China supplier visit report: what to record, section by section (with template structure)

This guide covers the buyer-side or agent-side visit report — the document produced after a qualification visit or a follow-up visit. It is not the formal third-party inspection report written against a purchase order and sampling plan; that document has its own anatomy and is covered separately in Factory inspection report — anatomy. A visit report helps decide whether to use the supplier; an inspection report helps decide whether to accept specific goods.

Visit report vs inspection report vs audit report

The three documents answer different questions; blurring them is the commonest reason visit reports disappoint.

A visit report borrows from both neighbours, but its job is a decision about a relationship, not a verdict on a batch.

The template, section by section

The sequence below doubles as the template — it mirrors the order of a well-run visit, moving from identity, to capability, to control, to evidence, to decision. For each section: what to record, what good evidence looks like, and the omission that recurs.

1. Visit logistics and attendance

Record: the date, the address actually visited (not the one on the website or the business card — the registered address can be a sales office in a different city), total time on site and, separately, time on the production floor. List attendees from both sides with roles, note which functions were absent, and whether an interpreter was used.

Good evidence: photographed business cards; the building signage on arrival; a distinction between the people who answered questions and the people who only attended.

Classic omission: the floor-time figure. Six hours on site with forty minutes on the floor is a different visit from six hours walking production — the reader should be able to tell which one happened.

2. Legal identity, verified on site

Record: the registered Chinese name and the 18-digit unified social credit code from the business licence presented, then cross-check both against the national enterprise registry after the visit. Note whether the registered address matches the plant you stood in, and whether the names on the gate, uniforms and internal paperwork match the licence — and match the entity that is actually quoting.

Good evidence: a legible photograph of the licence, paired with the registry extract it was checked against.

Classic omission: recording only the English trading name — which is unregulated and cross-checks against nothing. Seeing a licence on a wall is not the same as matching it to the registry.

3. The floor walk — production capability and in-process QC

The heart of the report. Walk the line in production order, not in the order the host wants to walk it.

Record: the process steps observed in-house against the steps the product requires; machines seen running (type, make, rough age, condition) versus those idle or under dust; whether in-process QC records were being completed live; and what was actually in production that day — and whether it belongs to the product family you intend to buy.

Good evidence: nameplate close-ups bracketed by wide shots; an in-process check sheet photographed with that day’s date on it; QC records pulled at random and read, not just observed.

Classic omission: transcribing the machine list handed over by sales instead of the machines observed — and failing to record the process step that was absent, which is usually the subcontracted one. A clean, idle workshop is not a working factory.

4. Workforce and capacity signals

Record: headcount claimed against heads roughly counted by area, with the basis for the estimate; the shift pattern; the proportion of workstations occupied; supervisors and QC staff visible on the floor; recruitment banners at the gate. Anchor it in date context — a visit two days before Chinese New Year says little about a normal Tuesday.

Good evidence: time-stamped wide shots of occupied lines, reconciled against the capacity claimed in the quotation.

Classic omission: a capacity claim (“X units a month”) recorded with nothing observed that supports or contradicts it — the bare number then migrates into later documents as if it were a finding.

5. Materials and traceability

Record: how raw material is stored — identified, segregated, labelled, or not; whether mill certificates were produced and whether heat or batch numbers on certificates match markings on stock; and the brands of critical bought-in components (motors, bearings, switchgear) actually present in stores against those promised in the quotation.

Good evidence: pull one certificate at random and walk it forward to a part in production, or pull one part back to its certificate. If neither direction works, traceability is a claim, not a system.

Classic omission: reviewing certificates in the office without checking that the material in the rack can be tied to them — or accepting “we use [brand]” verbally without sighting a single boxed unit.

6. Testing capability — witnessed, not listed

Record: the test and measurement equipment present, with calibration labels and dates; what was demonstrated, not merely pointed at; and which tests can actually be performed in-house versus sent out. Ask for a live measurement or test on a current workpiece and write down the result.

Good evidence: a short list of tests witnessed, with the instrument used and its calibration reference — and a separate list of tests claimed but not witnessed.

Classic omission: “they have a CMM” — with no calibration date, no fixtures and no recent measurement records nearby, that records a machine’s existence, not a capability.

7. Documentation review on site

Record: which documents were sighted in original, with certificate numbers, issuing bodies, scopes and expiry dates: ISO certification, product certificates (for CE, record the issuer and the directives — a self-declaration is not a notified-body certificate), recent internal inspection records, drawing-revision control, non-conformance and corrective-action records. Note what was promised but not produced, and what was refused.

Good evidence: a small table of certificate numbers ready for desk verification after the visit.

Classic omission: photographing the wall of framed certificates without a single legible number — decorative, and unverifiable later. Listing documents emailed afterwards as if they had been reviewed on site. Receiving a document is not the same as verifying it.

8. Management interview

Record: who actually answered each substantive question — general manager, production manager, quality manager or salesperson — with decision-relevant answers quoted against the named role: subcontracting policy, engineering change handling, order book and customer mix, capacity commitments, warranty handling. Note disagreements between interviewees.

Good evidence: attributed, checkable statements (“Production manager: welding is in-house; surface treatment subcontracted to a plater in [city]”) rather than impressions (“seems experienced”).

Classic omission: every answer filtered through the salesperson, never tested against the people who run the floor.

9. Red flags observed

Give warning signs their own section so they cannot be buried: entity-name mismatches; areas where photography or entry was refused — the refusal is itself a finding; answers that changed between people; a floor full of product from an unrelated industry; certificates not produced when asked; an immaculate and idle “showroom” workshop. Record each factually, without adjectives — a red flag described neutrally is harder to argue away — and carry every one into the conclusion.

Classic omission: scattering them as asides through other sections, where a reader scanning for risk will miss them.

10. Photographic evidence rules

Caption every photograph: what it shows, where it was taken, why it matters. Shoot wide before close so context survives; make nameplates, labels and certificates legible; keep time-stamps on; number files so they map to report sections. No supplier-provided photos unless explicitly labelled as such; no photographs of people without consent.

Classic omission: two hundred uncaptioned photographs in a folder. Evidence that cannot be located or interpreted is not evidence.

11. Conclusion and recommendation, with the explicit next step

The conclusion must answer three questions: is this supplier capable of the work in scope; what remains unverified after the visit; and what should the buyer do next, by when, owned by whom.

Good evidence: a single recommendation — proceed, proceed with conditions, disqualify — followed by the specific next step (“issue trial order limited to [scope], subject to pre-shipment inspection”; “obtain registry extract and re-assess”).

Classic omission: ending with “the supplier is suitable for further evaluation”. That is not a recommendation, it is a deferral — and it quietly pushes the decision onto people with less information than the visitor had.

The ten details visitors record but never use — and the five they forget

Visit reports bloat with supplier claims dressed as findings, while the observations that decide cases go unwritten.

Recorded, but never used in any decision:

1. The supplier’s introduction presentation, transcribed — it existed before you arrived and verifies nothing.
2. Claimed annual turnover — unaudited and unanchored.
3. Claimed export share and the list of export markets — unverifiable on site.
4. The machine list provided by the sales team — a different document from the list of machines you saw running.
5. Floor area in square metres — quoted, not measured, rarely decision-relevant alone.
6. Customer logos in the showroom — sometimes aspirational, never evidence.
7. Award plaques and local-government honours — uninterpretable without knowing who issues them.
8. The LED welcome banner with your name on it — it proves the visit happened; you already knew that.
9. Reception-room certificates with no scope, validity or number checked.
10. The founding year as told — the registry holds the real date.

Forgotten, but decisive:

1. Time on the floor versus time in the meeting room.
2. What was in production that day — and whether it belongs to the product family being sourced.
3. The names on things: gate, uniforms, internal documents, packing cases — and whether they match the licence and the quoting entity.
4. Which questions were dodged, deferred, or answered only by the salesperson. The unanswered question is a finding.
5. What was absent: the process step with no corresponding equipment, the laboratory with no calibration labels, the material store too empty for the claimed order book. Absence is the most under-recorded category of evidence on Chinese factory visits.

How findings map to a decision

The report should close on one of three recommendations, and the wording matters — procurement teams reading at a distance act on the verb.

Proceed. Use when identity is verified, the capabilities that matter were witnessed, and documentation checks out. Word it as scoped, never as blanket endorsement:

Qualification is for a scope, not for everything the supplier may ever quote — and do not write “proceed” just because the visit was friendly or the factory looked busy. Busy is not the same as controlled.

Proceed with conditions. The most common, and most honest, outcome of a first visit. Each condition must be a verifiable action with an owner and a deadline, mapped to a specific finding. Weak: “supplier should improve quality control.” Usable:

Disqualify. State the disqualifying findings as facts and keep adjectives out — facts disqualify; opinions invite argument:

Distinguish a scope mismatch, which is re-assessable if requirements change, from misrepresentation — an entity or capability presented as something it is not — which normally ends the matter regardless of price.

Frequently asked questions