EDIFACT Explained: Format, Envelope and X12 Mapping

If you already run X12 with US retailers, EDIFACT will feel familiar within an hour. The documents are the same ones you already exchange: purchase order, acknowledgment, ship notice, invoice. What changes is the punctuation, the envelope, the names, and the way versions are labelled.

This guide covers the history, the syntax, the envelope, how to read an interchange step by step, directory versions, the GS1 EANCOM subset most retail partners use, every message family, and a detailed comparison with X12 and with the UK's older TRADACOMS standard. The live example below is a complete, valid interchange you can open in the free EDIFACT inspector.

What is EDIFACT?

UN/EDIFACT (United Nations/Electronic Data Interchange for Administration, Commerce and Transport) is an international EDI standard developed under the United Nations and published by UNECE. It defines three things: syntax rules for structuring data, a library of standard messages (ORDERS, INVOIC, DESADV and about two hundred more), and the data elements and code lists those messages draw on.

In practice, "EDIFACT" covers three layers:

When a UK grocer or a European distributor says "we send EDIFACT", they almost always mean a specific subset plus their own implementation guide on top of it, the same way Walmart's 850 is X12 plus Walmart's rules.

A short history of EDIFACT

Maintenance and development sit with UN/CEFACT, the UN Centre for Trade Facilitation and Electronic Business, under UNECE. New directories are still published, but retail and logistics partners mostly stay on the older ones their systems were built for.

How to read an EDIFACT interchange, step by step

Faced with a raw file from a new partner, work through it in this order. The live example further down follows the same steps.

1. Find the separators. If the file starts with UNA, the next six characters declare the separators. If not, assume the defaults: : + . ? space '.
2. Read the UNB. It tells you the character set (UNOA, UNOC), syntax version, sender, recipient, date and the interchange control reference.
3. Split the messages. Each UNH starts a message and names its type and version, for example ORDERS:D:01B:UN:EAN010. Each UNT ends it.
4. Read the BGM. It gives the document code (220 order, 380 invoice, 351 despatch advice), the document number, and whether it is an original (9), a replacement (5) or a cancellation (1).
5. Decode the qualifiers. Most segments are generic. DTM+137 is the document date, DTM+2 the requested delivery date. NAD+BY is the buyer, NAD+SU the supplier, NAD+DP the delivery point.
6. Walk the lines. Each LIN starts an item. The segments after it (PIA, IMD, QTY, PRI, MOA, TAX) belong to that line until the next LIN or the UNS section control.
7. Check the controls. CNT totals in the summary, the segment count in UNT, and the message count and reference in UNZ must all match what is in the file.

EDIFACT syntax: segments, elements and separators

An EDIFACT file is plain text made of segments, each starting with a three-letter tag. Segments contain data elements, and some data elements are composites made of several component elements.

NAD + BY + 5012345000008 :  : 9 '
 |     |         |        |  |  |  |
 |     |         |        |  |  |  segment terminator
 |     |         |        |  |  code list agency: 9 = GS1
 |     |         |        |  empty component (code list)
 |     |         |        component separator
 |     |         party identifier: the buyer's GLN
 |     party qualifier: BY = buyer
 segment tag: NAD = name and address

The same pattern, a tag followed by a qualifier and a value, runs through the whole standard:

DTM + 137 : 20261011 : 102 '
      |     |          |
      |     |          format: 102 = CCYYMMDD
      |     the date
      qualifier: 137 = document date

The service characters

The UNA service string advice

A file can open with a UNA segment that declares which characters it uses. It is always exactly nine characters: UNA plus six service characters in a fixed order.

UNA:+.? '

Position by position: : component separator, + element separator, . decimal mark, ? release character, a space (reserved in syntax version 3), and ' segment terminator. GS1 adds that the same character must not be used in more than one position.

UNA is optional when a file uses the defaults with character set level A. The EANCOM manual says it must be sent when partners use a character set from B to F with the default separators, which covers the common UNOC case in European retail. In syntax version 4, the reserved fifth position becomes the repetition separator.

The release character

The release character ? lets a separator appear inside data. A product description of "Salt + Pepper Mill Set" has to be written:

IMD+F++:::SALT ?+ PEPPER MILL SET'

Without the ?, a parser would split the description at the +. The same applies to :, ' and ? itself (?? is a literal question mark). Unescaped apostrophes in names such as "O'Neill" are the most common cause of broken EDIFACT files in UK trade, because the ' ends the segment mid-name.

Empty and missing data

The EANCOM manual sets out how empty data is left out:

• Empty conditional segments are omitted entirely, tag and all.
• An empty element in the middle keeps its separator, so position is preserved: NAD+BY+5012345000008::9' has an empty code list component between the two colons.
• Trailing empty elements are truncated. The segment ends at the last element with data. Parsers must not assume a fixed element count.
• Line breaks are not part of the format. Real files are often one continuous line. Examples on this page put one segment per line for readability, and the inspector accepts both.

Numbers

EANCOM uses the point as the decimal mark, with at least one digit either side when a decimal is sent. Thousands separators are not allowed: 2500000, never 2,500,000. Values are treated as positive. A credit note carries positive amounts and the document code tells the receiver to treat them as negative. Where a negative value is needed, a minus sign goes immediately before the number.

The underlying UN/EDIFACT syntax versions 1 to 3 define the comma as the default decimal mark, which is why UNA declares it explicitly. If your partner sends UNA:+.? ', the dot is the decimal mark and 1406.40 means what it looks like.

Character sets

The first part of the UNB syntax identifier names the character set. GS1 recommends level A for EANCOM and asks partners to agree anything else first.

The EDIFACT envelope: UNB, UNG, UNH

EDIFACT nests data in three levels, like X12. The interchange is the outer envelope, an optional functional group sits inside it, and each business document is a message.

UNA                      Service string advice (optional)
UNB                      Interchange header
  UNG                    Functional group header (optional, rarely used)
    UNH                  Message header
      BGM, DTM, NAD, LIN ...   the business document
    UNT                  Message trailer: segment count
  UNE                    Functional group trailer
UNZ                      Interchange trailer: message count

UNB: interchange header

UNB+UNOC:3+5012345000008:14+5098765000002:14+261011:0915+HLS000451++ORDERS'

UNH and UNT: message header and trailer

UNH+1+ORDERS:D:01B:UN:EAN010'
...
UNT+26+1'

The UNH carries a message reference (1) and the message identifier: type ORDERS, version D (the code for a UN/EDIFACT directory), release 01B, controlling agency UN, and an association code. EAN010 identifies the EANCOM 2002 version of ORDERS. UNT carries the number of segments in the message including UNH and UNT themselves, plus the same message reference.

UNZ: interchange trailer

UNZ+1+HLS000451'

The count of messages (or groups, if UNG is used) and the control reference, which must match the UNB. A mismatch here, or a wrong UNT count, is the most common reason a partner rejects an interchange outright.

UNG and UNE

UNG/UNE functional groups exist in the standard but are optional and uncommon in retail EDIFACT. Most partners send messages directly inside UNB/UNZ. This is a real difference from X12, where GS/GE is mandatory.

A complete EDIFACT interchange

This is a valid EANCOM purchase order from the fictional UK retailer Harbour Lane Stores to the fictional supplier Kestrel Valley Homeware, with three order lines. GLNs and GTINs are made up, with valid GS1 check digits.

Read it with the seven steps above. The separators are the defaults, declared in UNA. The UNB says Latin-1, syntax 3, from one GLN to another. The single UNH is an EANCOM 2002 ORDERS. BGM+220 is an original purchase order, DTM+137 its date and DTM+2 the delivery date requested. Three LIN groups each carry a GTIN, the supplier's own code (PIA+1), a description, the ordered quantity (QTY+21) and a net price. CNT+2:3 confirms three lines, UNT+26 counts 26 segments, and UNZ+1 closes one message with the same reference as the UNB.

The EDIFACT ORDERS guide walks through every segment in this message.

EDIFACT directories: D96A, D01B and the rest

EDIFACT messages are versioned by directory. A directory is a dated release of every message, segment, data element and code list, named by year and release letter: D.96A, D.01B, D.21A.

What this means in practice:

• Your partner picks the directory. It appears in every UNH. You map to what they send, not to the newest release.
• Old directories stay in production for decades. D.96A and D.01B are the two you will see most in retail, because the two main EANCOM editions are built on them.
• Differences between directories are usually small for core messages, but code lists and segment groups do move. A map built for D.96A ORDERS will not always read D.01B cleanly.

EANCOM and GS1

GS1 EANCOM is GS1's subset of UN/EDIFACT for retail and consumer goods supply chains. GS1 describes EANCOM 2002 as a subset of selected messages from directory D.01B, including only the parts that are mandatory or relevant to GS1 users. It requires GS1 identifiers throughout: GLNs for parties and locations, GTINs for products, SSCCs for pallets and cartons.

EANCOM 2002 replaced EANCOM 1997 (built on D.96A) and exists in a syntax version 3 and a syntax version 4 edition. Each EANCOM message carries its own subset version number, EAN plus three digits, at the end of the UNH message identifier. In EANCOM 2002 syntax 3:

If a UK or European retailer sends you an "EDIFACT spec", expect an EANCOM-based guide with their own mandatory segments and codes layered on top.

EDIFACT message families

The EANCOM 2002 syntax 3 edition alone has 46 messages. These are the ones a retail or wholesale supplier meets most, grouped by the job they do. The EDIFACT message types reference lists every message by business flow.

The mapping to X12 is at the document level. Field-level mapping is never one-to-one: an X12 PO1 line carries quantity, price and product IDs in one segment, while EDIFACT spreads the same data across LIN, PIA, QTY and PRI.

EDIFACT vs X12

The two standards solve the same problem. The differences are syntax, governance and geography.

Envelope mapping

Same order, both syntaxes

The header of the order above, side by side:

X12:      BEG*00*SA*HL-PO-88231**20261011~
EDIFACT:  BGM+220+HL-PO-88231+9'
          DTM+137:20261011:102'

X12 packs purpose, type, PO number and date into one BEG. EDIFACT splits it: BGM for document type (220 order), number and function (9 original), and a separate DTM for the date with a qualifier (137 document date) and a format code (102 CCYYMMDD). Our UBL vs X12 vs EDIFACT comparison adds the Peppol UBL column if you also deal with e-invoicing.

EDIFACT vs TRADACOMS

UK suppliers sometimes meet both. TRADACOMS is the UK retail standard from 1982 that EANCOM is replacing.

The free EDIFACT inspector reads both, and the TRADACOMS file types reference lists every TRADACOMS file.

Other standards you may meet

GS1 UK's EDI guidance for new users lists the standards around EDIFACT that UK companies run into. Two matter outside retail:

• ODETTE, a regional standard for Europe's automotive industry.
• VDA, an automotive standard used mainly in Germany.

If an automotive customer asks for EDI, ask which of these it means before assuming EANCOM.

Who uses EDIFACT

• UK retail and consumer goods. UK grocers and retailers trade in EANCOM, and some still run TRADACOMS alongside it. Each publishes its own requirements: see our pages for Tesco, Sainsbury's, Asda, Morrisons, Ocado, Waitrose, Marks & Spencer, Boots and the Co-op.
• Continental Europe. EDIFACT is the usual EDI syntax for retail, wholesale and manufacturing across the EU, supported by national GS1 organisations.
• Global logistics, customs and transport. Carriers, ports and customs authorities use EDIFACT messages such as IFTMIN and IFTSTA, wherever their customers are based.
• Multinationals. A US company with European subsidiaries often runs X12 with US customers and EDIFACT with European ones from the same ERP.

How to start trading EDIFACT

1. Get the partner's implementation guide. Confirm the subset (usually EANCOM), the directory (D.96A or D.01B), the syntax identifier (UNOA, UNOC) and the message list.
2. Get GS1 identifiers in place. Most EANCOM partners require your GLN in the UNB and NAD segments and GTINs on every line. A UK supplier gets these from GS1 UK.
3. Agree the transport. AS2 and SFTP are common; some partners still use a VAN.
4. Map once to your ERP data model, not per message. Customer, product and unit-of-measure mapping is the expensive part, and it is shared with your X12 flows. See ERP EDI integration.
5. Test with real files. Paste the partner's samples into the EDIFACT inspector to check envelopes, counts and escapes before you go live.

Frequently Asked Questions

What does EDIFACT stand for?

EDIFACT stands for Electronic Data Interchange for Administration, Commerce and Transport. The full name of the standard is UN/EDIFACT because it is developed under the United Nations and published by UNECE. Its syntax rules are the international standard ISO 9735.

Is EDIFACT the same as EANCOM?

No. EANCOM is a subset of EDIFACT maintained by GS1 for retail and consumer goods. Every EANCOM message is a valid UN/EDIFACT message, but EANCOM removes segments that retail does not use, restricts code lists, and requires GS1 identifiers (GLN, GTIN, SSCC). Most European retail partners trade EANCOM, not raw UN/EDIFACT.

What is the difference between D96A and D01B?

They are two releases of the UN/EDIFACT directories, from 1996 and 2001. EANCOM 1997 is built on D.96A and EANCOM 2002, the current GS1 edition, on D.01B. Your partner's UNH tells you which one they use.

Is EDIFACT harder than X12?

No, once you know the separators. EDIFACT uses more generic, qualifier-driven segments (one DTM segment for every kind of date), so messages have more segments than the X12 equivalent. The optional UNA, the release character and the dropped trailing elements are the parts X12 teams most often get wrong.

Can one EDI system handle both EDIFACT and X12?

Yes. Both standards carry the same business documents, so the work splits into a syntax layer (parsing and generating each format) and a business layer (orders, products, customers) that is shared. OrderSync supports EDIFACT, EANCOM and TRADACOMS alongside X12, PDF, CSV and email, through the same pipeline into your ERP.

What is the EDIFACT equivalent of the 997?

CONTRL, the syntax and service report message. Like the 997, it reports whether an interchange or message was received and parsed, not whether the business content was accepted. Business-level rejections in EDIFACT usually travel as APERAK or as a response message such as ORDRSP.

Which EDIFACT version should I use?

The one your trading partner specifies. Retail partners mostly use EANCOM 2002 (D.01B) or EANCOM 1997 (D.96A). The version is in the UNH segment of every message they send, so a sample file tells you the answer.

Why does my EDIFACT file fail with an apostrophe in a name?

The apostrophe is the default segment terminator, so "O'Neill" ends the segment after the O. Escape it with the release character: O?'NEILL. The same goes for +, : and ? inside data.

Related guides

• EDIFACT message types: every EDIFACT message, grouped by business flow
• EDIFACT ORDERS: the purchase order message, segment by segment
• EDIFACT ORDRSP and ORDCHG: responses and changes
• EDIFACT DESADV and RECADV: despatch and receipt
• EDIFACT INVOIC: the invoice, including VAT
• EDIFACT CONTRL and APERAK: acknowledgements
• TRADACOMS: the older UK retail standard still in use
• EDI order lifecycle: how the X12 documents connect, which maps directly onto EDIFACT

Sources

• Introducing UN/EDIFACT, UNECE
• UN/EDIFACT syntax implementation guidelines, UNECE
• UN/EDIFACT syntax rules (ISO 9735), UNECE
• EANCOM 2002 syntax 3 manual, including Part 1 syntax rules and the message list, GS1
• GS1 EDI standards, GS1
• EDI guidance for new users, GS1 UK

Paste any EDIFACT file into the free EDIFACT inspector to check its envelope and segment counts, or use the X12 EDI inspector for your US partners.