When a potential €3 Million Recall Turned Into a Sorting Action
The call came in with the kind of urgency that makes quality teams freeze.
An electronics supplier had just discovered that a batch of components — already delivered, already integrated, already shipped — might be non-conforming. Not in one product. Not in one country. Across multiple product lines, distributed to end customers throughout Europe.
The initial estimate landed fast: a potential recall affecting thousands of units, with a cost exposure around €3 million. Beyond the financial hit, there was a harder question no one wanted to answer: did end customers need to be warned about a safety risk?
That is where we came in.
The problem with raw risk numbers
When a potential safety defect surfaces, the first instinct is to contain. Stop the bleeding. Pull the products. Inform the customers. It feels responsible. It feels safe.
But raw risk estimation is not the same as real risk. And acting on a worst-case number — before anyone has truly understood the failure mode — is how companies turn a manageable problem into a catastrophic one.
The root cause in this case was a machine setup error at the supplier level. A misconfiguration had led to a misinterpretation of a measurement signal during manufacturing. The concern: under certain conditions, a short circuit could produce a thermal incident. On paper, that sounds serious. In a specific context, with a specific application, the picture can look very different. That distinction matters enormously. We were brought in to find it.
Taking over the root cause analysis
Our first step was not to recommend a path forward. It was to understand the problem ourselves — deeply, technically, without assumptions.
We took over the root cause analysis entirely. That meant going back to the supplier data, the process parameters, the measurement chain, and the failure physics. It meant asking the questions that had not yet been asked: how often would this setup error actually produce a non-conforming component? What was the real population of suspect parts? What were the boundary conditions under which the failure mode could actually manifest?
These are not questions quality checklists answer. They require engineering judgment, hands-on investigation, and the willingness to challenge conclusions that feel settled. We challenged them.
Understanding the component in context
The critical insight in cases like this is that a component never fails in isolation. It fails — or doesn't — in a system. In an application. Under real operating conditions.
We worked closely with the engineering teams to model the component within its actual use case. What currents were flowing? What thermal margins existed? What protection mechanisms were already present in the end products? What would actually need to go wrong — simultaneously — for a thermal incident to occur?
This work required both sides to be uncomfortable. For our client's teams, it meant revisiting assumptions they had already signed off on. For the supplier, it meant full transparency about process data they would have preferred to keep internal. We pushed until we had the complete picture. What emerged was a risk profile that looked nothing like the initial estimate.
From €3 million exposure to a targeted sorting action
The structured risk assessment — grounded in physics, in real application data, and in the actual population of suspect components — led to a clear conclusion: the real-world risk to end users was negligible compared to what the raw numbers had suggested.
The conditions required for a thermal incident to occur did not align with how the products were actually used. The probability of the failure mode manifesting fell well below the threshold that would trigger customer notification or recall obligations.
That conclusion had to be communicated — and that communication had to be airtight. We led the risk assessment documentation and the communication strategy across the supply chain and with every stakeholder who needed to sign off. The argument was not "we think the risk is low." It was: here is the data, here is the failure physics, here is why the real risk is demonstrably different from the theoretical exposure — and here is why no recall or customer notification is warranted.
It held.
No recall. No customer notification.
Problem closed at the supply chain level.
Suspect parts identified and removed from warehouses across Europe. The crisis never reached the end customer.