Europe's refurbished marketplace, one country at a time, as JSON

refurbed's own, word for word. It grades device appearance in four categories and the product page prints them as Premium, Excellent, Very good and Good, while its data layer labels the same four AA, A, B and C. Both are returned: grade is the display name, grade_letter is refurbed's letter and grade_code is the lowercase short form. The pairing was checked on 32 live rows and agreed 32 of 32, with no exceptions. Nothing is translated into a scale of ours. A result card prints the letter rather than the name, so on a search row the name is resolved through refurbed's own letter-to-name table, the same one its product page prints; that pairing agreed on 32 of 32 live rows, so the name repeats refurbed rather than inferring anything.

That is the question the product action is built to answer. It returns grade_options for the device you asked about: every appearance grade refurbed currently has in stock, each with refurbed's own price difference and its own offer id. On the iPhone 13 we sampled on refurbed.de, Good was 264.99 EUR and refurbed printed Very good as +13.01, Excellent as +35.01 and Premium as +74.01 against it. Battery is a separate ladder on phones — Optimal or New, +10.02 there — and comes back as battery_options. When refurbed has only one grade of a device left there is no ladder to return, so grade_options is null and grade still tells you which one it is.

The one you asked for, and the response says so. country is required in practice — it defaults to de — and every response echoes meta.country, meta.locale, meta.currency and meta.currency_symbol. The currency is read from the storefront's own structured data, which matters: refurbed's analytics payload reports EUR on all 24 storefronts including refurbed.co.uk, so taking the obvious field would have published British prices labelled EUR. Across the 24 storefronts the live values are EUR on 19, GBP, CHF, SEK, DKK, PLN and CZK on the other five. If a storefront ever changes currency, the engine returns the live value and says so in meta.warnings instead of staying wrong quietly.

Yes, that is compare_countries. Give it a keyword or a category and 2 to 6 country codes and it reads each storefront's own search page inside one time budget, returning each one's cheapest matching offer with that country's currency and its own total. Six is the cap so a single call cannot outgrow its budget, and any storefront that did not finish is counted in meta.countries_failed with the reason rather than dropped silently.

price is what the storefront prints to a buyer, and price_new_reference is the brand-new retail reference refurbed shows struck through next to it. The second one is not a refurbed price and we do not call it one: refurbed's own tooltip describes it as the device's new retail price averaged from an independent price-comparison portal and recalculated daily. It is null whenever the storefront prints no strikethrough. There is also price_feed, which is the number in the storefront's own data layer for the same row — and it disagrees with the printed price on roughly a fifth of rows, always by a round 10 or 20 units and always lower. We publish the printed price, keep the data-layer number beside it, flag the row with price_mismatch and count the flagged rows in meta, because that disagreement is real and hiding it would mean quoting people prices that are too low.

No, and this is worth knowing before you build on it. refurbed's keyword search is fuzzy and it never reports a miss: the nonsense keyword zzqqxxnotathingqq still returned 300 rows on refurbed.de on 2026-10-01. A keyword result is a relevance list, not a containment filter. When you need an exact scope, browse a category slug and add brand, colour and a price range — those are real filters and every one of them was measured to change the total.

Every search response reports meta.total_results, refurbed's own exact match count, plus meta.filters_requested and meta.filters_applied_by_source, which is the storefront echoing back which filters it recognised. In one run against an unfiltered 506 German smartphone offers: brand=Apple 43, brand=Apple,Samsung 172, colour=Black 313, price_min=800 70, price_max=200 224, price 200 to 400 218, and Apple under 300 EUR 21. Sorting is verifiable the same way — sort=price_asc returned 35.66 EUR as its first row and sort=price_desc returned 2,844.60, which are exactly the lowest and highest price that category publishes for itself. A brand the storefront does not stock comes back as a real zero and meta.warnings says the storefront did not recognise the value, so you can tell that apart from an empty shelf.

No, because refurbed does not publish it. refurbed is a managed marketplace and its public product page names no merchant, no shop and no per-merchant rating anywhere — we probed for it rather than assuming. What the page does publish, and what you get, is the storefront entity (refurbed Deutschland, refurbed UK, and so on), the product rating with its review count, the stated warranty, the return window and the shipping and delivery-time ranges. Rather than fill a seller field with nulls, the API leaves it out.

The offer id and the slug, both of which every search row gives you, and both of which are in any product URL: refurbed.de/en-de/p/iphone-13/14162c/ is slug iphone-13, id 14162 and grade code c. The slug is mandatory because refurbed has no id-only product route and answers HTTP 400 for an id with the wrong slug — we measured that rather than offering you a handle that does not work. Pass the row's grade_code as grade and you land on exactly that offer; leave it out and you get refurbed's default grade for that device, which is usually a different price.

The full spec sheet as refurbed labels it (24 rows on the iPhone 13: battery capacity, camera, connectivity, connectors, dimensions, display type, operating system, processor, weight and so on), every gallery image in display order, all colour and storage variants with their own price and offer id (18 on that iPhone 13), the appearance-grade ladder and the battery option, the product rating with its review count, the storefront's stated guarantees including the minimum warranty, the return window in days and whether returns are free, the shipping cost and the handling and transit day ranges, and the breadcrumb path. Everything in the search row is in there too.

Two cheap actions. categories returns that storefront's live category tree read off its own navigation — 60 slugs with labels on refurbed.de, 58 on refurbed.co.uk — and a returned slug goes straight into search's category. filters returns the live filter taxonomy for one search or category: refurbed's own attribute ids and labels with the exact value list of every enum and the min and max of every numeric one, plus that result set's price range and currency symbol. Smartphones exposed 10 attributes, laptops 15 — the taxonomy is per category, which is exactly why you should read it rather than guess it.

You get a reason, never an empty success to interpret. An offer id that is not live in that storefront is NOT_FOUND. A slug that does not belong to the id is INVALID_PARAM with an explanation, because that is our URL being wrong and not the site refusing us. A country, category, grade or sort value refurbed does not have is rejected with the allowed list. Calling search with neither query nor category is MISSING_PARAM that says so, rather than dumping the whole catalogue on you.

Refurbed API is a ReefAPI endpoint group for europe's refurbished-electronics marketplace as json: 24 country storefronts, each with its own price list and currency, refurbed's own four appearance grades and the price step between them. It returns live JSON through POST requests under /refurbed/v1.

Yes. ReefAPI starts with 1,000 free credits, no card required. Refurbed calls use the same shared credit balance as every other ReefAPI engine.