Application Identifier (AI) – GS1 Data Elements

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

Updated October 2025 • GS1 Barcodes, Serialization & Traceability • Labeling, QA/RA, Supply Chain, MES/WMS

Application Identifiers (AIs) are the GS1 “data tags” that give meaning to the numbers and characters inside a barcode or 2D symbol. They tell scanners and systems what each piece of data is—GTIN, lot, expiry, serial, net weight, quantity, SSCC, and more—so labels can drive automated receiving, lot genealogy, and compliant global trade. Typical item‑level strings include (01) GTIN + (10) lot + (17) expiry + (21) serial; logistics labels center on (00) SSCC for case/pallets. In regulated sectors, AIs underpin UDI (with 21 CFR Part 830 for US devices) and fast, accurate recalls and traceability. Operationally, the rule is simple: if your AIs are wrong, your supply chain is blind.

“Barcodes without correct AIs are just stripes. With AIs, they become instructions your factory, warehouse, and regulators can trust.”

1) What AIs Cover—and What They Do Not

Covers: encoded data semantics across GS1 symbols (GS1‑128, GS1 DataMatrix, GS1 QR). AIs define product identity (GTIN), batch/lot (10), expiry (17), serial (21), quantities (30/37), variable measure (310n/320n), price (392x), locations (414 GLN), logistics unit (SSCC 00), and event hooks for EPCIS. They support in‑line verification, Goods Receipt, pick/pack, and compliant shipping.

Does not cover: inventing proprietary “meaning” inside a GS1 symbol or bypassing regulatory structure. AIs don’t excuse bad master data, sloppy Document Control, or broken retention. They also aren’t a substitute for QA release or QC evidence—only the data carrier that connects label to record.

2) Legal, System, and Data Integrity Anchors

AIs sit within a broader compliance frame: medical devices require UDI per 21 CFR Part 830; food traceability references structured Key Data Elements (FSMA 204) and interoperable eventing via EPCIS; cosmetics and OTCs demand clear identity and expiry control. Electronic evidence chains must meet Part 11/Annex 11 (unique users, signatures, time‑stamps, audit trails) and ALCOA+ data integrity. Labels and strings live under Document Control with validated generation in MES/WMS and integration to EDI (e.g., ASN) where required.

3) The Evidence Pack for AI Compliance

A solid “AI pack” includes: GTIN/packaging hierarchies; allocation/uniqueness rules for serials; AI usage matrix by level (unit, case, pallet); print templates and business rules; verification plans and results; scan test scripts (variable‑length fields, FNC1 separators); interface specs for EPCIS/EDI; error handling and reprint workflow; and retention of label/evidence under Record Retention. For devices, include UDI device package configuration and GUDID/UDIR submission evidence; for food, show FSMA 204 KDE mapping to the encoded AIs.

4) From Item Master to Scannable Label—A Standard Path

1) Master Data. Define product identity (GTIN), pack hierarchy (unit/inner/case/pallet), target AIs per level, and required variable data (lot, expiry, weight, serial). Govern in a controlled repository (master recipes often reference these values for label pulls).

2) Template Authoring. Build label templates with AI‑aware fields and validation. Include human‑readable text, correct date masks (YYMMDD for (17)), and check digit logic for GTIN and SSCC.

3) Print & Verify. Generate at source with the MES/WMS, capturing serialized identifiers and AI values into the eBMR/WMS. Verify print quality and syntax via Label Verification.

4) Scan & Post. At Goods Receipt, scan AIs to post put‑away and start lot genealogy. Enforce Quarantine/Hold until QA release; then pick/pack and ship with case/pallet SSCCs feeding compliant fulfillment and EPCIS events.

If a template, AI rule, or printer status fails, block execution and force correction—bad labels cost more than downtime.

5) Designing the AI String—A Practical Method

Start from downstream needs and work backward. For sellable units, encode identity and safety first: (01) GTIN + (17) expiry + (10) lot; add (21) serial for item‑level traceability/UDI. For logistics: (00) SSCC and optional counts (37). For variable‑measure products, use (310n) net weight in kg (set decimal precision by n) or (320n) in lb; for units‑per‑pack, (30). Terminate variable‑length fields (e.g., (10), (21)) with FNC1 unless they are last in the string. Keep human‑readable parentheses, but remember they are not encoded literally in GS1‑128/GS1 DataMatrix. Validate dates (YYMMDD) and check digits (mod 10) in the label engine; never rely on operators to fix math at a printer.

6) Variable‑Weight & Net Content AIs

For variable‑weight products controlled by scales and checkweighers, tie the encoded AIs to the executed net. When you claim mass/weight on label, drive (310n) directly from gravimetric measures; if you claim volume but fill by mass, ensure conversions are governed under UOM consistency and density logic within MES. Align AI content with Label Verification and downstream WMS rules to avoid mismatches between encoded weight and transactional weight. If you use guardbanded fills for regulatory reasons (e.g., TNE), ensure only the declared value is exposed in the AI where required, and keep executed nets in the eBMR for release and audit.

7) Data Integrity—Proving Label = Record

Everything on the label must be reconstructable from controlled records. Store the generated AI string, parsed elements, and source data (GTIN, lot, expiry, weights, serials) with audit trails and signatures per Part 11. Avoid free‑text edits; bind label content to master data and eBMR values. If a downstream system (ERP, EDI) disagrees with the label, the label loses—because scans at Goods Receipt set inventory reality. Don’t let “print shop” shortcuts undermine release or recall readiness.

8) Sampling, Print Quality & Scanner Cross‑Checks

Define inspection that covers both symbol quality and AI syntax. Use verifiers to score print quality and decodability, but also parse strings programmatically to catch missing FNC1s, wrong date masks, or invalid check digits. Cross‑check scanners against master data (does (01) match the SKU, does (17) fall within shelf life policies, does (10) exist). Route persistent issues through Deviation/CAPA and improve templates under Change Control.

9) Equipment Status—Printers, Verifiers, and Scanners

Printers and vision/verifier systems are GxP‑relevant when labels affect release. Keep devices qualified (IQ/OQ/PQ) and in known calibration status. Enforce warm‑up, media specs, and preventive maintenance; block label print steps in the eBMR if a device goes out of status. If scanners can’t read at the dock, expect real operational pain and customer complaints—fix the source, not the warehouse.

10) Labels, Claims & Conversion Logic

Labels must consume the same values your systems rely on. Pull lot/expiry from the eBMR; pull GTIN/pack levels from master data; ensure serialization rules match business logic (uniqueness scope, lifetime). For variable measure, apply consistent UOM and rounding (UOM); document conversions and date rules under Document Control. If labels re‑calculate what MES already computed, you’ve created a second truth—don’t.

11) Warehouse Status & Logistics Units

Use AIs to make the warehouse smarter. Cases/pallets carry (00) SSCC; items carry (01)/(10)/(17)/(21). On scan, the WMS should validate identity, lot, and shelf life, enforce Hold until QA release, and guide FEFO/FIFO picking with accurate locations (bin/zone topology). Publish shipping events via EPCIS and exchange EDI ASNs that reference the same SSCCs your labels carry. No SSCC = no visibility after the dock door.

12) AIs in Daily Operations—Keep Flow High and Errors Low

Use scans to prevent mistakes, not record them. Gate critical steps—material issue, mixing/dispensing, pack/label, ship—on valid AIs that match the eBMR/WMS context. Trend no‑reads and mis‑encodes by line/supplier; work upstream. Establish tight verification triggers when mis‑encode rates rise. The goal is a boring label system: consistent AIs, clean scans, zero rework.

13) Metrics That Demonstrate Control

• Scan Success Rate at receipt/picking (first‑pass decode %).
• AI Syntax Error Rate (bad FNC1, date mask, check digit) by template and site.
• No‑Read Incidents per 10k labels and reprint rate.
• UDI/EPCIS Consistency (label vs. submission/event payload match).
• Recall Drill Latency from GTIN/lot to shipped SSCCs (recall readiness).
• Supplier Label Conformance (dock rework minutes per ASN).

If these indicators are red, your AIs are costing margin and credibility—fix the templates and the governance, not just the scanners.

14) Common Pitfalls & How to Avoid Them

• Missing FNC1 separators. Variable‑length AIs (e.g., (10), (21)) must be terminated unless last; otherwise downstream parsing fails.
• Wrong date format. (17) is YYMMDD—do not print local masks in the barcode.
• Check digit mistakes. Compute and validate mod‑10 for (01) and (00) in the label engine.
• GTIN/SKU mismatches. Keep GTIN assignments under Document Control and sync to MES/WMS before print.
• Recalculating at the printer. Print devices should render, not compute; feed them validated values from MES/WMS.
• Supplier non‑conformance. Enforce inbound AI standards and push back on vendors who ship “generic” labels—dock rework is a tax on your time.

15) What Belongs in the AI/Label Dossier

Include GTIN assignments and pack hierarchies, SSCC rules, serial generation/uniqueness policy, AI usage matrix, print templates and logic, verifier specs and routine results, scanner parsing tests, EPCIS/EDI interface specs, exception handling/reprint SOPs, and approvals/effective dates under Document Control. Retain label images, parsed payloads, and scan logs per retention rules. If you can’t prove what was encoded, you can’t defend what was shipped.

16) How This Fits with V5 by SG Systems Global

AI Master Data. In the V5 platform, GTINs, pack levels, and AI rules are versioned, effective‑dated objects with approvals captured in the audit trail. Serialization policies (unique unit identification) define scopes and ranges.

Execution & Interlocks. V5 generates AI strings from the eBMR/WMS context and blocks print if a device is out of calibration status, if GTIN/pack hierarchy is missing, or if syntax fails validation. Scans at Goods Receipt post lots and locations automatically and enforce Hold until QA release.

Traceability & Shipping. V5 emits EPCIS events bound to the same AIs your labels carry and builds EDI ASNs using those identifiers—so customers receive precisely what you encoded. Dashboards show scan quality KPIs and supplier conformance in real time.

Bottom line: V5 turns GS1 AIs into operational leverage—accurate labels, clean scans, instant traceability, and faster release with fewer chargebacks.

17) FAQ

Q1. What’s the difference between a GTIN, an AI, and UDI?
The GTIN is the product identifier (often encoded with AI (01)). The AI is the prefix that defines each data element’s meaning (e.g., (10) lot, (17) expiry). UDI is a regulatory construct that typically uses GS1 AIs (e.g., (01)+(17)+(10)+(21)) plus database submissions (e.g., GUDID) for medical devices.

Q2. Which AIs are “must‑have” on sellable units?
It depends on your sector and customers, but (01) GTIN, (10) lot, and (17) expiry are common minimums; add (21) serial for item‑level traceability/UDI. Cases/pallets generally require (00) SSCC.

Q3. How do I handle variable‑length fields like (10) and (21)?
Terminate with FNC1 unless the field is last in the symbol. Set maximum lengths in your template, and validate at print time to avoid truncation or parsing failures at the dock.

Q4. When should I use (310n) versus (320n) for weight?
Use (310n) for kilograms (SI), (320n) for pounds. Choose the decimal precision n to match your process capability and labeling claim, governed under UOM consistency.

Q5. How do I guarantee serial uniqueness?
Govern serialization under Unique Unit Identification with clear scope (per GTIN, per site, or global), collision checks before print, and reconciliation at pack/ship. Never generate serials in spreadsheets.

Q6. How do AIs relate to EPCIS and ASNs?
The same identifiers you encode—GTIN, lot, serial, SSCC—become the keys in EPCIS events and the line items of your EDI Advance Shipping Notice. If the label is wrong, your digital messages will be wrong too.

Related Reading
• Identifiers & Logistics: GS1 GTIN | SSCC | EPCIS | Serialization
• Labeling & Systems: Label Verification | WMS | MES | EDI
• Quality & Governance: Document Control | Audit Trail | Record Retention | Quality Control | QA