By Prof. Dr. S. Filo Ambrosini
On a semiconductor line, a rising defect rate can trigger a familiar debate. Do we need a new inspection system? More sensors? An additional engineer?
Those may all be valid responses. But before making the decision, someone needs to establish when the change began, where it first appears, and whether the signal itself is reliable.
Technical companies often make an expensive mistake before choosing a solution: they accept the first description of the problem as if it were a diagnosis.
A measurement becomes unstable. Is the instrument failing, or is the sample preparation inconsistent? Is the process drifting, or has the measurement protocol changed? Is a defect causing the yield loss, or is it simply the first defect that became visible? A customer asks for higher resolution. Do they really need resolution, or do they need repeatability, throughput, or evidence for a qualification decision?
These are not semantic differences. They lead to different purchases, different projects, and different commercial outcomes.
A company may buy a more capable instrument and discover that contamination, vibration, operator practice, software settings, or an undefined protocol was responsible for the original variation. Another company may collect thousands of measurements without deciding what result would trigger a process change. The data grows, but the uncertainty remains.
This is where diagnosis becomes a business instrument.
Diagnosis turns scattered signals into a decision.
Diagnosis is not just finding what is wrong. It is defining the situation well enough to decide what should happen next. Before choosing a solution, it is worth asking four less exciting questions:
What has actually changed?
Where in the process does the change first appear?
Which decision is currently blocked?
What evidence would justify a different action?
The questions are simple, but they force different groups to describe the same situation in operational terms. Research may be focused on improving resolution while manufacturing needs repeatability. A sales team may be promising a capability that service cannot deliver consistently. Management may be asking for growth while the real constraint is capacity, training, or a process with too much variability.
Everyone can be working hard while the company is still solving the wrong problem.
Technical complexity makes this more likely. Specifications create the impression of precision, even when the decision itself is vague. Discussions move quickly towards bandwidth, resolution, sensitivity, automation, software, and other technical features. Those details matter, but only in relation to a job they are expected to do.
The useful question is not always: “What is the best tool?” It may be: “Which decision is impossible today, and what is preventing us from making it?”
That question changes the role of measurement. Measurement is no longer just a way to produce more numbers. It becomes a way to distinguish among possible causes and reduce the risk of a wrong decision. The value of an instrument therefore depends not only on its specification, but also on whether its output can be connected to a real process and a clear action.
The same principle applies to projects and customers. A delayed project may not need more people; it may need a definition of done. A request for a new capability may not justify a new development program; it may reveal that the customer has not yet identified the decision the capability is meant to support. A service problem may look like a hardware problem until the entire workflow is examined.
This is why technical and commercial discussions should meet earlier. Three layers need to stay connected: what is physically happening in the product or process; what the available evidence can actually tell us; and what business decision depends on that evidence.
When these layers are separated, companies tend to compensate with activity. They buy, test, hire, redesign, and schedule more meetings. Sometimes the activity solves the problem. Sometimes it only makes the wrong direction more expensive.
Good diagnosis does not remove uncertainty. It makes uncertainty visible, and therefore manageable. It tells the team what is known, what is not known, and what should be tested before committing more time or money.
The strongest technical organizations are not necessarily those with the most advanced equipment or the largest volume of data. They are often the ones that can stop for a moment, identify the first point at which reality diverged from expectation, and agree on what decision the next measurement is supposed to support.
In technical business, solving the problem is only half the work.
The other half is making sure it is the right problem.
Prof. Dr. S. Filo Ambrosini is a physicist and nanotechnology specialist working at the intersection of scientific instrumentation, materials characterization and business. He currently leads European sales for NT-MDT, focusing on AFM, Raman and TERS, and also teaches physics, mathematics and microeconomics. His work spans research, technical sales and international business development, with a particular interest in how scientific and engineering problems are translated into practical commercial decisions.
Also Read:
Nvidia Sees AGI While OpenAI Sees Danger
Samsung Might Be the Most Unusual Company on Earth
TSMC’s 2026 OIP Forum to Spotlight the Technologies Shaping the Next Era of Chips
Share this post via:
The Invisible Bottleneck Is Costing You Time You Can’t Get Back