How to Move a Class II Device to a New China Manufacturer Without Backdating

Transfers fail in a predictable order. The commercial decision is made in January. The tooling moves in March. First articles arrive in April and look acceptable. Purchasing, under pressure, releases a production order in May. In June someone in quality asks for the process validation supporting product that has already shipped to customers.

At that point there are three options, and two of them are bad. You can stop shipping. You can perform the validation now and explain the gap honestly. Or someone can produce a document with a date in March.

Backdating is not usually a decision anyone makes deliberately. It is what a sequencing failure looks like once it reaches the paperwork.

The sequencing rule

There is one rule that prevents nearly all of it:

Evidence qualifying the new factory must exist before product from the new factory reaches a customer.

Everything below is an elaboration of that sentence. The reason transfers go wrong is not that people disagree with it. It is that the commercial calendar and the qualification calendar are set by different people, and only one of them has a customer waiting.

Which means the real intervention is early: the transfer plan has to be agreed before the tooling moves, and the first shippable date has to be derived from the qualification sequence rather than from the tooling arrival date. That is a conversation with operations and sales, not with quality.

A sequence that produces real evidence

1. Establish what you are transferring

You cannot transfer a design you have not defined. Before anything moves, capture the as-built configuration of the current product: actual BOM with actual approved manufacturers, current drawings at current revisions, firmware identity, critical dimensions and characteristics, labeling, packaging, and the acceptance criteria the outgoing factory was actually using.

This step routinely takes longer than expected because the documented configuration and the shipping configuration have drifted apart. Find that out now rather than during first article inspection, when the new factory's parts will differ from the drawing and nobody will be able to say which is correct.

2. Identify what is critical before you qualify anything

Not every dimension matters equally. Work from your risk analysis: which characteristics, if wrong, produce a hazard? Which processes cannot be fully verified by inspecting the finished device? Those two questions determine where validation effort goes and which characteristics need production controls.

A process whose output you cannot fully verify by inspection, such as welding, sealing, bonding, sterilization, potting, and some overmolding, requires validation rather than inspection. Getting this list right early is what keeps the validation scope proportionate.

3. Qualify the factory before you qualify the process

Supplier qualification is a separate gate and it precedes the technical work. A certificate is a starting point, not the assessment. What matters is whether this factory can build your device: relevant process capability, equipment, measurement systems, environmental controls if you need them, document and change control that will actually notify you, and a quality organization with authority to stop production.

This is also where the quality agreement gets written, before the relationship has momentum and while you still have leverage. Who approves changes, who owns validation, what requires your written approval, how deviations are reported, how long records are retained, and your right to audit.

4. Transfer the design, deliberately

Design transfer is not sending a folder. It is confirming that the receiving factory can produce the device correctly and repeatably, and it produces its own evidence: transferred documentation with acknowledgement, process documentation the factory authored and you reviewed, tooling and fixture qualification, measurement system verification, and operator training records.

The most useful artifact here is a transfer checklist that both companies sign, not as ceremony but because it forces someone to state that each element is complete.

5. First articles, then pilot, then validation

First article inspection confirms the factory can make the part to specification once. It is necessary and it is not sufficient, and treating it as sufficient is the single most common error in the sequence.

A pilot build exercises the actual process, with production tooling, on production equipment, run by production operators, not by engineers. Engineers can make almost anything work once. The pilot exists to reveal what happens when they are not standing there.

Process validation then demonstrates the process consistently produces conforming output: installation qualification for equipment, operational qualification across the operating range, and performance qualification over multiple runs under production conditions. Where the process is not fully verifiable by inspection, this is not optional.

6. Verify the product, not just the process

A validated process at a new factory can still yield a device that behaves differently. Different resin lot, different supplier for a component you did not designate critical, different assembly sequence with different residual stress.

Run the verification tests that matter for safety against product from the new line. For most transfers this is a subset, chosen by risk, not the full original battery. If your safety certification is site-specific, the certification body has to be engaged here as well. Their lead time is frequently the longest single item in the plan, which is a good reason to ask early.

7. Release deliberately, and say so on the record

Manufacturing release should be an explicit decision with a date, made by named people, referencing the evidence that supports it. Not a purchase order.

If you ship before that decision, you have a gap. Record the gap, assess whether shipped product is affected, and close it. That is an unpleasant conversation and a functioning quality system. The alternative is a document with a false date, which converts a schedule problem into an integrity problem.

What goes wrong even when the sequence is right

• Silent substitutions. The new factory sources a component from a different manufacturer without telling you, because your agreement never said it had to. Designate critical components explicitly and require written approval by name.
• Validation performed on the wrong configuration. Validation runs during a design change, and the validated process describes a product you no longer ship. Freeze the configuration for the duration.
• Training records created after the fact. Operators are trained during the pilot; records get written the week before an audit. Capture them at the time, with real dates.
• The old factory's tribal knowledge never transfers. The undocumented adjustments that made the process work leave with the outgoing team. Interview the outgoing line before it disperses; this is cheap and the window closes permanently.
• Nobody owns the evidence. Your side assumes the factory is compiling it; the factory assumes it is filing to its own system. Both are true, and neither produces a package you can hand an auditor.

Why the last one is the hard one

The first four failures are addressed by discipline. The fifth is structural, and it is why transfers with competent parties on both sides still produce incoherent evidence.

The factory documents in its system. You document in yours. Approvals happen in email. Six months later, reconstructing who approved which version of what requires an archaeology project across two companies, and the answer is frequently that nobody formally approved it at all. A manager replied "looks good" to a message with an attachment, and production started.

A transfer runs better when both companies work from one shared record: the factory proposes the process change, your quality lead reviews the exact version proposed, both sign that version, and the qualification evidence attaches to the decision rather than living in an inbox. The evidence package then exists as a by-product of running the transfer properly, instead of being assembled retrospectively under deadline.

That is also the difference between a transfer you could defend in an inspection and one you would rather not discuss.