From 200 to 600 SKUs in Spain, then three new EU marketplaces: master catalogue and AI localisation

The problem

The brand had a catalogue to sell across Europe, but its product content lived in the marketplace itself, one listing at a time. Every new market meant rewriting listings by hand, in another language, from scratch. That does not scale to hundreds of SKUs and several countries.

Context

I was Amazon Country Manager for Dnipro-M Spain from March to September 2026, responsible for the Amazon account and its expansion. When I joined, the brand had 200 SKUs, only on Amazon.es. The work started there and then extended to Germany, France and Italy, and beyond Amazon to ManoMano, Leroy Merlin and Carrefour in Spain and France: the same 600 SKUs on ten storefronts.

Approach

Treat the catalogue as data, not as listings. Build one master catalogue that is independent of any single marketplace, check it against the brand’s official information, and generate every market’s content from it.

Tools

Process

1. Grow the base market. The Amazon.es catalogue went from 200 to 600 SKUs.
2. Extract. A Python scraper pulled the live listing content into a structured master catalogue.
3. Verify. Each SKU was cross-referenced against the brand’s official website to flag data discrepancies before anything was published elsewhere.
4. Localise. An AI pipeline generated native Spanish, Italian, French and German content from the catalogue — written from the data, never beyond it — with human review. The approach is explained in Amazon listing localization for Europe.
5. Publish. Content was exported as marketplace-ready flat files for each store.
6. Localise images. Text inside product images was localised in batches, keeping design and layout.

Result

• Amazon.es catalogue grown from 200 to 600 SKUs: 400 new listings.
• The same 600 SKUs published per market on Amazon DE, FR and IT from the master catalogue.
• Listing production went from a few products per day to hundreds.
• Catalogue discrepancies caught before publication, not after.
• Expansion extended beyond Amazon: the same 600 SKUs on ManoMano, Leroy Merlin and Carrefour, in Spain and France.

Lessons

• The catalogue is the product. Once product data is structured and verified, each new market is an export, not a rewrite.
• AI is a drafting engine, not a source. It is only as good as the attributes it is given, and it needs native review.
• Rules belong in the pipeline. Limits such as Amazon’s 75-character titles are enforced at generation time and validated again before upload — the Amazon Title Checker runs the same checks.
• Verify against the brand’s own data. Marketplaces drift; the brand’s official information is the reference.

Since September 2026 I work as an independent Amazon and marketplace consultant, and I am open to senior roles: see my profile as an Amazon Marketplace Manager.