Why Some Customers Lead to Losses — Allocating Quality Costs by Customer and the Five-Step Decision-Making Method

The monthly quality cost report is out: the COQ across the entire factory is 3.1% of sales, a 0.4 percentage point decrease from last year, and it passed the meeting smoothly. The boss casually asked, "What about the breakdown by customer?" The report couldn't answer. So, the finance and quality departments were tasked to break it down, and the results were not pretty: 2.6% for Customer A, 3.1% for Customer B, and 4.8% for Customer C. More glaringly, out of the 510,000 yuan in external failure costs, 270,000 yuan came from Customer C—despite contributing only 15% of sales, it accounted for more than half of customer complaints, returns, and claims.

This is the most common failure mode of quality costs: not that they can't be calculated, but that they are only calculated as a total sum. The total sum looks good for external reporting, but it provides almost no useful information for internal decision-making. Quality costs are inherently uneven, and the costs of customer complaints, returns, sorting, rework, and production stops never occur in proportion to sales. They are concentrated on a few customers, a few models, and a few processes. If not broken down, good customers quietly subsidize bad ones, and management always sees "overall control."

This article provides a practical approach that can be implemented this month: how to allocate the four types of quality costs to customers, what the report should look like, and how to make decisions after breaking it down.

1. Why Averages Can Be Deceptive

The quality cost rate (COQ ÷ sales) is a ratio. The numerator and denominator of this ratio follow completely different rules: sales follow orders, while quality costs follow nonconforming products. Dividing two different things naturally smooths out the differences.

There are three direct consequences of this smoothing:

First, the investment is diluted. A 3.1% COQ for the entire factory seems fine, so the budget for improvements is unlikely to be approved. However, if broken down, Customer C's 4.8% is already 1.5 times the factory average. This type of customer is where the money is truly leaking, and where improvement investments should be directed.

Second, responsibility is blurred. Customer complaint costs are recorded by the quality department, rework hours by the manufacturing department, return logistics by sales expenses, and claims by non-operating expenses—each department records a segment, and no one has the full picture. The essence of allocation is to reassemble these scattered segments based on the same object.

Third, decision-making lacks a basis. Decisions such as quoting, order acceptance, and capacity allocation depend on whether "this customer is profitable." If quality costs are not allocated to customers, the customer profitability analysis is incomplete: a customer with a 4.8% quality cost rate might appear as a "high-quality customer" on the gross margin sheet but be a loss-making customer in real cash flow.

A counterintuitive arithmetic: a customer with a 15% gross margin and a 4.8% quality cost rate has a net contribution rate of only 10.2%. If the quality costs could be reduced by 300,000 yuan, it would be equivalent to generating an additional 2,000,000 yuan in sales—without the need to allocate more capacity, materials, or delivery times. Which is easier: reducing quality costs or taking more orders? The answer is often clear.

2. Distinguishing Between Direct Allocation and Indirect Allocation

When creating customer-level quality costs, the first step is to abandon the idea of "average allocation" and adopt a priority for direct allocation: all costs that can be directly allocated should be, and only those that cannot be directly allocated should use driver allocation. Estimation should be the last resort. The order of these three levels cannot be reversed.

The judgment criteria are straightforward: can this cost be traced to a specific customer, model, or work order number through a document? Customer complaint forms, return forms, rework work orders, scrap forms, and claim letters all have associated objects and can be directly allocated. Shared investments such as factory depreciation, gauge calibration, and system maintenance require allocation.

Two key ratios to remember:

① The direct allocation ratio should reach 70% or more. Among the four types of costs, internal failure and external failure typically account for more than half of the quality costs, and these two types can almost always be directly allocated through documents. If a company's report has more than 30% of the allocation, it indicates a problem with document management. Address the document issues before building the model.

② External failure costs must be allocated by customer, not by product. The same model sold to different customers can have different acceptance standards, packaging requirements, and claim clauses, leading to quality cost differences that can be several times higher. Allocating by product would completely obscure these differences.

3. Five Steps to Create a Customer-Level Quality Cost Report

Step One: Define the Calculation Object and Analysis Period

The object is recommended to be "customer × product family," not detailed to individual part numbers—too much detail can lead to small data in each cell and higher allocation ratios; nor should it be too broad, only by customer—different product families of the same customer often have very different issues. The period should be monthly, synchronized with financial closing; a quarterly summary and review should be conducted.

Step Two: Build a Subject Mapping Table

Map the subjects in the financial accounts to the four types of quality costs, clearly defining the allocation method and data source. This step must involve the finance department; otherwise, the口径 (caliber) won't match, and the report will be rejected in the management meeting.

The mapping table should have at least five columns: financial subject, quality cost category, allocation method (direct/indirect/estimated), data source (which system and which form), and responsible person. Once the mapping table is finalized, it should not be changed arbitrarily within six months; otherwise, the trend line will break, and improvement effects cannot be back-tested.

Step Three: Select Allocation Drivers

The principle for selecting drivers is "who consumes, who bears," and the driver data itself must be obtainable. Common drivers include:

• Inspection hours: used when sharing inspection lines or test equipment, more reasonable than allocation by sales;
• Rework hours: used for shared rework or mixed-line scrap, reflecting actual usage;
• Customer complaint work order numbers: a reasonable driver for customer complaint handling and after-sales travel expenses.

Sales should be avoided as a driver. Allocating by sales is equivalent to assuming "quality costs are proportional to sales," which is incorrect—it would make large customers bear costs that do not belong to them and make problematic customers appear less severe.

Step Four: Generate the Report

The report rows are customers, and the columns are suggested to be fixed at nine items, no more:

(Unit: 10,000 yuan; example data from a monthly breakdown of a plastic injection molding company)

The most informative column in this table is not the COQ rate but the external failure ratio. Internal failures are "money burned at home," while external failures are "money burned at the customer," often accompanied by reputation loss, production stop penalties, and relationship costs. Customer C's external failure ratio is 31.4% of its quality costs, more than five times that of Customer A, indicating the problem is not "high costs" but "many leaks."

Add two derived indicators for more accurate judgment:

• External failure cost per unit of sales = external failure ÷ sales × 100, used to compare customers horizontally;
• Customer complaint PPM and claim amount, used to determine whether it is a major incident or continuous minor leaks.

Step Five: Back-Testing and Calibration

The report should not be used for decision-making directly; first, conduct a back-test: randomly select two or three known major customer complaints and check if they fall into the correct customer and the correct cost category. If an 80,000 yuan customer screening fee is allocated to three customers, it indicates that the allocation process is still at the indirect stage. Return to Step Two to revise the mapping.

Only after passing the back-test should the report be allowed into the management meeting.

4. After Breaking Down: The Four-Quadrant Decision Matrix

With customer-level data, decision-making has coordinates: the horizontal axis is the COQ rate (relative to the factory average), and the vertical axis is the net contribution rate (gross margin minus quality cost rate).

In the example, Customer C falls into the fourth quadrant: a COQ rate of 4.8% (more than 1.5 times the factory average), a gross margin of 15%, and a net contribution rate of 10.2%. For such customers, there are only four paths, each requiring a cost analysis before discussion:

① Raise prices. Pass half of the quality cost difference through price adjustments to at least cover the increase in external failures. The prerequisite is to have the allocation report as evidence—this is the primary use of the customer-level report, where "we have calculated" and "we feel" are two different things.

② Change specifications or redesign. Many external failures stem from customer specifications not matching the process capability: tolerances tighter than the process capability, appearance standards without limit samples, packaging requirements exceeding the norm. These issues can be resolved through specification clarification, limit sample confirmation, and packaging solution changes, with low costs and high benefits.

③ Transfer or分流 (divert). Transfer the high-difficulty models of this customer to a more capable production line, or switch to a higher-grade supplier. Note that the transfer itself can generate new appraisal costs and ramp-up risks, which should be analyzed on a project basis.

④ Exit. This step should only be taken if the net contribution rate is negative and the three improvement paths have been analyzed and found unfeasible. Before exiting, calculate the capacity idle costs and the impact on customer relationships.

The common premise for these four paths is: decisions should be based on the net contribution rate, not the gross margin or sales. The customer with the highest sales is not necessarily the most profitable, which is the easiest reason for the sales department to resist the allocation report.

5. Five Common Misconceptions

① Allocate by sales. The easiest and most inaccurate method. It systematically pushes costs to large customers while protecting truly problematic customers, ultimately leading to the absurd conclusion that "the larger the customer, the less we dare to accept."

② Only calculate the costs recorded by the quality department. Rework hours are recorded by the manufacturing department, production stop losses by the manufacturing department, customer screening fees by sales expenses, and claims by non-operating expenses. Customer-level cost allocation must draw data from four departments; only taking the quality department's data usually results in a figure that is only one-third of the actual value.

③ Set the allocation model once and for all. Product structure, production line layout, and customer requirements are constantly changing, and the reasonableness of drivers will also change. It is recommended to review drivers every six months, especially when the proportion of shared equipment changes by more than 20%.

④ Treat prevention costs as expendable. Among the four types of costs, prevention costs are the only ones that can reduce total costs; appraisal costs and failure costs are its results. In the allocation report, customers with low prevention costs are often those with high failure costs—this correlation can be seen in the data of most companies.

⑤ Use internal management口径 (caliber) for external communication. Customer-level quality costs are for internal decision-making and do not align with contract claim terms or quality agreement terms. External claims should follow the agreement clauses and evidence chain; do not directly hand over the management numbers after allocation to customers or suppliers, as this will only lead to口径 (caliber) disputes.

6. Monthly Self-Checklist

Check each item before generating the monthly report. If any item fails, the report is still in the "estimation" stage and should not be used for decision-making:

• The calculation object is clearly defined as "customer × product family" and is consistent with the previous month's口径 (caliber);
• The financial subject mapping table has been updated, and new subjects have been classified;
• The direct allocation ratio is ≥70%, and the indirect part has noted the drivers;
• External failure costs are 100% allocated by customer, with traceable complaint work order numbers;
• Rework and scrap are allocated by work order, and mixed-line parts are allocated by rework hours;
• The report includes the external failure ratio and external failure cost per unit of sales;
• 2-3 major customer complaints have been selected for back-testing, and the allocation is correct;
• Customers in the fourth quadrant have a disposal plan and responsible person.

Only after all eight items are passed can the report be confidently presented at the management meeting and used to negotiate with customers.

Quality costs not allocated to customers will always remain an average.

Knowledge code: 4.3.1

Version: v20261022

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.