Product data
Amazon product data requirements: get your catalogue ready first
Daniel Pawłowski · Amazonway · · approx. 12 min read
For a few hundred products to go live on Amazon in days rather than weeks, one thing has to be true: the data has to already exist on your side. That means product identifiers or a deliberate decision to seek an exemption, the attributes required in your specific category, a clean variation structure, images in a format Amazon will accept, content in the language of every store you sell in, prices and stock per marketplace, and compliance data. The source format is irrelevant: an online store, a CSV file, an ERP or an export from another marketplace are all equally good starting points, as long as the data is complete and consistent. You do not need to roll out a PIM before you can start. What sets the pace is the state of your catalogue, not the tooling on the agency side.
Amazon requirements as of 13 August 2026. Requirements differ by category and by country store, and Amazon changes them, so treat the specific numbers in this article as reference points rather than universal rules. What binds you is the template for your category and the notices on your own account. Every claim below carries its source and date.
Why your catalogue sets the pace, not the tooling
Listing a product on Amazon is a data operation. Someone takes your description of the product and maps it onto the structure Amazon expects for a specific product type in a specific country store. If a field Amazon requires is missing from the source, it cannot be invented. It has to be asked for.
That question loop is the only real brake. Uploading a few hundred rows takes minutes and Amazon's validation runs in hours. What turns into weeks is the back and forth over gaps: missing EANs, images at web-shop resolution, variations described in a way that no parent-child relationship can be built from, no contact details for the EU responsible person.
This is why the first step on a new account is not logging into Seller Central but reviewing the catalogue for completeness. That is an observation from our own work rather than an Amazon rule: catalogues that pass such a review without comments go live quickly, and catalogues with holes go live exactly as slowly as it takes to fill those holes.
Catalogue readiness checklist: three tiers
Not every gap weighs the same. Below is the split we use in a review: what blocks the launch outright, what slows it down and what can safely wait until after publication.
| Tier | What it covers | Why it sits here |
|---|---|---|
| 1. You cannot start without it | A product identifier or an approved exemption; an assigned product type and category; a product name; at least one image meeting the technical requirements; price and availability; compliance documents wherever Amazon asks for them | Amazon rejects the row or blocks publication. There is no workaround and no temporary version |
| 2. It will slow you down | The full set of attributes required in the category; a correct variation structure; secondary images; content in the language of each market; prices and stock per store | The listing goes live but stays incomplete and harder to filter, and you will keep coming back to it with every fix |
| 3. Fix it later | A+ Content; lifestyle photography and infographics; backend keywords; polished bullet points; video | It moves conversion, not whether the product is on sale at all |
A practical rule: collect tier one in full before anyone opens a template. Collect tier two in parallel with the first uploads. Do not start tier three before publication, because you are then polishing something nobody has seen yet. Items from that last group, such as A+ Content, only make sense once the detail page is live and pulling traffic.
Identifiers: GTIN, EAN, UPC and when you can go without
Amazon uses GTIN as the umbrella term and accepts four number types in that role: UPC (12 digits), EAN (13 digits), JAN (13 digits) and ISBN for books. In most categories an identifier is a precondition for creating a new listing, and the exceptions are documented per category (Amazon, guide to product IDs and GTINs).
GTINs come from GS1, through its national member organisations, and the lead time for getting a company prefix depends on which one you register with. In Poland, for example, GS1 Poland quotes up to three working days from application and cleared payment. Whatever your country, treat it as a lead time to plan for rather than a formality.
Every variant in your catalogue therefore falls into one of three situations:
- The product carries a manufacturer code. The simplest case. Check only that the code in your system is complete, with no leading zero lost to a spreadsheet, and that it belongs to the variant rather than to the whole product family.
- The product is yours and has no code. Two routes: buy a range of numbers from GS1, or apply to Amazon for an exemption from the GTIN requirement. The first is predictable in time, the second depends on Amazon's decision.
- The product already exists in Amazon's catalogue. You are not creating a new detail page, you are joining an existing ASIN, and your own copy is irrelevant because the page belongs to whoever controls it.
Three things about exemptions are worth knowing before you put them in a plan. Not every category allows them, applications are filed separately for each brand and each category, and enrolling a brand in Brand Registry does not by itself remove the identifier requirement. An Amazon moderator confirms as much in the European Seller Forums: whether an exemption is possible is category-dependent, and Amazon publishes the category list separately.
For the catalogue this boils down to one column rule: the identifier field must be filled for every SKU, or carry an explicit "exemption" decision. An empty field with no decision behind it is the single most common reason a spreadsheet goes back to the client.
Required attributes by category, and how to find out which they are
There is no single list of mandatory fields on Amazon. Requirements are defined per product type and per country store, so the same physical item can carry a different set of required attributes in Germany and in Spain, and certainly does across two different categories.
Amazon exposes those requirements in two places:
- In Seller Central, as the inventory file template for the category, with its Instructions, Template and Valid Values tabs and a marker for which columns are required.
- Through the API, where the Product Type Definitions API returns a JSON schema describing the attributes, their conditionality and the requirements for a given product type in a given store. Amazon states plainly that validating data on your own side before submission prevents most listing issues from happening at all.
The practical move: before you start gathering data, pull the list of required attributes for your categories and compare it with what you already hold. The difference is your work list, and it, not the number of products, tells you how much is left.
Typical divergences between categories look like this. Apparel needs size charts, material composition and target audience; consumer electronics needs power ratings and compatibility; household chemicals need hazard information. One universal file will not clear two different categories and is not worth building.
Variations: parent, child and variation theme
Variations are where store exports most often fall apart. Web shops describe them freely. Amazon does not.
Amazon builds variations on three fields: Parentage (parent or child), Relationship Type (variation on the child rows) and Variation Theme, which must hold an identical value on the parent and on every child. The permitted themes are listed in the Valid Values tab of the category template, and where a category offers no theme you simply cannot build variations in it (Amazon reply in the European Seller Forums).
The gaps that recur most often in source data:
- Variants as separate, unrelated products. Every size is its own row, but nothing records that they belong together. Without a parent key that relationship has to be rebuilt by hand.
- All variants in one cell. "Size: S/M/L" in a single row reads fine to a human and is useless to an import. No script can guess which combinations genuinely exist and at what price.
- Inconsistent value naming. "Black", "black", "BLK" and "Jet black" in one column are four different variants as far as the system is concerned.
- No separate SKU per variant. Without its own internal identifier you cannot run stock, prices or images at variant level.
- A variant that is not a variant. A newer model, a different brand or a bundle with another product are not versions of the same item, and Amazon does not accept those relationships.
The test to run on your own catalogue: a parent and its children are one product in several versions that a buyer expects to find on a single page. If you hesitate over whether something is one product or two, that is a sign the data needs a decision, not a mapping.
Images: technical requirements and what they mean for your data
Images are the second most common bottleneck, because unlike a missing field they cannot be filled in from a spreadsheet.
The technical requirements Amazon sets out in the Seller Forums:
- 500 to 10,000 pixels on the longest side, in JPEG, TIFF, PNG or non-animated GIF, sharp, with no pixelation or jagged edges;
- above 1,000 pixels on the longest side the zoom function works, which makes it the practical minimum rather than the formal one;
- a listing needs one image to be sellable, but Amazon recommends at least six images plus a video.
On top of that come the main-image standards Amazon restated in a separate thread on product photography standards: a pure white background (RGB 255, 255, 255), the product filling 85 to 100 per cent of the frame, the whole product in shot, nothing that is not part of the purchase, and no text, graphics, logos, watermarks or promotional messaging.
For a catalogue that translates into three things that are easy to overlook. Files have to be tied to a SKU, either through the filename or a URL column. If you upload by feed, they have to sit at a publicly reachable URL. And they have to exist separately for each variant that looks different, because one shot per product family is fine in a web shop and not on a page with a colour picker.
If your products only have images rendered for the store front, or only in styled settings with a background, that is not a correction. It is a photography project or a retouching pass across the whole assortment, with its own schedule and budget, and it is precisely the difference between launching in days and launching in weeks.
Content: titles, bullet points and several markets at once
Here is the change that is easy to miss and that invalidates a good part of catalogues prepared earlier. From 27 July 2026, titles in every category except media must be 75 characters or less, including spaces. Amazon announced it on 10 June 2026 in separate Seller Forums posts for the stores in the UK, Germany and the US, on the grounds that the title should display in full on mobile.
Three more points follow from those announcements:
- Item Highlights. An extra 125 characters for materials or recommended use cases. The content is searchable and shows below the title.
- Titles still over 75 characters are being replaced gradually with an AI recommendation. Listings stay active throughout, and you can change the title at any point.
- Brand owners get 14 days to review, amend and approve the AI suggestions before they go in.
The previous limit was 200 characters and had applied since 21 January 2025. Plenty of guides and internal templates still repeat it, so if your catalogue holds names built for 200 characters, you have rewriting ahead of you, not copying. The structural lesson matters more than the number itself: what you want in the catalogue are the components of a name (brand, product type, key feature, variant) that can be assembled into a title of any length, rather than one long finished string.
Bullet points carry their own requirements, updated on 15 August 2024, restricting special characters, emoji and phrasing such as refund guarantees. Length limits vary by category, so do not hard-code a single number into a template covering the whole assortment.
Languages are the separate trap that most multi-market plans fall into. Amazon requires detail pages to be translated into the local language of every country store you sell in, and that covers not just the title but the description, bullet points, specifications and even text inside images and video (Amazon on translating listings, 21 February 2025). Amazon offers Build International Listings, which can create pages in a target store using machine translation, but that is a starting draft, not a finished version. Five markets means five sets of content to maintain, including on every future change.
How to turn complete data into an offer that actually converts, meaning keyword selection, bullet writing and page structure, is covered in our guide chapter on listing a product and optimizing the offer. This article stops exactly where that one begins.
Prices, stock and markets
Price and availability are mandatory fields, yet catalogues are rarely prepared for multi-market selling. Four things to check:
- The currency belongs to the store, not to you. Euro-zone stores price in euro, Sweden in krona, Poland in złoty, the UK in pounds. A catalogue with a single "price" column needs a decision on how to convert it and who owns that decision.
- Price is the gross price of that market. VAT rates differ between countries, so the same margin produces a different shelf price in each store.
- Stock is held per variant SKU, not per product. Availability at product-family level is useless once the buyer is picking a size.
- The fulfilment model can differ per market. Amazon's warehouse in one country, your own shipping in another, which changes both cost and the delivery promise.
Amazon has its own tool for this, Build International Listings, which syncs prices from a source store to the others and accounts for exchange rate movements as well as differences in taxes and fees. It is a sensible way to start, but a conversion rule is not a decision about price positioning in each market. Where the cost differences come from we break down separately in our pieces on Amazon FBA fees and on the PAN-EU model.
Compliance data: part of the product record, not paperwork for later
Since 30 September 2024, where Amazon requires compliance documents for your product, they must be submitted and approved before you can list it. Previously you could list first and supply documents afterwards. Now it is a precondition. The change is confirmed in a thread in the European Seller Forums, where an Amazon staff member points to the help page on creating new selection with compliance requirements.
Documents are uploaded through the Manage Your Compliance dashboard, one at a time or in bulk. Amazon says a single spreadsheet can cover up to 5,000 products, and warns that missing documentation leads to listings being removed and can affect Account Health (Amazon on the Manage Your Compliance dashboard).
Alongside the documents sit data points that are not attachments at all but fields in the product record:
- GPSR. A responsible economic operator established in the EU, the manufacturer's contact details, and warnings and safety information in the language of each market. The full scope is in our piece on GPSR on marketplaces.
- EPR. Registration numbers for packaging, electrical equipment and batteries, separately in each country. Which registers and which numbers end up in Seller Central is covered in EPR registration in Germany and France.
The scheduling consequence is unforgiving: registrations, agreements and verifications run on their own clocks, and a complete catalogue does not shorten them. Start them on the day the data is ready and they, not the data, will set your launch date.
The source format does not matter
The most common worry before any conversation about data is "we would have to implement a PIM first". You do not. What counts at launch is the content, not the system holding it. Any of these is enough:
- An export from your online store, as CSV, XML or a product feed.
- A spreadsheet, even a manually maintained one, provided there is only one of it and it is current.
- An ERP or warehouse system you can pull a product master out of.
- An export from another marketplace you already sell on.
Whatever the format, we look for the same four things:
- A key. Every variant has its own stable internal identifier that lets copy, images, prices and stock be joined together.
- Completeness. The fields required in your categories are filled in, not "mostly filled in".
- Consistency. The same value always written the same way, units in one system, dimensions in the same order.
- A single source of truth. It is clear which file or system wins when two places disagree.
A PIM starts paying off later: with several channels, several languages and frequent assortment changes, once holding those four things together by hand stops being realistic. The technical side of that setup, one source catalogue mapped to each platform's format, is run by our sister brand Seedlight as part of product feed automation. For a first entry onto Amazon it is not a prerequisite.
The most common gaps and what they really do to the timeline
Below are the gaps that come back most often in our catalogue reviews, with what actually happens when they show up. This is an observation, not a statistic: the order reflects how frequently we see them, not a measurement.
- No identifiers on own-brand products. The launch waits either on a GS1 registration, with a lead time set by your national GS1 organisation, or on an exemption decision whose timing neither you nor we control.
- Images only at web-shop resolution. Not a spreadsheet fix but a shoot or a retouching pass across the assortment, with its own schedule and budget.
- Variants collapsed into one row. Splitting them runs manually, product by product, because only you know which combinations genuinely exist and at what price.
- Content in one language while the plan covers several markets. Translation moves onto the critical path instead of running alongside it.
- Gaps in only part of the catalogue. The most deceptive case, because the catalogue looks ready. Treat it as one batch and the worst-described product sets the pace for everything.
- Compliance data left until last. Everything else is done and the listing still will not go live, because a document is sitting in Amazon's approval queue.
The conclusion we draw from that: launch in batches. A category or product group with complete data goes live immediately and starts building sales history, and the rest follows in later waves. That is almost always better than waiting for the whole assortment to be ready at once.
What you prepare and what we take on
An honest split, so nobody is waiting on anybody at launch.
| Your side | Our side |
|---|---|
| Product identifiers, or the decision to apply for an exemption | Checking whether the category allows an exemption and preparing the application |
| Product photography at original quality | Cropping, preparing the main image to the standards and tying files to SKUs |
| Technical product data: dimensions, weight, material, specifications | Mapping it onto the attributes required in that category and that country store |
| Compliance documents, manufacturer and responsible operator details, registration numbers | Placing them in the right fields and documents in Seller Central and handling Amazon's requests |
| Prices, stock and the decision on the fulfilment model | Pricing rules per market and the stock update setup |
| Decisions: what counts as a variant, how the brand name reads, which markets go first | Parent-child structure, titles and bullet points against current requirements, translations, validation error handling |
What we will not do for you: invent an identifier, issue a document the manufacturer never produced, or photograph a product we do not have. Practically everything else can move to our side.
How this runs on our side
At Amazonway a catalogue review is the first step of any engagement, before any work inside Seller Central. It produces a list of gaps split across those three tiers and an order for closing them, so the first batch of products can go live before the rest is finished. Where the accounts are to be run on an ongoing basis, the same scope sits inside marketplace account management.
The honest caveats. With complete data, listing a few hundred products usually takes us a matter of days, but we do not promise a date, because several variables sit outside us: account verification, how long compliance documents spend in approval, category approvals and Amazon's own pace. Nor do we promise sales results or search placement, because those depend on the product, the price and the competition. What we are accountable for is data that is complete and consistent, and a listing that does not stall over a missing field.
FAQ: Amazon product data requirements
Does every product need an EAN?
In most categories an identifier is a precondition for creating a new listing, and Amazon accepts UPC, EAN, JAN and ISBN in that role. The alternative is an exemption from the GTIN requirement, but not every category allows one, applications are filed per brand and per category, and holding a brand in Brand Registry does not remove the requirement by itself.
Is an export from my online store enough?
Usually yes, as a starting point. What matters is whether the export carries a stable variant identifier, whether variants sit on separate rows, whether images can be tied to SKUs and whether the attributes required in your categories are populated. Gaps in those four areas cannot be made up for on the platform side.
How many images do I need to launch?
Formally one image meeting the technical requirements and the main-image standards. Amazon recommends at least six images plus a video, but missing extra shots does not block publication, it only lowers the quality of the page.
Will Amazon translate my copy for other markets?
Amazon offers Build International Listings, which can create pages in a target store with machine translation. The requirement, though, is that the detail page is in the local language of that country store, so treat machine-translated titles, bullet points and specifications as a draft for editing rather than a version ready to sell.
Do I need a PIM before selling on Amazon?
No. One current source of data in any format is enough: a store export, a spreadsheet, an ERP or an export from another marketplace. A PIM becomes justified with several channels, several languages and frequent assortment changes, once maintaining consistency by hand stops adding up.
What most often blocks a launch despite a complete catalogue?
Compliance data. Since 30 September 2024 the documents Amazon requires have to be submitted and approved before you can list, and GPSR and EPR registrations run on timelines owned by authorities and registers. That is the most common reason a complete product data set does not translate straight into active listings.
Sources
- Amazon, What are product IDs or GTINs? (identifier types and exemption basics)
- Amazon Seller Forums (Europe), GTIN Exemption Brand Approval (exemptions are category-dependent; Brand Registry does not remove the requirement)
- GS1 Poland, ordering a barcode (an example of national GS1 registration lead times, in Polish)
- Amazon SP-API, Product Type Definitions API (attribute requirements per product type and store)
- Amazon Seller Forums (Europe), Guidance on Parent-Child Variations Setup (Parentage, Relationship Type, Variation Theme, Valid Values)
- Amazon Seller Forums, Listings Lounge: Product Image Requirements and Product Photography Standards (technical file requirements and main-image standards)
- Amazon Seller Forums, title update announcements of 10 June 2026: UK, Germany, US
- Amazon Seller Forums, Review updated bullet point requirements (effective 15 August 2024)
- Amazon, How to translate product listings for Amazon Global Selling (21 February 2025)
- Amazon Seller Forums (Europe), New product compliance coming in 30th September 2024 and Amazon, Manage Your Compliance
Find out what your catalogue is missing
If you have a few hundred products and no clear view of whether the data is enough to launch, do not guess. A catalogue review answers one question: what blocks publication today, what will slow it down and what can safely wait until after launch. It also gives you the list of products that could go live immediately, as a first batch.