Which customers got a bad batch when shipping records have gaps?

This method covers chemical lots shipped in drums, IBCs or bulk to industrial customers, traced forward from one suspect lot. It does not replace a recall procedure that a regulator orders for foods, drugs or consumer products, which sets its own records, notices and deadlines.

What records do you need before tracing a bad batch?

Before tracing a bad batch, you need every record that names the suspect lot or the material that went into it: the certificate of analysis (CoA), the batch or production record, the raw material receiving records, the shipping documents and any customer use records.

• CoA: the lot number and the test results released with it. A customer may keep the CoA on file even when the delivery note is lost, which makes it a backup link.

• Batch record: the production run, the raw material lots consumed, and the tanks or lines used.

• Shipping documents: delivery notes, bills of lading, packing lists and invoices with the lot written on them.

• Customer use records: work logs, product use records and site visit notes that show which product was used where.

• Retained samples: the lot's retained sample, for confirming the defect.

The US Food Traceability Rule gives a useful list of what can carry the link. It defines a reference document as a record that may contain key data elements, with types including "bills of lading, purchase orders, advance shipping notices, work orders, invoices, database records, batch logs, production logs" (FDA, 21 CFR Part 1 Subpart S, 2026). The same document types exist in chemical supply chains.

How do you trace a bad batch from lot to customer step by step?

You trace a bad batch from lot to customer in four steps, one action each, starting from the suspect lot and moving downstream one link at a time. GS1 calls this direction tracking, the capability "to locate or follow the path of a particular traceable object downstream" (GS1, 2017).

1. Trace the lot forward. Fix the suspect lot and its production window. Follow each link: raw material lot, production run, finished lot, shipment, customer, use. Write down the record that proves each link.

2. Mark each broken link. Where no record proves the next link, mark the break, the record you expected and why it is missing.

3. Rank customers by exposure. Class each customer as received, not received or unconfirmed, and order them by whether the product is still on site or in use.

4. Confirm the unconfirmed. Contact each unconfirmed customer, ask which lots they hold, and update their class with the answer and its date.

GS1 recommends that each member of a supply chain can at a minimum "trace back to the direct suppliers of traceable objects and to track forward to the direct recipients" (GS1, 2017). For a single lot, that minimum is the whole forward trace.

Which records link a chemical lot to the customers that used it?

The records that link a chemical lot to the customers that used it form a chain of five links, and each link needs its own record: raw material lot to production run, production run to finished lot, finished lot to shipment, shipment to customer, and customer to use. A missing record at any link leaves every customer beyond it unconfirmed. Write each link and its record into the trace table, so a reviewer can check it without reopening the files.

The table below shows which record proves each link and where that link most often breaks in chemical supply.

The link that breaks most often in practice is finished lot to shipment: when the delivery note does not carry the lot number, the shipment proves only that the product arrived, not which lot. A drug distribution record shows what a complete shipping record holds: consignee name and address, date and quantity shipped, and lot or control number (FDA, 21 CFR 211.196, 2026).

How do you mark where the bad batch record breaks?

You mark where the bad batch record breaks by writing, at the exact link that failed, which record you expected, what you found instead and how you will close the gap. Marking the break at the link, not at the customer, shows which customers share one gap and can be confirmed with one question. A break at the finished lot to shipment link, for example, affects every customer who received that product in the window, so one question to the warehouse or carrier may close many rows at once.

The check table below decides each customer's class for the suspect lot and the next action when a record is missing.

The rule that decides most rows is the last one: a customer is not received only when records show they got other lots. A missing record means unconfirmed.

Which customers should you contact first when the shipping record is missing?

When the shipping record is missing, contact first the unconfirmed customers who bought this product during the production window and are still using it, then those likely to hold stock, then those whose last order falls near the edge of the window. Customers confirmed as received are already notified through the confirmed route, so the contact order here is for the unconfirmed.

Ask each one a closed question: which lot numbers of this product they received or hold, with the CoA or container label to prove it. A CoA lot number held by the customer can rebuild the missing shipment link, because the CoA names the lot it was issued for. Record the answer, the date and who gave it.

Order the list by exposure, which is an illustrative ranking, not a rule from any standard:

1. Unconfirmed, product in active use on the line.

2. Unconfirmed, product in stock but not yet used.

3. Unconfirmed, order date near the start or end of the window.

4. Unconfirmed, low volume or a single delivery.

Do not wait for every reply before notifying confirmed customers. The two lists run in parallel, and each answer moves a customer from unconfirmed to received or not received.

When do you have to widen the bad batch scope to the whole production window?

You have to widen the bad batch scope to the whole production window when a link between the suspect lot and its shipments cannot be rebuilt from any record, because no customer in that window can then be shown not to have received it. Kognitos states that if lots cannot be tied to shipments, "the answer defaults to everything produced during the suspected window, because nobody can responsibly certify that the rest is clean" (Kognitos, 2026).

Three conditions usually force the wider scope in chemical production:

• Shared tanks, silos or lines. Material from several runs or supplier lots mixes, so no record separates the suspect lot from its neighbours.

• Blended or reworked lots. Material from the suspect run went into later lots, so those lots join the trace.

• No lot on shipping documents for the whole window, so every shipment in that period becomes a candidate.

Widen by lot, not by product: list every finished lot made in the window, then trace each one forward with the same four steps. ISO 9001:2026 places the control of nonconforming outputs in clause 8.7, separate from traceability in clause 8.5.2 (ISO 9001:2026, contents), so record the scope decision in your nonconformity record as well.

When should quality and regulatory specialists join a bad batch trace?

Quality and regulatory specialists should join a bad batch trace when the defect could harm people or property, when the product falls under a regulated recall regime, or when the trace has to widen to a whole production window. Each of these turns a records exercise into a decision about notices, regulators and liability.

A regulated product changes the clock. For foods on the US Food Traceability List, records must be available to FDA within 24 hours of a request, or within a time FDA agrees to (FDA, 21 CFR Part 1 Subpart S, 2026), once the rule is enforced: Congress directed FDA not to enforce it before 20 July 2028 (FDA, Federal Register, 2026). In the EU, a chemical supplier's revised safety data sheet must go to all recipients supplied in the preceding 12 months (Regulation 1907/2006, Article 31), and that duty depends on the same recipient list the trace builds.

Lubinpla traces the customers one lot reaches through your work logs, product use records and CoA text and gives the scope and contact priority; because lot and shipment have no dedicated fields, records it cannot find stay as gaps marked unconfirmed. Give the specialist the trace table with each break marked.

Common mistakes in bad batch tracing

The most common mistakes in bad batch tracing treat a missing record as a negative answer, or trace in the easier direction. Each one below narrows the scope without evidence or slows the trace.

• Classing a customer as not received because no shipping record names the lot. Missing means unconfirmed.

• Tracing backward from complaints instead of forward from the lot. Complaints show only the customers who noticed; the forward trace finds every customer the lot reached.

• Confusing ingredient lists with lot links. Knowing which raw materials a product contains does not show which lot went into which run.

• Ignoring distributors and warehouses. Their onward shipments are part of the forward trace.

• Forgetting blended and reworked lots made after the suspect run.

• Waiting for all replies before notifying confirmed customers.

• Closing the trace without dating each answer.