Most sellers I talk to think the Nordics are a translation problem. Get the copy into Swedish, Danish, Norwegian and Finnish, ship it, done. Then they wonder why their listings sit on page three while a competitor with a plainer product outranks them. The gap is almost never the translation. It’s the data underneath it.
Language is a field, not a paragraph
I watched a wholesaler try to launch a kitchen range on CDON last year. Good products, fair prices, decent photos. Their agency had translated the marketing descriptions into all four Nordic languages and everyone felt good about it. Two weeks in, barely any placement.
The prose wasn’t the problem. CDON reads listings as structured records and ranks them on content completeness, and their feed had one description field doing quadruple duty while the localized versions lived in a spreadsheet nobody had mapped to a language indicator. Swedish shoppers were getting Danish snippets. The brand and MPN drifted from variant to variant. The category mapping was off. CDON’s quality model saw an incomplete product and treated it like one.
That’s the thing I keep coming back to. On a Nordic marketplace, “speaking four languages” isn’t a copywriting task. It’s a data task. Every language is a field that has to be populated, validated and tied to the right locale, per variant, or the ranking engine downgrades you. That is the same discipline behind keeping product data quality under control: the field either exists, is consistent and is in the right place, or the channel treats the record as incomplete.
Why CDON rewards the boring stuff
CDON is the biggest marketplace in the region, spanning Sweden, Norway, Denmark and Finland, with 1,500+ merchants and roughly 12 million products. It’s a pure third-party platform now: about 94% of GMV flows through third-party sellers, with a category take rate around 20% in 2026. You don’t get a storefront and hope. You get a listing that competes inside a ranking algorithm.
And that algorithm runs on content. CDON scores listings by how complete and clean the content is, then buckets merchants into four performance tiers. Better content, higher placement. It’s not a slogan. It’s the mechanic.
The mandatory pieces have been in place since October 2021: a barcode (EAN/GTIN), the manufacturer part number, and the brand, all in the product data. Descriptions go in the local language of each market (Swedish for SE, Danish for DK, Norwegian for NO, Finnish for FI, with English as the international fallback), up to 10,000 characters. Titles cap at 150 characters. Nordic shoppers browse visually and compare hard, so your images and structured attributes carry more weight than any clever tagline. If you want those images to pull their weight, optimising your product images for size and format matters as much as the copy sitting next to them.
Here’s what a CDON-ready record actually needs, per product:
None of it is glamorous. All of it moves you up the page.
Finnish is where the wheels come off
If you remember one thing, remember this. Swedish, Danish and Norwegian are close cousins. Finnish isn’t. Different language family entirely, and AI models tuned on Scandinavian languages routinely trip over it. Norwegian has its own trap too: two written standards, Bokmål and Nynorsk, that need handling as separate outputs, not one “Norwegian” bucket.
So when a vendor tells you their tool “does the Nordics,” ask exactly how it handles Finnish, and whether it treats Bokmål and Nynorsk as distinct. That one question separates the tools that understand the region from the ones that ran everything through a single model and hoped.
The old fix was brute force. Litium’s 2026 B2B report (915 Nordic decision-makers, surveyed January 2026) puts multilingual product data at the center of the operational bottleneck for wholesalers going digital, and industry sources put a professional translation of a 3,000-SKU catalog across the five Nordic languages at €50,000 to €100,000, running to months of work. That’s a real number, and a real reason people quietly drop Finnish and leave a market on the table.
“Speaking four languages” isn’t a copywriting task. Every language is a field, and Finnish is the field most tools leave empty.
This is exactly the work that pays off once the source is clean rather than every time you export. Localized descriptions and attributes can be produced with AI enrichment across your product data, and AI product descriptions and translation carry the volume, as long as a human checks the languages a model is most likely to get wrong.
Your prices live on comparison engines too
CDON isn’t the end of it. Nordic buyers price-check obsessively, and the comparison engines run on the same feed you think of as “just data.” Prisjakt (and its international arm PriceSpy) wants nine required fields, including a category mapped to its taxonomy, images at least 500×500 over HTTPS, and a sale price that’s genuinely lower than the regular one. PriceRunner, owned by Klarna since 2022, wants every variant as its own feed row with its own EAN and SKU. A shoe in seven sizes is seven lines.
| Channel | What it insists on | Where the gap comes from |
|---|---|---|
| CDON | EAN/GTIN, MPN, brand and a localized description per market | One description field reused, brand/MPN drifting per variant |
| Prisjakt / PriceSpy | Nine required fields, category mapped to its taxonomy, images ≥ 500×500 over HTTPS | Your internal category tree and low-res or HTTP images |
| PriceRunner (Klarna) | Every variant as its own row, each with its own EAN and SKU | Variants collapsed into one line in the source feed |
One catalog, three rule sets. Clean, structured, per-variant data keeps you visible wherever your buyers land.
One thing worth knowing for 2026: Prisjakt raised fees around 1 January and set off a real merchant revolt, with Inet, Cyberphoto, Jollyroom and CDON among those pulling out, and its planned March listing got postponed. Where your buyers compare prices is shifting under your feet. The one constant is that clean, structured, per-variant data keeps you visible wherever they land. Splitting a catalog into proper per-variant rows is exactly the job behind merging simple products into variants, which is what a channel like PriceRunner quietly assumes you’ve already done.
How we actually approach it
I’ll be straight about what we do and don’t do. We let AI carry the heavy lifting: localized descriptions and attributes, per language, per variant, at a speed no human team matches. But it runs with our people alongside it, never AI-only, because Finnish and Nynorsk are exactly the cases where an unchecked model quietly ships something wrong.
We read your existing shop and supplier structure and build on it instead of making you start over. We never overwrite your source. Enrichment sits in layers, and you’re always one click from the original. That matters more than it sounds when you’re pushing the same catalog to CDON, Prisjakt and PriceRunner and one bad overwrite propagates everywhere. Pulling scattered supplier files into a single consistent record is the groundwork, and merging supplier data into a golden record is where per-variant EAN, MPN and brand stop drifting.
A quick sanity check before you push a Nordic feed:
Run that list and you’re ahead of most of the sellers you’re up against. If your catalog still lives in spreadsheets, the honest first step is turning that into a structured source, which is what going from Excel chaos to a clean product catalogue is about. From there you can see how SyncRefine works on your existing shop setup without overwriting it.
The honest question I’d put to you is this: if you exported your catalog right now and opened the Finnish column, would there be anything in it? For most people the answer is no. And that empty column is exactly where a fourth market is hiding.
In half an hour we’ll walk through how your product data travels today, from suppliers all the way to CDON, Prisjakt and PriceRunner, and tell you honestly which fields derive cleanly and which languages need a real human check. If you want more on the same theme, product feed management for Amazon, eBay and Google covers how one clean catalog projects into different channel formats, do I need a PIM helps you decide how much structure you actually need, and a clean product catalogue is where all of this starts.
