A customer complaint always arrives at the worst possible time. You handle the return, send the replacement material and close the case. Cost recorded in the system: 800 euros.
What does not appear in that record is the real cost of that complaint. And it is probably ten times higher.
How much a quality complaint really costs
The cost of a complaint has two layers. The first is visible. The second is the one that destroys margins without anyone accounting for it.
Layer 1: direct costs (visible) | Layer 2: hidden costs (invisible) |
|---|---|
Defective product (material, labour, scrap) | Engineering hours investigating the root cause of the failure |
Return and reshipment transport | Customer line downtime while waiting for material |
Rework or replacement of parts | Extraordinary customer audit at your plant |
Contractual penalties | Loss of order volume in the next negotiation |
Quality team hours managing the case | Reputational damage if the failure reaches other customers in the sector |
According to common cost of poor quality estimates (the American Society for Quality cites ranges between 5% and 35% of turnover depending on the maturity of each company's quality system), in organisations with reasonably mature systems the total cost usually lies between 5% and 15% of annual turnover. For a company with 10 million euros in sales, that means between 500,000 and 1.5 million euros a year. Most are hidden costs that never appear in any report.
What happens in practice: three sectors, three scenarios
Abstract figures help very little. What does help is seeing what happens in specific sectors when a defect escapes quality control.
Sector | Typical cost of a failure | Contextual data |
|---|---|---|
Food and consumer goods | ~10 million € in direct recall costs (without counting future sales or brand) | 2024 was a record year for recalls in Europe: more than 7,000 alerts in the first half of the year across the five sectors tracked by Sedgwick |
Automotive | Between 500 and 2,000 € per affected vehicle in a recall of safety parts (figure to be confirmed with you) | Global warranty and recall costs have risen by 63% since 2014 (from 35,000 to 57,000 million dollars among major manufacturers) |
Metallurgy and industrial components | Cost of material, machine time and round-trip logistics, plus customer line downtime | An hour of downtime in a continuous process can cost between 5,000 and 50,000 € depending on the sector (figure to be confirmed with you) |
In the food sector, the problem is not always contamination: there is also a significant proportion of packaging defects, incorrect labelling or visible foreign bodies—exactly the type of defect that a machine vision system can segregate on the line before it leaves the plant.
In automotive, the problem does not start with the finished part: it starts in process inspection, where a surface or dimensional defect that should have been segregated on the line ends up fitted in the vehicle.
In metallurgy, the real damage is not the rejected batch itself, but the customer line downtime it causes. If that downtime is attributable to a component supplier, the penalty is usually reflected in the following year's contract.
Why your current inspection lets defects slip through even if it doesn't seem like it
Before looking for solutions, it is worth understanding why traditional systems fail. It is not a question of lack of rigour: it is a question of physical limits.
Sampling inspection has a structural gap. If you inspect 10% of the parts in a batch, the remaining 90% goes uninspected. It is statistically inevitable that some defects will slip through, and those that slip through are precisely the ones that reach the customer.
Visual criteria are subjective. The acceptance threshold for a surface scratch is not the same first thing in the morning as it is at the end of a long shift. Not because the operator is untrained, but because human visual judgement varies with fatigue and concentration.
Without automatic traceability, investigating a failure is expensive. Can you tell in which exact shift the defect occurred? With which batch of raw material? On which machine? Without an automatic log of each inspection, investigating the root cause of the failure can take days and consume many engineering hours.
What changes when you inspect 100% of parts automatically
A machine vision system resolves the three previous issues at root: it inspects every single part, not a sample; it always applies the same criteria, regardless of shift or operator; and it automatically logs every inspection: part image, result, type of defect detected, and exact timestamp.
With Rely, you configure inspection criteria in natural language, code-free. The camera captures the image of each part on the line, the system analyses it in real time, and the defective part is segregated before reaching packaging. If a complaint arrives, you have the complete visual history of that batch available in minutes: which parts passed inspection, which did not, and why.
That changes the conversation with the customer. Instead of explaining what went wrong and what corrective measures you will take, you can present data: complete traceability of every unit shipped. That is not just operational efficiency; it is an advantage in every audit and every contract negotiation.
Before evaluating an automated inspection system, ask yourself these four questions
Not all machine vision systems fit equally well into every environment. Before speaking to any provider, it is worth clarifying these points:
What type of defect do you want to detect? A surface defect (scratch, stain, pinhole) is not the same as an assembly error (missing component, incorrect position) or a dimensional defect (measurement out of tolerance). Each type requires a different approach.
What is your line cadence? Production speed determines the time available per part and conditions the choice of image capture system.
Do you have cameras installed? Many systems, including Rely, work with industrial or IP cameras already installed on the line. You don't always have to start from scratch.
Who will configure and maintain the system? If you need a specialised integrator for every reference change, operating costs can be prohibitive. No-code systems, where the process engineer configures directly without programming, completely change the equation.
If you have answers to these four questions, you are in a position to evaluate a specific solution. If not, it is the first step before speaking to anyone.
Frequently asked questions
What is the true average cost of a quality complaint?
There is no universal figure, because it depends on the sector and the type of defect, but the pattern repeats: the direct cost recorded in the system is usually only a fraction of the total cost once engineering hours, customer line downtime and the impact on the next contract negotiation are added up.
Isn't sampling inspection enough if we already have a good quality team?
The problem is not team training, it is statistics: if only a sample is checked, part of the batch always remains uninspected, whatever the rigour of the process.
Does Rely need new cameras or can we use those already installed?
In many cases, industrial or IP cameras already on the line will work; you don't always have to start from scratch with hardware.
How long does it take to configure a new inspection criterion?
As it is configured in natural language, without code, a process engineer can adjust a criterion directly, without relying on an external integrator for every reference change.
How much is the last complaint you handled really costing you?
Book a free 30-minute demo and we will see how Rely inspects your specific product type, with no prior installation.
Sources: Grocery Manufacturers Association / Covington & Burling / EY, 2011; Sedgwick European Recall Index, 2024; Autoline, "The $57 Billion Secret", 2025; cost of poor quality ranges according to American Society for Quality.