The record before the product — why structured feeds decide multi-market visibility

The observation. A single product can exist across a dozen countries, in eight languages and six currencies, and still fail to appear in any shopping comparison engine — not because it is a poor product, but because its data is incomplete, ambiguous or badly localised. Shopping engines do not rank products ; they rank structured data records. The quality of the product feed has become, in only a few years, a distribution asset in its own right — as decisive as price or photography.

What is a structured feed?

A product feed is a file — usually XML, TSV or delivered via API — that describes each item through a set of standardised attributes : identifier, title, description, price, availability, image, brand, category. The Google Merchant Center specification, like most comparison engines, distinguishes required from recommended attributes. The former govern eligibility ; the latter govern ranking and the relevance of search matching.

At the centre of the system sits the product identifier. The GS1 standard — the body that has administered barcodes since 1974 — defines the GTIN (Global Trade Item Number), of which the European EAN-13 and North American UPC are variants. A valid GTIN lets an engine attach competing offers to the same physical product, deduplicate them, and enrich a record with third-party reviews or specifications. Without a reliable identifier the product stays an island : invisible to grouping, excluded from consolidated product pages.

Completeness and quality: two distinct axes

Completeness measures the proportion of attributes filled in. Quality measures their accuracy and consistency. A feed can be complete and wrong : a well-formed GTIN assigned to the wrong item, a colour declared “silver” for a black product, an availability of “in stock” for a sold-out reference. The data-quality literature — from the ISO 8000 standard to academic work on data quality management — converges on the same dimensions : accuracy, completeness, consistency, timeliness, uniqueness.

Shopping engines penalise inconsistency more harshly than absence. A missing attribute degrades query coverage ; a contradictory attribute — a feed price different from the landing-page price — triggers a mismatch that can suspend the offer. The rule is constant : the record must reflect exactly the page it points to.

Localisation: one record per market, not a translation

Serving several markets is not translating a single feed. Each market imposes its language, its currency, its price format and — often forgotten — its displayed taxation. A submitted price must include or exclude VAT according to local conventions, and the declared currency must match the landing page : an offer in euros pointing to a page in Swiss francs is a classic mismatch. Units (centimetres versus inches), sizes and regulated categories also vary from one country to the next.

Localisation also covers the language of search. A German, Spanish or Italian customer does not phrase a query with the words of a French one. A title translated literally, without regard for the terms actually searched in each market, loses matching power. Data must be localised, not merely translated.

Why engines reward structure

A shopping engine seeks to match a query to a buying intent, then to rank comparable offers. The more structured the data — reliable identifier, standardised attributes, correct category — the more the engine can understand, group and present the offer with confidence. A rich record improves match relevance, feeds faceted filters and enables enriched information display. Conversely, a poor record is not merely ranked lower : it is less often eligible for the query at all.

The roots of disapproval

The most frequent causes of disapproval are well known and broadly documented by comparison engines : a price or availability mismatch between feed and page ; an invalid or missing identifier ; a missing, low-quality or watermarked image ; an unreachable landing page or one without a clear returns policy ; non-compliance of a regulated category. Almost all reduce to a break between what the feed says and what the page shows. To fix a feed is, first of all, to restore that concordance.

Where we stand

Montandor runs a multi-market product feed, served in several languages and currencies, to make its professional catalogue visible where HoReCa customers search — without confusing markets or tax regimes. Our discipline is simple : a reliable identifier per product, a record that reflects exactly the landing page, and a localisation that respects each country's search language. The feed is not one more technical export ; it is how a product exists, or fails to exist, in comparison commerce.

“A fine product, badly described, does not exist for a shopping engine. Feed quality is not a technical formality ; it is the condition under which a customer in another country, searching in their language and paying in their currency, lands on the right record, at the right price, with no surprise.”
— Wouter Meijboom, CEO, Montandor Andorra.

Sources

• GS1 — GTIN (Global Trade Item Number) and the GS1 barcode system, identification standards since 1974 (gs1.org).
• Google Merchant Center — product feed specification (required and recommended attributes, price/availability matching rules), Google documentation.
• Google — Shopping ads policies and product disapproval causes (Merchant Center help centre).
• ISO 8000 — international standard for data quality, International Organization for Standardization.
• Academic literature on data quality management — dimensions of accuracy, completeness, consistency, timeliness (Wang & Strong, and subsequent work).

Published 26 May 2026 by the Montandor team — research led by Dorota Sawicka (Merchant Feed Manager), in collaboration with Hélène Vincent (Growth & Analytics Lead) and Céline Faure (Content & SEO Lead).