Most factory visits generate a large amount of low-resolution information. The factory has machines. The team seems friendly. The floor appears organized. Samples look good. A presentation lists international customers. The visitor leaves with photographs and a general feeling.
Those observations are not worthless. They are weak because they are not connected to the decision the buyer must make. A useful visit needs an evidence architecture: a deliberate path from business question to observable proof.
Start with the decision, not the itinerary
“Visit three factories” is a travel plan. “Determine which supplier can own the pilot build without uncontrolled design changes” is a decision objective.
Before outreach, write the decision and the uncertainties that prevent it. A supplier-selection visit may need to resolve process capability, engineering ownership, commercial scope, or launch priority. A technology-company visit may focus on integration maturity, product roadmap, service coverage, or fit with the buyer’s operating environment.
Every agenda block should resolve one uncertainty. If a showroom walk does not change the decision, it should not dominate the visit.
Build a chain of evidence
1. Claim
The supplier states a capability: “We perform end-of-line functional testing on every unit.”
2. Record
Ask for a recent test record. Can it be traced to a unit, date, operator or station, software version, limits, and result?
3. Process
Go to the station. Does the observed sequence match the record? Are limits controlled? What prevents a skipped step? What happens when the result fails?
4. Owner
Speak with the person responsible for maintaining the test and deciding on exceptions. Can that person explain calibration, false failures, rework, and change approval?
5. Exception
Trace one failed unit. The exception tests the system more strongly than a passing result because it reveals containment, diagnosis, authority, and learning.
A claim becomes decision-grade only when it survives contact with the record, the process, the owner, and an exception.
Use vertical slices instead of broad tours
A broad tour covers every department lightly. A vertical slice follows one important characteristic deeply across departments.
For a machined component, the slice might begin with a functional tolerance, then move through drawing control, fixture strategy, machining, in-process measurement, tool-life management, final inspection, and nonconformance disposition. For an electronics product, it might follow firmware revision, programming, calibration, serialized test, rework, and shipment record. For furniture, it could trace moisture, finish approval, hardware installation, assembly strength, packaging, and transit validation.
Two or three vertical slices usually reveal more about operating control than an hour of walking every aisle.
Separate existence, use, and effectiveness
Many visit observations establish only that something exists: a machine, certificate, procedure, test fixture, quality board, or engineering team. Diligence should move through three levels.
| Level | Question | Example |
|---|---|---|
| Existence | Is the control present? | A work instruction is posted. |
| Use | Is it current and followed? | The revision matches the released drawing and operator sequence. |
| Effectiveness | Does it prevent or detect the intended risk? | Recent failures were caught at the step and triggered the defined reaction. |
A beautiful document that is not used is decoration. A faithfully followed control that does not detect the failure mode is activity, not effectiveness.
Triangulate across roles
Ask the same operating question at different levels. Sales may describe the intended model, engineering the technical model, production the daily model, and quality the exception model. Differences are not automatically dishonesty. They show where assumptions or handoffs need clarification.
For example, ask how a drawing change reaches production. Then trace a recent change through the commercial team, engineer, document controller, line instruction, inspection program, inventory disposition, and first revised shipment. The path matters more than the procedure title.
Record evidence and uncertainty separately
Visit notes often mix observation with interpretation. “Good quality system” may mean the floor was clean, documents were visible, and the quality manager was confident. Those are three observations, none of which proves the conclusion.
Use four columns:
- Question: the uncertainty being tested.
- Evidence: what was observed, shown, or stated.
- Interpretation: what the evidence suggests and with what confidence.
- Open action: what must be provided, tested, or clarified next.
This structure prevents a strong impression from silently becoming a factual conclusion. It also makes supplier follow-up precise.
End with a decision memo, not a photo archive
Within a day of the visit, translate the evidence into a short decision memo. State what the visit changed, which assumptions were confirmed or weakened, what remains unresolved, and the smallest next commitment that can resolve it.
Possible outcomes include proceeding to samples, changing the scope, adding a technical audit, visiting a subcontractor, running a capability trial, redesigning an interface, negotiating an ownership boundary, or rejecting the supplier. “We liked them” is not an outcome.
A good factory visit may still feel impressive. The difference is that the impression is supported by a chain the buyer can explain, test, and use after leaving the gate.