Lot/Batch Number (AI 10) – Source TraceabilityGlossary

Lot/Batch Number (AI 10) – Source Traceability

Q: How long can AI (10) be and which characters are allowed?

AI (10) is variable-length up to 20 alphanumeric characters. Because it is variable-length, include an FNC1 separator unless (10) is the last AI in the symbol. Keep the human-readable format simple and consistent.

Q: Should I encode supplier lot or my internal lot?

Scan and store supplier lots at receipt and map them to your internal lots. For outputs you manufacture, encode the internal lot on labels and maintain supplier-to-internal mappings in the eBMR for upstream linkage.

Q: What’s the difference between AI (10) and AI (21)?

AI (10) identifies a batch shared by multiple units; AI (21) is a serial unique to a single unit. Many regulated items carry both along with (01) GTIN and (17) expiry.

Q: Do I need AI (10) on cases and pallets?

Items usually carry (10); logistics units carry (00) SSCC. Include (10) on cases when required by customers and always keep the association in WMS and in ASN/EPCIS messages.

Q: How do I manage lot splits and merges?

Use MES workflows to generate sub-lots on splits and to record all parent lots on merges. Update labels and genealogy automatically; avoid manual editing of lot numbers at printers.

Q: How does AI (10) support FSMA 204 and recall readiness?

AI (10) ties KDEs across CTEs—from receipt to transformation to shipping. With clean (10) in labels, WMS, EPCIS, and eBMR, you can locate affected product and customers in minutes.

This topic is part of the SG Systems Global regulatory & operations glossary.

Updated October 2025 • GS1 AIs, Traceability, KDE & EPCIS • Labeling, QA/RA, Supply Chain, MES/WMS

Application Identifier (AI) 10 encodes the lot/batch number in GS1 barcodes. It is the anchor that stitches together supplier provenance, in‑process records, and downstream logistics for rapid, defensible lot genealogy. In practice, (10) is a variable‑length, up‑to‑20‑character alphanumeric field that identifies the manufacturing lot or batch to which a unit belongs. It is commonly paired with (01) GTIN, (17) expiry, and optional (21) serial for item‑level uniqueness. At the case/pallet level, (10) may appear alongside (00) SSCC to drive compliant shipping and EPCIS eventing. Within regulated manufacturing, (10) binds the physical label to the Batch Manufacturing Record (BMR) and Master Batch Record (MBR), so investigative questions (What went into this? Who touched it? Where is it now?) can be answered in minutes, not days.

“If GTIN says what it is, AI (10) proves where it came from—and how fast you can find every unit like it.”

TL;DR: Encode the manufacturing lot with (10) in every GS1 symbol that needs traceability. Treat it as master data governed under Document Control, generated and consumed by MES/WMS, verified by Label Verification, and persisted in the eBMR. Use consistent schemas, handle splits/merges in real time, and publish EPCIS/EDI with the same lot keys to enable instant recall readiness.

1) What AI (10) Covers—and What It Does Not

Covers: the lot/batch identifier assigned to a defined manufacturing quantity under a single set of conditions, used throughout product life. AI (10) appears in GS1‑128, GS1 DataMatrix, and GS1 QR to connect physical items to digital records: raw‑material receipt lots; in‑process blend lots; intermediate and finished goods; rework lots; and packaging component lots (when relevant). It enables upstream supplier linkage, in‑plant batch‑to‑bin traceability, and downstream distribution visibility.

Does not cover: item‑level uniqueness (serialization uses AI (21)), logistics unit identity (SSCC uses AI (00)), or product identity (GTIN uses AI (01)). AI (10) also does not replace QA release, QC evidence, or legal declarations; it simply keys those records together.

2) Legal, System, and Data Integrity Anchors

Lot identification is embedded in quality requirements across sectors: 21 CFR Part 211 (pharma), 21 CFR Part 117 (food), ISO 13485 (devices), and ISO 9001 (general QMS). Electronic chains that generate and consume (10) must satisfy Part 11/Annex 11 with validated software (CSV), user accountability, and immutable audit trails guided by ALCOA+. For food, FSMA 204’s KDEs and CTEs lean on (10) to tie source, transformation, and shipping events; for devices, (10) complements UDI when lot traceability is required.

3) The Evidence Pack for AI (10) Compliance

Build a reusable “lot dossier” that proves control of (10) from creation to consumption: lot schema and sequence rules; supplier‑lot mapping and transformation (e.g., blend lots, repacks); label templates and verification results; scan parsing tests (variable‑length field with FNC1 when not last); MES/WMS interfaces; rules for lot splits/merges and rework; disposition and QA release evidence; and retention aligned to product life and jurisdiction. Include QC test links (e.g., COA references), NCMR/MRB decisions by lot, and change‑history under Change Control.

4) From Goods Receipt to Shipping—A Standard Path

1) Goods Receipt. Scan supplier labels (usually (01)+(10)+(17)) at Goods Receipt. The WMS validates master data and starts lot genealogy. Place in Quarantine/Hold pending QC/QA review.

2) In‑Process. As materials are weighed/dispensed, capture source lots (scanned (10)) in the eBMR. When creating a blend or intermediate, issue a new internal lot (10) for the output and record full genealogy. Enforce line clearance to prevent cross‑lot carryover.

3) Pack & Label. Generate item/case labels from eBMR context. The label engine must populate AI (10) from the executing lot record, not from operator free‑text. Verify quality and syntax via automated Label Verification.

4) Release & Ship. QA disposition moves inventory to shippable; cases/pallets carry (00) SSCC and inherited (01)/(10)/(17). Publish ASNs via EDI and shipping events via EPCIS using the same identifiers to keep digital and physical aligned.

If any prerequisite fails—printer out of status, label mismatch, missing (10)—block execution and remediate before continuing. Bad lots on labels become expensive problems at customers’ docks.

5) Designing the Lot Schema—A Practical Method

Lot schemas should be human legible and machine robust. Practical design steps:

Scope & rollover: Define uniqueness scope (site vs. enterprise) and rollover rules to avoid collisions.
Structure: Use components with operational meaning (e.g., year/Julian day, line or plant code, sequence). Keep total length ≤20; avoid ambiguous characters in human‑readable print.
External vs. internal: Preserve supplier (10) on inbound; generate internal (10) for outputs and maintain a mapping table in MES for full upstream linkage.
Splits/merges: When splitting, carry forward parent references; when merging, enumerate all consumed source lots in the eBMR and, if space allows, consider embedding a suffix that distinguishes sub‑lots.
FNC1 & variable length: Because (10) is variable‑length, terminate with FNC1 unless it is the last AI in the symbol.

Lock the schema under Document Control with changes governed via Change Control and validated in templates before go‑live.

6) Chain of Custody—Upstream and Downstream

Upstream: Map supplier lots to internal lots at receipt and keep the mapping with QC disposition and COA. For materials with safety implications (e.g., allergens), enforce priority allergen control by lot and storage zone.

In‑plant: Capture (10) scans at weigh/dispense (batch weighing) and maintain complete consumption/production links in the eBMR aligned to S88 procedures/phases. This is the heart of end‑to‑end genealogy.

Downstream: Use (10) with (01)/(17) on items and (00) SSCC on logistics units to power FEFO/FIFO and to accelerate recall drills. Publish events via EPCIS.

7) Data Integrity—Proving Label = Record

Every (10) printed must be reconstructable from controlled data. Store the encoded string, parsed AIs, and source eBMR context with audit trails and signatures per Part 11. Keep labels synchronized to execution—labels should consume the lot from MES, not allow free‑text edits on printers. Manage all artifacts under Record Retention.

8) Sampling, Lab & Cross‑Checks

Traceability is only as good as your evidence. Tie sampling plans to lot risk (statistical sampling) and manage tests in LIMS. Where environmental or identity tests apply, bind results to the executing lot ID. If lab and shop‑floor disagree on lot assignment or results, treat it as a data‑integrity signal and route via Deviation/CAPA.

9) Equipment Status—Printers, Scanners & Verifiers

Labeling is GxP‑relevant when it determines identity/traceability. Keep printers, scanners, and verifiers qualified (IQ/OQ/PQ) and in known calibration status. Block print if status is unknown; capture device IDs in the eBMR for auditability. Use verification both for print quality and AI syntax (FNC1, date mask, check digits on companion AIs).

10) Labels, Claims & Conversion Logic

Make labels consume the same lot the eBMR is using. If multiple lots feed a unit (e.g., allergen advisory), follow your sector rules for disclosure in human‑readable text while keeping the encoded (10) representative of the governing lot. Ensure consistency with Labeling Control and avoid recalculations in printer scripting. When net content AIs (e.g., (310n)) are also present, keep claim logic under Document Control and validated test cases.

11) Warehouse Status & Logistics Units

On scan, the WMS must validate (10) against known lots, enforce storage rules (temperature maps, segregation), and keep inventory in Hold until release. Cases and pallets carry (00) SSCC to bind pick/pack to shipment documentation (EDI ASN). Use dynamic lot allocation to meet customer date/lot requests without manual searches.

12) AI (10) in Daily Control—Fast Flow, Low Error

Gate critical steps on valid (10): material issue, blend close, pack label, and ship. Trend no‑read and mis‑encode rates; fix templates or supplier labels proactively. Keep scheduling aware of lot expiry (17) and coordinate with Pack & Ship to minimize rework. The goal is a boring system: reliable lots, clean scans, zero relabels.

13) Metrics That Demonstrate Control

Scan Success Rate for (01)/(10)/(17) at receipt and pack (first‑pass decode %).
AI (10) Syntax Errors (missing FNC1 termination, illegal characters) per 10k labels.
Supplier Label Conformance (dock rework minutes per ASN; % labels auto‑accepted).
Recall Drill Latency (lot → shipped SSCCs → customers) under recall readiness.
Genealogy Completeness in eBMR (inputs per lot fully captured).
Hold/Release Cycle Time and wrong‑lot picks (should be near zero).

These indicators tie label quality to operational speed and regulatory defensibility.

14) Common Pitfalls & How to Avoid Them

Free‑text lot entry. Disable manual edits; drive (10) from MES/WMS context.
Missing FNC1. Because (10) is variable‑length, it must be FNC1‑terminated unless last; otherwise downstream parsing fails.
Lot collisions. Poor schema/rollover management leads to duplicates across plants—centralize assignment and validation.
Printer logic doing math. Printers should render, not compute; keep logic in validated systems with audit trails.
Label ≠ record. If the label’s (10) diverges from eBMR, the label wins at the dock—prevent divergence by design.
Supplier non‑conformance. Enforce inbound AI standards and penalize chronic offenders; dock rework is a hidden tax.

15) What Belongs in the Lot/AI (10) Dossier

Lot schema and governance; sequence allocation and uniqueness scope; supplier mapping rules; label templates and print/verify history; device status; parsing/scan tests; split/merge and rework SOPs; WMS/MES interface specs; release/disposition policy; retention schedule; and approvals/effective dates under Document Control. Include links to BMR/MBR, Data Integrity, and Record Retention policies.

16) How This Fits with V5 by SG Systems Global

Lot Master Data. In the V5 platform, (10) is a first‑class object: schema, sequence, and uniqueness are versioned and effective‑dated with approvals in the audit trail. Inbound scans at Goods Receipt create supplier‑to‑internal lot mappings and place stock in Hold pending QA.

Execution & Interlocks. V5 blocks weigh/dispense and pack steps unless a valid lot context exists. Label generation pulls (10) from eBMR; print is blocked if a device is out of status or if syntax checks fail (FNC1, length). Splits and merges are guided with automatic genealogy updates.

Traceability & Shipping. V5 emits EPCIS events and builds EDI ASNs with the same (10)/(01)/(00) you printed. Dashboards show scan quality, supplier conformance, and recall readiness—from lot to shipped SSCCs—in one click.

Bottom line: V5 operationalizes AI (10) across labeling, eBMR, WMS, and shipping, shrinking recall latency and eliminating hand‑keyed errors.

17) FAQ

Q1. How long can AI (10) be and which characters are allowed?
AI (10) is variable‑length up to 20 alphanumeric characters. Because it is variable‑length, include an FNC1 separator unless (10) is the last AI in the symbol. Keep your human‑readable format simple and consistent.

Q2. Should I encode supplier lot or my internal lot?
At receipt, scan the supplier lot and map it to your internal lot. On outputs you create (intermediates/finished goods), encode the internal lot and maintain full mapping in the eBMR for upstream linkage.

Q3. What’s the difference between AI (10) and AI (21)?
(10) identifies the batch that multiple units may share; (21) is a serial unique to a single unit. Many regulated items carry both: (01) GTIN + (10) lot + (17) expiry + (21) serial.

Q4. Do I need AI (10) on cases and pallets?
Items typically carry (10); logistics units carry (00) SSCC. If cases are homogeneous and customers require it, include (10) on the case. Always keep the association in WMS and in the ASN/EPCIS payload.

Q5. How do I manage lot splits and merges?
Use MES workflows to generate sub‑lots on splits and to record all parent lots on merges. Update labels and genealogy automatically; never hand‑edit lot numbers at printers.

Q6. How does AI (10) support FSMA 204 and recall readiness?
(10) is the key that ties KDEs across CTEs—from receipt to transformation to ship. With clean (10) in labels, WMS, EPCIS, and eBMR, you can find affected product and customers in minutes.