Industrial procurement has long relied on documents: material certificates, inspection reports, declarations, audit reports and certificates of origin. These remain useful. The change is that customers and regulators increasingly ask how each document connects to a specific product and supply-chain event.
A certificate can be authentic and still be irrelevant to the delivered batch. A supplier can be approved while an undisclosed sub-tier performs the critical process. A database can be tamper-resistant while storing an unsupported input. The emerging requirement is not simply “more documents”; it is a more granular evidence chain.
What an evidence chain contains
A practical manufacturing evidence chain connects four elements:
- Identity: product, part, model, revision, material lot, facility and legal entity.
- Transaction: purchase, receipt, transfer, subcontract and shipment records.
- Transformation: what process converted the input, where, when and under which specification.
- Claim or result: composition, conformity, origin, emissions, test result or other statement supported by records.
Each link needs a scope. Does a certificate cover one heat, one furnace load, one batch, one model or an entire facility for a period? Without scope, readers may apply evidence more broadly than it supports.
Chain of custody is not one universal model
ISO 22095:2020 establishes general terminology and models for chain of custody across materials and products. ISO — ISO 22095:2020 In March 2026, ISO announced new standards ISO 22095-2 and ISO 22095-3 concerning mass balance and book-and-claim models. ISO — New standards bring clarity to chain of custody
These references underline an important procurement distinction: physical segregation, controlled mixing, mass balance and certificate trading do not make the same claim. Buyers should identify which chain-of-custody model a scheme uses and whether that model is acceptable for the legal, customer or engineering purpose.
A mass-balance claim, for example, should not be described as proof that a specific physical unit contains a particular source unless the applicable model supports that statement. Conversely, physical unit tracing may be unnecessary where a legitimate programme expressly uses another model.
Traceability is moving from facility-level to product-level questions
Traditional supplier qualification often asks whether a factory has a certified management system. Newer questions may ask which facility made the item, which material batch was used, what transformation occurred, who supplied the input and which evidence supports a sustainability or origin claim.
Regulatory developments such as CBAM, forced-labour controls and Digital Product Passports illustrate different reasons for this shift. They should not be merged into one compliance checklist: each has its own scope, actors, evidence rules and dates. Their common operational need is the ability to connect upstream information to the product or transaction.
NIST's IR 8536, *Supply Chain Traceability: Manufacturing Meta-Framework*, was issued as a second public draft on 31 July 2025. NIST describes it as a cross-sector manufacturing traceability meta-framework; its draft status must remain visible and it should not be presented as a binding standard. NIST — IR 8536 Second Public Draft
Technology follows data discipline
Barcodes, QR codes, ERP systems, shared platforms and distributed ledgers can improve retrieval and integrity. They cannot establish whether the initial data was accurate or whether a supplier omitted an operation.
NIST's publication on blockchain and related technologies for manufacturing traceability examines technical and non-technical challenges and emphasises the need to understand traceability objectives and system context. NIST — IR 8419 overview
Before buying a platform, define:
- what question the traceability system must answer;
- the smallest required traceable unit;
- the events and transformations to record;
- who creates, checks and corrects each field;
- evidence-retention and access rules;
- how identifiers survive processing and repacking; and
- how data from different suppliers will interoperate.
If these rules are unclear, technology digitises ambiguity.
Design traceability according to consequence
Full unit-level tracing can be expensive or physically impossible for some bulk materials. A risk-based design selects the required granularity: unit, batch, heat, load, purchase order, facility or reporting period.
The selection should consider safety, regulatory obligation, customer contract, material mixing, process variability and the cost of containment if something fails. A low-risk standard item may need supplier and lot records. A safety-critical custom part may need material heat, serial identity, special-process load and individual inspection results.
State where traceability legitimately ends. An honest limitation is more useful than an unsupported “fully traceable” claim.
Test the chain with retrieval and reconciliation
Choose a delivered item and trace backward to material and processors, then choose an incoming lot and trace forward to finished shipments. Reconcile quantities, identities and dates. Record missing links and the time required.
This bidirectional test reveals common weaknesses:
- one certificate reused across unrelated batches;
- labels removed during cleaning or finishing;
- supplier names changed between local and English records;
- subcontract operations absent from the route;
- mixed inventory with no allocation method;
- spreadsheet claims with no source method; and
- valid records that cannot be retrieved within a decision deadline.
The corrective action may be better identifiers, controlled containers, document indexes, supplier notification, retained source records or a narrower claim. Blockchain is not the default answer.
Procurement checklist
- What business, quality or regulatory question must traceability answer?
- What is the required traceable unit?
- Are product, revision, batch, facility and legal-entity identities controlled?
- Are transactions and transformations recorded, not only final certificates?
- Does every claim identify its evidence, method, scope and date?
- Is the chain-of-custody model stated accurately?
- Are quantities and yields reconciled across transfers?
- Can identifiers survive subcontract processes?
- Are corrections and superseded records preserved?
- Can the chain be traced backward and forward within the required time?
- Are confidentiality and access rights controlled?
- Are limitations disclosed instead of hidden behind “fully traceable” language?
What this means for procurement
Supplier evaluation is moving from document possession to evidence connectivity. Buyers should not abandon certificates; they should ask what each certificate proves and how it connects to the delivered item. A modest, well-tested lot-level system is often more credible than a sophisticated dashboard with unclear inputs.
Morning Sunlight Asia can help map China-side suppliers, processes, lots and documents into an order-level evidence chain. Discuss Your Requirement to define the traceability question and the practical granularity before selecting tools.
證書可以真實卻不屬於交付批次；供應商可以通過審核，關鍵工序卻由未披露次級供應商完成；資料庫可以難以竄改，輸入仍可能沒有證據。變化不是單純需要更多文件，而是需要更細的連接。
證據鏈的四個元素
- 身份：產品、零件、型號、版本、材料批次、設施與法人。
- 交易：採購、收料、轉移、委外和出貨記錄。
- 轉換：哪個工序在何地、何時、按什麼要求改變投入。
- 聲明或結果：成分、符合性、原產、排放或測試數值。
每個證據都要說明範圍：一個爐號、一個裝爐批、一個型號，還是某設施的一段期間。沒有範圍，使用者容易把證據擴大到它不能支持的對象。
監管鏈不是單一模型
ISO 22095:2020 建立材料與產品監管鏈的一般術語和模型。ISO — ISO 22095:2020 2026 年 3 月，ISO 宣布新的 ISO 22095-2 和 ISO 22095-3，涉及 mass balance 與 book-and-claim 模型。ISO — Chain of custody
實體隔離、受控混合、質量平衡與證書交易支持的聲明不同。採購方要確認某制度使用哪種模型，以及它是否適合具體法律、客戶或工程目的。不能把質量平衡聲明誤寫成每件實物都來自特定來源。
問題正從工廠層走向產品層
傳統供應商評估常問工廠有沒有管理體系認證；新的問題則可能問哪個設施製造、哪個材料批次被使用、發生哪些轉換、投入由誰提供，以及哪份證據支持永續或原產聲明。
CBAM、強迫勞動管制和 DPP 有不同法律範圍、角色、日期與證據規則，不能合併成一張通用清單；它們共同需要的是把上游資訊連到具體產品或交易。
NIST IR 8536《Supply Chain Traceability: Manufacturing Meta-Framework》於 2025 年 7 月 31 日發布第二次公開草案。它是跨行業製造追溯框架草案，不能被描述為強制標準。NIST — IR 8536 Second Public Draft
技術必須跟在資料規則之後
條碼、QR code、ERP、共享平台和分散式帳本可以改善檢索與完整性，卻不能證明初始資料正確，也不能發現被刻意遺漏的工序。
選平台前先定義要回答的問題、最小追溯單位、需要記錄的事件、每個字段由誰建立與更正、保存和權限、身份如何穿過加工與重新包裝，以及不同供應商資料如何互通。規則不清楚，技術只會把模糊數位化。
按後果選擇粒度
逐件追溯對某些散裝材料成本過高或不可行。應依安全、法規、合約、混料、製程變差和失敗後隔離成本，選擇逐件、批次、爐號、裝爐、訂單、設施或期間。誠實說明追溯終點，比無證據宣稱「完全可追溯」更可信。
用雙向取證測試
選一件交付品向後追到材料和加工方，再從一批來料向前追到成品出貨，核對數量、身份和日期。這能暴露證書重複使用、標識在表面處理中消失、次級工序未記錄、混合庫存沒有分配方法，或有效文件無法在時限內取回等問題。
採購清單
- 追溯要回答什麼品質、商務或法規問題？
- 最小追溯單位是什麼？
- 產品、版本、批次、設施和法人身份是否受控？
- 是否記錄交易與轉換，而不只保存最終證書？
- 每項聲明是否有證據、方法、範圍和日期？
- 監管鏈模型是否準確表述？
- 各層數量與良率能否核對？
- 身份能否穿過外協工序？
- 更正和被替代記錄是否保存？
- 能否在時限內雙向追溯？
- 是否清楚披露限制？
對採購決策的意義
供應商評估正在從「擁有文件」轉向「證據可連接」。採購方不應放棄證書，而要問每份證書證明什麼，又如何連到交付品。一套經過測試的批次級系統，通常比輸入不清楚的華麗儀表板更可信。
晨陽亞洲可協助把中國境內供應商、工序、批次和文件整理為訂單級證據鏈。歡迎 Discuss Your Requirement，先定義追溯問題和實際粒度，再選擇工具。