A factory list answers where you might go. It does not answer why each visit deserves a day of the team’s time.
The common planning failure is to collect companies first and invent the purpose later. A well-known brand is added because it sounds impressive. A large factory is added because it is nearby. A supplier is added because someone replied quickly. By the end, the route is full but the decision logic is empty.
A decision tree reverses the process. It begins with the decisions the team expects to make during or immediately after the trip, then identifies the evidence and company types needed to support those decisions.
Write the decision in operational language
“Understand the China market” is too broad. “Decide whether to source a complete subsystem, license a technology, or assemble from selected components” is useful. “Find good factories” is vague. “Select one supplier for pilot tooling and identify a credible second source before launch” can be tested.
A good decision statement includes an object, alternatives, timing, and consequence. It also makes clear who will act after the evidence is collected.
Identify the uncertainty blocking the decision
Teams often believe they need more information when they actually need one particular uncertainty resolved. Examples include:
- whether the required process capability belongs inside the lead factory, with a systems integrator, or with a specialist subcontractor;
- whether the dominant risk is technical performance, supplier attention, compliance, service, or ramp capacity;
- whether a named company can support the buyer’s use case outside its standard market;
- whether a component-level approach gives better control than buying a complete system;
- whether the product is mature enough for supplier selection or still requires architecture work.
Each uncertainty suggests a different visit. A production-line tour cannot answer a software ownership question. A corporate showroom cannot prove field service. A component factory cannot explain who will own final system integration.
Translate uncertainty into evidence requests
| Uncertainty | Useful evidence | Best visit counterpart |
|---|---|---|
| Can the supplier own NPI? | Recent engineering change, pilot records, test development, named project team | NPI engineer, project manager, quality lead |
| Can the technology deploy at our site? | Interface map, exception handling, service path, reference architecture | Application engineering and field service |
| Is capacity real? | Current loading, cycle times, bottleneck process, shift model, expansion trigger | Production planning and operations |
| Is subcontracting controlled? | Approved-provider list, incoming records, change approval, recent issue | Supplier quality and purchasing |
| Can we commercialize the design? | BOM control, interface maturity, DFM issues, tooling and validation plan | Engineering and manufacturing teams |
Build branch conditions before the visit
A decision tree needs branches. Define what happens if the evidence is strong, mixed, or weak.
If a supplier demonstrates process control and engineering ownership, the next action may be a paid trial or sample build. If the process is credible but an external capability remains unclear, the next action may be a subcontractor visit. If the factory is strong but the project will be too small to receive attention, the team may keep it as a future scale option and continue evaluating specialists.
Predefined branches reduce the chance that hospitality, brand prestige, or one charismatic executive overwhelms the actual evidence.
Choose a portfolio of company types
A useful route may include different roles rather than several near-identical factories. For a technology project, the portfolio could include a product company, core component supplier, systems integrator, and deployed application. For a sourcing program, it may include a scaled manufacturer, responsive specialist, critical external process, and test or tooling partner.
The purpose is not variety for its own sake. Each company should test a different strategic or operational alternative.
Leave space for the network to teach you
Decision discipline does not mean rigidity. One of China’s strengths is that a good conversation can expose a capability or supplier the buyer did not know to request. Keep route space for follow-up, but connect the new opportunity to the decision tree before adding another meeting.
Ask: which uncertainty does this new company resolve? What would we need to observe? What scheduled activity becomes less valuable if we add it?
Convert every visit into a decision record
At the end of each day, separate evidence, interpretation, uncertainty, and action. Update the decision tree while the conversations are fresh. The route may need to change because a supplier exposed a new dependency or because a company type has already been eliminated.
The final output should not be a diary of where the group went. It should show which alternatives remain, why, what evidence supports them, and the next smallest commitment.
China rewards prepared teams because specificity attracts the right people and makes access more useful. Arrive with questions that can change the plan, and the factories become part of a decision system rather than stops on a tour.