the numbers, the positions must be determined. The three criteria, ranked by priority, are:

Conversely, three types of points are typically not worth setting up: after processes with defect rates consistently below the threshold, before processes where downstream interception is stronger (redundant inspection is like buying two insurance policies), and end-of-line verification for "peace of mind."

4. Five-Step Calculation (Including Example)

Step 1: Map the process and list candidate points. Arrange the processes in order, mark the irreversible points, special characteristic points, and cost escalation points, forming 5 to 8 candidate positions.

Step 2: Collect data. Fill in c, p, F, Fi, and d for each candidate point. Mark the source and confidence level of any missing data; it's better to label it as "estimated" than to leave it blank.

Step 3: Calculate the expected loss per unit for each point. For example, consider a process with a daily output of 2000 pieces: p = 0.8%, c = 0.6 yuan, d = 0.9, F = 120 yuan, Fi = 12 yuan.

• Without an inspection point: 0.008 × 120 = 0.96 yuan per piece, or 1920 yuan per day.
• With an inspection point: 0.6 + 0.008 × 0.1 × 120 + 0.008 × 0.9 × 12 = 0.6 + 0.096 + 0.086 ≈ 0.78 yuan per piece, or 1564 yuan per day.
• Daily savings of about 356 yuan, or about 7500 yuan per month for 21 days; if the inspection tool costs 20,000 yuan, the payback period is about 2.7 months, so it can be set up.

Step 4: Rank and verify capacity. Rank the points by "daily savings amount" from highest to lowest, and set up the top three. Simultaneously verify one point: will the new station become a bottleneck, and will additional work-in-progress (WIP) buffering be needed? Once an inspection station queues up, hidden downtime and rhythm issues can be masked, with costs far exceeding the savings.

Step 5: Regularly recalculate after setting up points. Review three things 4 to 8 weeks after implementation: whether the miss rate has decreased, whether the inspection cost meets expectations, and whether the defect rate has been reduced by improvements. Using the same example, if front-end improvements reduce p to 0.2%: without an inspection point, 0.24 yuan per piece; with an inspection point, 0.6 + 0.024 + 0.022 ≈ 0.65 yuan per piece—removing the point would save 811 yuan per day. Inspection points are temporary costs that fluctuate with defect rates, not permanent positions.

5. Four Common Misjudgments

• Focusing only on detection rate, ignoring downstream flow cost. A detection rate of 95% sounds impressive, but if the defect rate is 0.05% and the downstream flow cost is 20 yuan, the annual savings won't even cover the social security for one inspector.
• Treating inspection as a corrective action. Inspection does not produce conforming products; it only identifies nonconforming products. Setting up an inspection point can only reduce downstream flow, not the occurrence of defects, so p will not decrease just because an additional station is added.
• Using inspection stations as buffers. Piling up materials before the station and rushing work after it may appear to control quality, but it actually masks upstream downtime and changeover issues.
• Making a one-time calculation and setting it in stone. When processes change, suppliers change, or automation is introduced, k and p change, but the inspection point configuration does not. This can either waste money or expose risks.

The configuration of inspection points is essentially an investment decision: spend money where defects can be intercepted and where it is worth doing so. Calculate k and p first, then determine the positions, and finally remove points based on data—this way, reducing inspection points won't be seen as "relaxing quality," and adding points won't be seen as "strong quality awareness."

Inspection points are not safer in greater numbers: use p × d × (F − Fi) > c to determine whether to set up a point, use irreversible points, special characteristic points, and cost escalation points to determine where to set them, and regularly remove points based on changes in defect rates.

Knowledge code: 11.1.1

Version: v20261017

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.