Product sourcing guide

Power Bank Batch Traceability: Can One Affected Lot Be Identified?

A fictional tabletop exercise starts with a shipped power bank and an authorized suspect-input reference. Can the team work backward to the unit’s build, then forward to every represented destination without hiding missing links? This is a documentary challenge, not a recall initiation or a finding about ChargeKeku products.

The broader power-bank supplier evaluation guide already asks how quickly a suspect cell lot can be isolated. Here, the deliverable is a versioned population report: who matches, who has evidence of non-match, and who remains unresolved.

Read the selector, not just the product name

OPSS’s IKEA VARMFRONT report, case 2407-0105, describes potentially faulty cells, a fire risk and an end-user recall by IKEA. Its Annex A gives these exact combinations:

Published modelArticle numberListed date stamps
E2037705556472318, 2319 or 2322
E2038105556452313, 2316, 2318 or 2319

The official companion notice supplies the 31 July 2024 publication/recall date; the PDF itself is undated. The notice’s title says 10500 mAh, while its table and the PDF say 10400 mAh for E2038. Keep that discrepancy visible rather than silently “correcting” it. Neither document supplies internal cell-lot or serial-number genealogy.

For this published case, a shared 2318 stamp alone cannot select the affected combinations. Preserve model, article number and listed stamp together. Do not decode the stamp into an assumed calendar rule or generalize this case to other products.

Pass one: recover the shipped unit’s as-built identity

Start the two-way trace worksheet with a de-identified unit or explicitly bounded cohort. Record the marked identifiers, label-evidence reference and actual shipment/carton membership. Follow controlled links to the assembly build order, as-built revision and cell supplier/model/lot evidence. Do not invent serial ranges when records identify only cartons.

Preserve original and subsequent rework or replacement records with their effective membership. A current BOM is not, by itself, evidence of what an earlier unit contained. If several input sources fed one assembly order, retain the unresolved join rather than choosing the most convenient source.

The UN38.3 test-summary identity checklist owns design/report matching. This exercise asks a different question: which physical units and destinations are represented by the available records?

Pass two: query forward with a versioned selector

Record the authorized selector’s reference, revision, scope, approving owner and snapshot time. Query every represented assembly, rework and onward-distribution membership, not just the first finished-goods lot found. Preserve the query criteria and supporting manifests so another reviewer can reproduce the population. State the searched record universe and evidence cut-off: a complete query result is not proof of complete supply-chain coverage.

Use three documentary results: match, evidence-supported non-match for this selector, and unresolved. Non-match needs positive identifying evidence that fails the stated selector; a blank cell-lot field is not such evidence. Non-match is not general safety clearance. Unknown membership cannot justify narrowing a real safety action. A responsible safety owner may require a broader action population than the records can identify precisely.

Proposed selector buckets with separate possession, notification and remedy evidence.
Original blank coverage model, not measured results. Match, evidenced non-match and unresolved membership stay distinct; possession, notification and remedy are separate dimensions, not quantities to add.

Health Canada’s Section 13 document guidance is dated June 2011 in its body, with page metadata modified 2022-01-06. It distinguishes upstream/downstream trading-party records, as applicable, from retailer supplier, sale-location and sale-period records. It expressly does not describe a universal retailer requirement to record every consumer transaction or identity. The guide is subordinate to legislation, not a current global compliance opinion. Known retail distribution therefore must not be presented as individually identified or reachable consumers.

Reconcile one population at one declared time

The companion coverage/gaps matrix uses one non-overlapping segment per row at one as-of time. Assign each represented physical unit once to a selector bucket and possession/location segment. Preserve a membership manifest for each cohort. Repeated shipments, rework entries and returns are events concerning units, not automatically new units.

Compare the deduplicated population with the sum of its mutually exclusive known segments. Show missing quantities, overlapping memberships and unexplained differences explicitly. Unknown is not zero; keep estimates separate from known-unit counts and state their basis. If current possession is unknown, retain an unknown-location segment instead of counting an earlier distributor transfer and later retailer sale as simultaneous holdings.

The two blank records hold lookup evidence and population reconciliation separately. Link them through case ID, selector revision, snapshot time and manifest references. Restrict populated records to authorized access.

Use exceptions to test the joins

These six additional cases are fictional documentary exercises, not incident findings or instructions to conduct physical operations.

Fictional gapDocumentary responseConclusion withheld
One suspect input spans buildsInclude every verified membershipFirst finished lot is complete coverage
Mixed inputs, missing carton membershipPreserve an unresolved segmentMissing evidence proves non-match
Replacement after original buildRetain both histories and effective membershipLatest BOM proves the earlier build
Retail destinations known, buyers unknownSeparate distribution from consumer reachEvery buyer can be contacted
Identified units, incomplete action evidenceTrack each action dimension separatelyIdentification or return means closure
Safety owner broadens scopeVersion selector; rerun and explain changesEarlier population remains sufficient

Keep identification, communication and remedy separate

The January 2025 edition of 16 CFR §1115.13(d) distinguishes identifiers, quantities, lifecycle dates, possession tiers, distribution, communications and contemplated remedial actions. Its full-report information is qualified as reasonably available and/or applicable, and staff may modify the requirements. This is a historical field-category example—not a statement of current 2026 or worldwide duties, a reporting deadline or a mandatory spreadsheet.

In the proposed matrix, evidence of identity, current possession, notification and remedy stays separate. An email does not prove receipt, a return does not prove an authorized remedy, and neither closes every open question. Assign each gap an owner and evidence reference; preserve corrections and previous snapshots. Necessary authorized safety action or reporting must not wait for perfect records.

For procurement review, discuss a de-identified batch-traceability evidence request: label schema, record keys, one backward lookup and the corresponding forward coverage report. Request boundaries and unresolved joins, not private customer lists or a recall-service guarantee.

These worksheets confer no production, safety or legal authority. This article provides no battery opening, testing, charging, discharging, handling, return-shipping or destruction instructions. Actual reporting, communications, action scope and remedy belong to the responsible qualified parties.

Sources