An importer uploads a record. The system accepts it. The broker receives three identifiers. Everyone expects the next shipment to be simpler.
Then someone notices that the shipment is blue and the test file identifies the black version. The certificate PDF was updated, but the registry record and broker instruction still reference an earlier version. The team has two problems: an evidence-scope question and a data-version question. Treating both as a clerical correction can make the record look cleaner while leaving the substantive problem untouched.
The most useful eFiling readiness test is therefore not “can we submit?” It is: can we trace this entry line to the right product and supported certificate, and prove that the broker is using that same version?
The deadline has arrived for the general requirement
As checked on 9 November 2026, CPSC's certificate guidance identifies 8 July 2026 as the general imported-product eFiling date and 8 January 2027 for applicable products entered from a foreign-trade zone. Determine whether certification and eFiling apply to the actual product and transaction; not every product needs the same certificate.
The agency's current FAQ describes initial ACE warnings rather than automatic rejection solely for missing eFiling data, while certificate enforcement continues. That enforcement posture does not make the requirement optional. “The system let the entry through” is not a substantive compliance conclusion.
A color change exposes two different failures
Fictional teaching case. P42, all companies, reports, dates, identifiers and review outcomes below are invented. The case assumes a qualified product review has already established that P42 requires a General Certificate of Conformity. It does not select a rule for a real product, provide a certificate or report an actual test result.
The importer is preparing 120 blue P42 units, lot B07. Its internal file has:
| Layer | Current record | What is unresolved | Who can resolve it |
|---|---|---|---|
| Physical shipment | P42 blue, lot B07 | Does the reviewed evidence cover this variant and production? | Product/compliance owner with the appropriate technical reviewer |
| Test evidence | Report RBLACK01 identifies the black variant | Blue coverage has not been established | Qualified scope assessment; additional testing if needed |
| Certificate | PDF revision 3 names blue after a sales-team edit | A changed description does not establish evidentiary support | Authorized certifier after the evidence issue is resolved |
| Registry and broker | Registry revision 2; broker holds the earlier reference | Even a justified certificate change would not update the handoff automatically | Registry owner and broker, with explicit version reconciliation |
The team must stop using the revised PDF as proof that blue is covered. A color change may or may not affect relevant requirements, depending on materials, coatings, manufacture and the applicable rule. The case supplies no technical basis for deciding that by inspection.
There are two legitimate paths. If a qualified review establishes that existing evidence supports the blue variant under the applicable requirements, record that basis before certification. If it does not, obtain the necessary additional evidence or testing before issuing the corresponding supported certificate. Renaming the report or incrementing a registry version cannot substitute for either path.
For the worked example, assume the review finds the existing record insufficient. The importer holds lot B07, obtains additional testing documented in fictional report RBLUE02, and has the appropriate qualified review confirm the resulting evidence supports certification for the defined blue product and lot. That supplied outcome enables the data repair below; it is not an inference that all color changes require—or can be solved by—the same test.
Choose the filing route after deciding who maintains the record
CPSC offers a Full PGA Message Set, transmitting the required certificate data with the entry, and a Reference PGA Message Set, using a Product Registry record. The reference route relies on Certifier ID, Product ID and Version ID. The eFiling FAQ explains both options.
For an importer with a small number of stable products and repeated entries, a controlled registry record may reduce repeated transcription. For an infrequent or changing assortment, sending full data through a capable broker may be more practical. Neither route makes weak evidence stronger. Compare maintenance responsibility, change frequency, broker capability and how corrections will be communicated—not just how many fields one person types.
The Product Registry is separate from ACE; storing a record does not itself send the broker's entry. Someone must give the broker the correct reference or full data and confirm the intended entry line uses it.
P42's importer chooses Reference because it expects repeat entries of the now-defined blue product and has assigned a registry owner. It will reconsider if variants multiply faster than that owner can maintain the product-to-certificate mapping. An unattended registry is not an efficiency gain.
Repair the chain, then hand over a finished instruction
In the hypothetical, the authorized certifier issues supported certificate revision 4 after reviewing the resolved evidence. The registry owner enters and checks the corresponding data, preserving the superseded record and its history. A second person compares the shipment, certificate and registry before the broker receives the new instruction.
| Link | Final example state | Verification before tender |
|---|---|---|
| Shipment identity | 120 × P42 blue, lot B07 | Physical/packing record matches the accepted product and lot scope |
| Evidence and certificate | RBLUE02 plus other evidence relied upon; certificate revision 4 | Authorized review confirms all applicable requirements and relied-on evidence are represented |
| Registry reference | Certifier ILLUSTRATIVE42; Product P42BLUE; Version 202611A | Registry record matches supported revision 4; IDs are illustrative, not live credentials |
| Broker handoff | New reference assigned to the P42-blue entry line | Broker acknowledges replacement of the earlier reference and confirms the line mapping |
The finished message is:
P42 blue, lot B07, 120 units — superseding filing instruction. Use the reviewed Reference PGA record identified by Certifier ID ILLUSTRATIVE42, Product ID P42BLUE and Version ID 202611A for this product line. This reference corresponds to certificate revision 4, supported by the reviewed evidence pack including RBLUE02. Do not use the previous revision-2 reference or certificate revision 3. Please confirm the product-line mapping and that the earlier instruction has been superseded. If identity, quantity, variant, lot or required data differ from the attached reviewed pack, return the discrepancy for resolution before filing. Do not alter product coverage to make the record submit.
This is a broker handoff example, not a full certificate or a complete electronic message specification. The actual record must contain the required information, use the current technical specification and accurately reflect the evidence relied upon. The identifiers above are invented labels for teaching the connection.
Test the failure, not just the happy path
Before the next repeat shipment, ask the team to handle a deliberate mismatch: the packing list says blue but the instruction says black. A useful rehearsal succeeds when the operator returns the discrepancy to an accountable owner, rather than copying the newer-looking reference and assuming it is correct.
Also test a legitimate update. If a new supported certificate is issued, can the registry owner identify which open shipments still carry the previous instruction, and can the broker confirm each replacement? Preserve enough version history to reconstruct what was actually filed; “latest” is a moving target.
The controlling distinction is simple: technical acceptance tests a transmission; product review supports a certification; a controlled handoff connects the two. A reliable process needs all three. Its next investment should target whichever connection is currently unproven, rather than buying more automation around a record no one can defend.