How do fashion brands syndicate product data to retailers and marketplaces?

KEY TAKEAWAYS Summary by the editors

1. Product data syndication means maintaining one master record per style, colour and size in a PIM and transforming it into each retailer's or marketplace's required format, rather than keeping separate spreadsheets per channel.
2. GS1 states that every size, every colour and every combination of size and colour needs its own GTIN, which makes clean variant structures the foundation of any syndication setup.
3. Wholesale partners still often receive article master data through the EANCOM PRICAT message, which GS1 defines as transmitting pricing and catalogue details from seller to buyer.
4. Marketplaces such as Zalando use a three level product model (model, config and simple) with category specific mandatory attributes, image sets and validation reports, so a PIM must map internal attributes to each channel's taxonomy.
5. AI is most useful in syndication for attribute mapping suggestions, enrichment of missing values and pre-submission checks, but it needs a governed master record and human review of anything customer facing.

Fashion brands syndicate product data by keeping one governed master record per style, colour and size in a product information management (PIM) system, then mapping and exporting it into each partner's required format: EDI catalogue messages for department stores, API submissions for marketplaces and feeds for retailer webshops. The work is less about the export itself and more about structure, identifiers and attribute mapping, which decide whether a partner accepts the data the first time.

What does product data syndication mean in fashion?

A brand selling through its own webshop, a dozen wholesale accounts and two or three marketplaces typically faces a different data specification at every partner. One retailer wants a PRICAT file with prices and GTINs, another uploads spreadsheets into its own supplier portal, and a marketplace expects a JSON submission with its own category tree and attribute values. Without a central system, teams rebuild the same information several times per season and every copy drifts.

Syndication flips that model. The PIM holds the single source of truth for customer facing product data (names, descriptions, composition, care, sizes, images), while channel specific transformations live in mapping rules. When a composition value is corrected in the master record, every outgoing channel picks up the change at the next export.

Which data formats do wholesale partners and marketplaces expect?

Formats fall into three broad families, and most brands need all three at once.

GS1 defines the PRICAT message as a way to transmit "pricing and catalogue details for goods and services offered by a seller to a buyer", and notes that buyers can also use it to respond with acceptance or rejection. For many traditional wholesale relationships it remains the reference format for article master data.

Marketplaces work differently. Zalando's developer documentation describes a three tier product hierarchy with merchant identifiers at model, config and simple level, and names the EAN as the primary product identifier. New products are submitted through a product submissions API against category specific attributes, with a mandatory image set, and validation results come back in a product status report. Zalando's partner onboarding material also requires category specific information such as silhouette and material composition, and a minimum run of three consecutive sizes for textiles.

Read also

Endless aisle and drop shipping between brands and retailers

How should a PIM be structured for many channels?

Most syndication problems trace back to the data model rather than the connector. A model that survives many channels usually follows a few principles.

1. Model variants explicitly. Keep style, colour and size as separate levels. GS1 states that each size, each colour and each combination needs its own GTIN, so the size level must carry one identifier per sellable unit.
2. Separate master attributes from channel attributes. Store composition, care and measurements once; store marketplace category codes, retailer specific names and channel copy in clearly labelled channel fields.
3. Use controlled vocabularies. Free text colour names or fit descriptions are hard to map. Internal value lists that map to each partner's enumerations make exports predictable.
4. Align with a shared classification where possible. GS1 Netherlands publishes a fashion data model based on the Global Product Classification, which links specific attributes to product categories and gives a neutral reference point for mapping.
5. Keep prices and stock out of the product record. Zalando, for example, manages prices and stock per sales channel through separate APIs, and wholesale price lists change on their own cycle.

How does AI help with syndication?

AI does not replace the data model, but it removes a good share of the repetitive mapping and enrichment work once the model exists.

• Attribute mapping suggestions: language models and classifiers can propose which internal value matches a partner's enumeration (for example an internal "relaxed tapered" fit to a marketplace silhouette value), with a human approving new mappings once.
• Enrichment of missing values: image recognition can suggest neckline, sleeve length or pattern from product images, and text models can draft channel specific descriptions from structured attributes.
• Pre-submission checks: rules plus models can flag records that would fail a partner's validation (missing mandatory attributes, composition not adding up to 100 percent, implausible measurements) before they are sent.
• Translation and localisation: machine translation of descriptions and care texts, reviewed for regulated wording.

The limits are practical. Suggestions are only as good as the master data they start from, generated copy can describe features the garment does not have, and composition or care information has legal weight. A sensible rule is that AI may propose and pre-fill, but a named owner approves anything that leaves the company.

What should brands check before connecting a new partner?

Onboarding a new retailer or marketplace is where the structure is tested. Useful questions to answer before the first export:

• Which identifier does the partner use as primary key, and do all sizes already have a GTIN?
• Which attributes are mandatory per category, and which have no equivalent in the master record yet?
• Which image specifications (angles, background, resolution) apply, and do current assets meet them?
• How does the partner report validation errors, and who on the brand side reads and fixes them?
• How often does the partner's specification change, and how will the brand be informed?

How do you measure whether syndication works?

A small set of indicators shows whether the setup is paying off: the share of records accepted on first submission per channel, the time from product approval to live on each partner, the number of manual corrections per season and the volume of partner queries about product data. Tracking these per channel shows where mapping rules or master data need attention, and gives a sober baseline for judging any AI tooling.

Read also

What a modern tech stack looks like for a mid-sized fashion brand

Where do teams usually go wrong?

Three patterns recur. First, enrichment happens downstream in channel tools rather than in the PIM, so improvements never reach other partners. Second, marketplace requirements drive the master model, which then fits one partner and breaks the others. Third, ownership is unclear: nobody is accountable for a partner's rejection report. Fixing these is organisational as much as technical, and it is the precondition for AI to add value rather than amplify inconsistencies.

Frequently asked questions

What is product data syndication?

It is the process of distributing product information from one master source, usually a PIM, to many sales channels in the format each channel requires. The PIM holds the governed record, and mapping rules transform it for retailers, marketplaces and own webshops.

Do I need a separate GTIN for every size and colour?

Yes. GS1 states that each size, each colour and each combination of size and colour needs its own GTIN. Retailers and marketplaces rely on this to identify the exact sellable unit.

What is a PRICAT message?

PRICAT is an EDI message in the EANCOM standard that GS1 defines as transmitting pricing and catalogue details from a seller to a buyer. In fashion wholesale it is commonly used to send article master data such as GTINs, descriptions and prices to retail partners.

Can AI map my product attributes to a marketplace automatically?

AI can suggest mappings between internal values and a marketplace's attribute enumerations and can flag records likely to fail validation. It works best on a clean master model with controlled vocabularies, and new mappings should be approved by a person before they are used at scale.

GuideThe complete guide to AI in fashion design and product developmentRead the complete guide