How we built a China supplier path for a German manufacturer's inspection bottleneck.

Client details anonymized

The first request sounded simple: automate final inspection and packing.

The manufacturer was a mid-sized German company producing small precision assemblies used in industrial fluid-control equipment. The parts were not glamorous: palm-sized plastic-and-metal components with a molded body, a small metal insert, two O-rings, a printed batch code, and a protective cap before packing.

The production issue was familiar. The upstream assembly process was reasonably stable, but the end of the line still depended on operators for visual inspection, orientation checks, tray transfer, label verification, and final packing. On the day shift, the team could keep up. On the late shift, quality escapes and work-in-process started to build.

The target was not full factory transformation. The customer wanted to reduce manual inspection load, stabilize pack-out quality, and understand whether Chinese automation suppliers could build a practical cell for Germany without creating a support problem after installation.

The hard part was not moving the part. It was deciding what had to be proven.

At first glance, the automation looked like a pick-and-place problem. Parts arrive in small trays, a robot or gantry moves them, a camera checks them, and good parts go to packing. But once we wrote down the actual acceptance criteria, the risk shifted.

The cell needed to check O-ring presence, cap seating, molded flash, metal insert orientation, printed batch code readability, label match, and basic surface defects. Some defects were obvious. Some were borderline. Some only appeared under certain lighting angles because the black plastic body reflected light differently by mold cavity.

The customer also had German plant standards to respect: preferred Siemens controls, CE documentation expectations, German-language operator screens if possible, spare-parts clarity, remote access rules, and a factory acceptance test that would be meaningful before shipment.

We split the project into four supplier conversations.

The mistake would have been sending the same RFQ to every "automation supplier" and comparing prices. We separated the problem by risk ownership, then spoke with different supplier types in China.

The supplier responses told us which route was realistic.

The early supplier calls were useful because they disagreed in productive ways. One vision supplier wanted to start with a bench test and lighting study before discussing machine layout. That was a good sign. Another supplier quickly quoted a complete machine but assumed one top-down camera would catch every defect. That was a warning sign.

A custom equipment builder proposed a rotary indexing machine with multiple inspection stations, but the layout made changeover awkward and treated label verification as a small accessory. A robot integrator proposed a flexible robot cell, but the robot time looked marginal once reject handling, code reading, and packing steps were added.

The strongest direction came from a supplier team that did not try to make one camera do everything. Their concept separated part presentation, O-ring and cap checks, code reading, label verification, and final packing into testable stations. It was less flashy than a robot demo, but it made the acceptance test much clearer.

The solution path became staged instead of one big leap.

We recommended a staged route. First, run a vision feasibility package with real samples: good parts, bad parts, borderline O-rings, shifted caps, weak print, glossy surfaces, cavity variation, and packaging labels. The output should be annotated images, lighting setup, false reject assumptions, and a written pass/fail method.

Second, use that evidence to choose the machine architecture. The likely path was a compact custom inspection and packing cell: tray infeed, controlled part presentation, multi-angle inspection, code and label verification, reject bins with traceability, good-part packing, Siemens PLC/HMI, guarded access, and a defined FAT using the customer's sample set.

Third, keep the robot question open until the vision and handling proof was done. A cobot sounded attractive because the part was small and the cell was human-adjacent. But if the cycle time required faster motion and the safety case became guarded anyway, a SCARA or Cartesian gantry could be simpler, cheaper, and easier to maintain.

The China visit plan changed after the supplier map.

Instead of visiting every supplier that claimed inspection automation experience, the customer could use three focused visit types. One visit would review vision testing and sample evidence. One would inspect custom machine build quality, wiring, guarding, HMI work, and FAT discipline. One would test controls support, documentation, spare parts, and remote-service assumptions for Germany.

That gave the visit a purpose. The buyer was no longer touring factories and asking, "Can you build this?" They were asking, "Can you prove this inspection, own this handoff, support this controls standard, and run this FAT before shipment?"

The sourcing logic we used.

This is the kind of project where a normal supplier shortlist is not enough. A German buyer could easily receive five confident quotes and still not know which supplier understood the actual risk. The work was to turn the vague category - inspection and packing automation - into a set of proof points.

For this case, those proof points were sample coverage, lighting, reject handling, label match, controls standards, documentation, remote support, and FAT criteria. Once those were visible, the supplier comparison became much cleaner.

The best China sourcing work often looks like this: slow down the first question, separate the supplier types, force the proof into the open, and only then decide what kind of machine should be built.

Have a similar inspection or packing bottleneck?

If your plant is relying on manual inspection, tray handling, labeling, or final packing and you want to evaluate Chinese automation suppliers, start with product photos, defect samples, current process video, target cycle time, label requirements, controls standards, and installation country.

Send us the project outline and we will map the supplier categories, RFQ evidence, and validation steps before you start comparing quotes.