UPC (Universal Product Code)

A UPC-A barcode encodes 12 digits. Read left to right, they break into three parts.

Part	Digits	What it means
GS1 Company Prefix	First 6-10	Licensed to the brand owner by GS1. Identifies the company, not the product.
Item reference	Digits 11 minus prefix length	Assigned by the brand owner to one specific sellable item.
Check digit	Last 1	A mod-10 calculation over the first 11 digits.

The check digit exists to catch scanner misreads and keying errors. If someone types 012345679905 instead of 012345678905, the math returns 2 where the code claims 5, and the record can be rejected before it enters the catalog. Mod-10 catches every single-digit error, but it does not catch every transposition: some swapped-digit pairs still compute a matching check digit. Any validation rule you write should recalculate the check digit rather than trust the source file.

You will also see UPC-E, an eight-digit compressed symbol carrying six data digits, printed on small packages: a tube of thread sealant, a single automotive fuse. UPC-E is not a different number. It is a UPC-A with runs of zeros suppressed to save print width, and it expands back to the full 12 digits. Store the expanded UPC-A. Never store the compressed print form as the identifier of record.

This is the question behind most identifier confusion, and the answer is simpler than the vocabulary suggests.

Identifier	Length	Who assigns it	Where you see it
UPC-A (GTIN-12)	12 digits	Brand owner, from a GS1 prefix	US and Canadian retail packaging
EAN-13 (GTIN-13)	13 digits	Brand owner, from a GS1 prefix	Europe, Asia, most of the world
GTIN-14	14 digits	Brand owner	Case and inner-pack labels
ASIN	10 characters	Amazon	Amazon listings only
MPN	No fixed format	The manufacturer	Spec sheets, cut sheets, distributor catalogs

UPC and GTIN are not competing standards. UPC-A is GTIN-12. EAN-13 is GTIN-13. GTIN is the umbrella term GS1 uses for the whole family, and every member can be expressed at 14 digits by left-padding with zeros.

• A UPC of 012345678905 stored as GTIN-13 is 0012345678905.
• The same UPC stored as GTIN-14 is 00012345678905.
• Same product, same number, different field width.

That padding rule is why a European trading partner asking for an EAN and a US retailer asking for a UPC are usually asking for the same digits.

UPC data almost never fails because the standard is hard. It fails because of handling.

• Leading zeros vanish. A spreadsheet reads 012345678905 as a number and stores 12345678905. The code is now 11 digits and fails a length check. It is recoverable by re-padding to 12, but only if you know a zero was dropped.
• Scientific notation. Longer codes display as 8.71234E+11. In the workbook this is a display format, not damage: precision survives to 15 significant digits, so a 12- or 14-digit value is still intact underneath. The corruption happens on round-trip. Export that cell to CSV and re-import it, and the rounded display string is what lands in the file.
• The wrong code in the field. Vendors drop the ITF-14 case code into the UPC column, or send the compressed UPC-E.
• Reuse. A brand retires a SKU and reassigns its number to something else. GS1 sets waiting periods for exactly this reason, and not everyone honors them.
• One UPC across variants. A supplier ships the same UPC for a 3/8-16 Grade 8 hex bolt in a 25-count box and a 100-count box. Retailers reject the second listing as a duplicate.
• Invented numbers. Internal SKUs padded to 12 digits, or codes using prefixes reserved for in-store and variable-weight use. They scan at the shelf and fail everywhere else.

A UPC tells two systems that they are looking at the same product. It tells neither system what the product is.

012345678905 might be a 3/8-16 x 2 in. Grade 8 hex bolt, zinc-plated, sold 50 to a box. The number carries none of that. Not the thread pitch. Not the grade. Not the finish. Not the box quantity. Not the proof load.

That gap is where catalogs stall. A distributor can have a validated, GS1-legitimate UPC on every SKU in the file and still have most of the attribute columns empty. Matching by UPC is the easy half. The hard half is everything the match does not hand you.

Your PIM stores the UPC and the empty attribute fields next to it. Anglera does the work of finding the thread pitch, the grade, the finish, and the pack count in the manufacturer's spec sheet and filling those fields, so the identifier and the content land together instead of years apart.

Six habits cover almost every failure mode above.

• Store it as text, fixed width. Never a numeric column, in the PIM, the ERP, or the export.
• Recalculate the check digit on every ingest, not once at initial load. Vendor feeds change.
• Normalize to GTIN-14 internally. Render back down to 12 digits for US channels that want a UPC. One internal shape, many outbound shapes.
• One UPC per sellable variant. Parent and grouping records do not get one.
• Keep the case GTIN in its own field, separate from the each-level UPC. They are different products to a trading partner.
• Record the source and the date. When two systems disagree, provenance is the only thing that settles it.