Marketplace image compliance at catalog scale: catching failures before they publish

Marketplace image requirements are relatively easy to find. Every marketplace publishes its image requirements somewhere, so you can find Amazon's, Walmart's, eBay's, Etsy's, Shopify's, and Google Shopping's requirements within minutes, add them to a shared document, and feel like your work is done.

But knowing the spec is only the beginning. If you have one master image feeding six channels, you already have six files per SKU before you count secondary angles or regional variants. At 500 SKUs, that’s 3,000 files. At 50,000, that’s 300,000 files, each of which must meet different combinations of dimensions, aspect ratio, file-size cap, background rule, and color profile. For brands, retailers, and distributors, the number grows further once you add supplier images, regional storefronts, and seasonal refreshes.

That’s a lot of derivatives. Manually inspecting every file at that volume is impossible, meaning mistakes can slip through. And the consequences vary by platform, ranging from listing suppression to unpublishing.

The answer isn’t another sprawling spreadsheet crammed with image specs for your team to memorize. Instead, it’s an automated workflow that handles marketplace image compliance on your behalf, with just the right amount of human oversight, and no more. With the right workflow, you can keep up at scale, generating the right derivative for every SKU and channel, catching failures before publication, and adapting as marketplace requirements change.

Table of contents

What are the current image requirements across major marketplaces?

Before you start creating images for different marketplaces, you need to understand what each one requires. Platforms regularly change their image rules, whether that means changing accepted parameters or adding entirely new requirements.

Here’s a quick look at where the specs stood for some of the most popular marketplaces as of September 24, 2026.

Note that these requirements don’t stay fixed. Treat this table as a working reference, but keep checking each marketplace’s first-party documentation so your image workflow stays aligned with the current rules.

What happens when one image has to meet different requirements?

The differences in the table are why a single image file shouldn’t be expected to serve every channel unchanged.

Walmart, for example, recommends a 2200 × 2200 image with a pure-white background, while Zalando recommends a 1801 × 2600 JPG at a 1:1.44 aspect ratio. Instacart, meanwhile, recommends 1000 × 1000 and caps files at 3MB.

Those requirements don’t necessarily mean you need three different source images. It just means you need a sufficiently rich master image that can produce the right derivative for each destination and the right tools and workflow to generate, validate, and route those derivatives reliably at scale.

How do you reformat one product image to meet every marketplace’s spec at scale?

Start with one high-resolution, content-rich master image. It should be designed for reuse rather than optimized for one marketplace. Give it enough resolution and crop headroom to support different aspect ratios and placements without cutting off the product or important details.

Then, create a named export recipe for each marketplace. Each recipe should define the dimensions, aspect ratio, format, file-size limit, background requirements, and additional channel-specific rules the derivative image needs to satisfy.

With an AI-powered image workflow, you can apply those recipes to the master and generate the required derivatives without manually rebuilding each file. For example, your master image could automatically spin off into:

• A Walmart derivative at 2200 x 2200 pixels, with the required white background and a file size of 5 MB or less.

• An Amazon derivative that stays within Amazon’s pixel limits and uses the required pure-white main-image background.

• An Etsy derivative at 2000 x 2000 pixels in a landscape or square ratio.

• A Zalando derivative at the required portrait-oriented 1:1.44 ratio.

Where do you add validation gates?

Creating images and generating derivatives is only part of the job. You also need to check that your image meets the right requirements before you multiply it into derivatives or send those derivatives to a marketplace.

That means validation belongs in two places: when the image first enters your system, whether that’s from a studio photograph, seller-submitted image, or user-generated asset, and after you’ve generated the channel-specific derivative but before publication.

At ingest, you’ll need to check that the initial image is usable because one bad image can lead to many unusable derivatives. This is especially important when images come from different external sources. Does it have sufficient resolution and crop headroom? Is the background appropriate? Is the color profile correct? Is the file in an accepted format? Is the image sharp enough to support the channels you need to serve? Catching those problems here prevents you from building a library of derivatives on top of a bad source.

Once you’ve generated a derivative, you’ll also want to validate it against the specific channel recipe before publication. This catches any problems introduced during resizing, cropping, or compression.

The checks can be divided into two broad groups.

Deterministic checks

These are conditions a system can evaluate directly, such as:

• Pixel dimensions.

• Aspect ratio.

• File format.

• File size.

• Color profile.

• Transparency.

• Background color.

• Filename requirements.

These checks are good candidates for hard blocks (meaning the image can't progress to the next stage in the workflow if they fail these checks) because there is little reason to send an objectively non-compliant file to a marketplace.

Visual checks

Other requirements require a different kind of evaluation.

Is the product blurry? Is it sufficiently visible? Has the crop removed part of the product? Does the background meet the marketplace's expectations? Is there unwanted text, a watermark, or another object in the frame?

Those conditions are less binary. Rather than treating every visual assessment as an automatic pass or fail, the right workflow can route uncertain cases directly for human review. It doesn't explicitly judge pass or fail, but the likelihood of the image being a fail, so it can redirect it for manual review if necessary. This keeps manual review to a minimum while making the risk of a non-compliant or low-quality image making it through to publication as low as possible.

How do you handle supplier and vendor images?

At enterprise scale, many source images come from outside your team: suppliers, agencies, regional offices, and marketplace sellers. You can't control how they were shot, so you need to control what gets in.

Give suppliers a short, written image brief covering minimum resolution, accepted formats, background, color profile, and naming. Run every upload through the ingest checks above. When a file fails, return it with the specific reason and what to change, rather than fixing it quietly on your side. Suppliers then learn the standard, and the same failure stops coming back.

What should you do when an image fails?

A good validation workflow needs a defined reject path. When a file fails a check, don’t simply flag it and leave someone to figure out what happens next. A comprehensive workflow should capture the failure, explain which rule was violated, assign an owner, and facilitate regeneration and rechecking of the file.

For example, if a derivative fails because it is 6 MB and the channel allows only 5 MB, that can be handled automatically through compression and revalidation. However, if the source image is too small to produce the required derivative without losing important product detail, the workflow should send it back for a new source rather than repeatedly trying to resize it.

In general, there are three ways to handle failures:

• Correct automatically: Use automation when the fix is deterministic and doesn’t require a judgment call. For example, a derivative that exceeds a file-size limit can be re-exported at a lower quality setting. In these cases, the system knows what the requirement is and which action will bring the file into compliance.

• Block and return: Some failures can’t be fixed by manipulating the derivative because the problem is with the source itself. If a supplier image is below the required resolution, for example, no amount of resizing will create the missing detail. Generating derivatives from it simply creates more files that will fail. Here, the best course of action is to block the image at ingestion and return it with a specific reason and a clear request for what needs to change. That lets the supplier or source team correct the problem once, before it gets multiplied across channels.

• Escalate to review: Sometimes, there are failures where the system can identify a potential problem but can’t reliably make the final call. For example, a product may be too close to the edge of the crop, a watermark may be suspected, or a background may be technically close to white without clearly meeting the requirement. Those cases should go to a named reviewer who can approve the image, reject it, or make the required correction before feeding the result back into the workflow.

The point is to give every failure somewhere to go. Automatically fix what can be fixed, return what cannot be fixed from the source, and send ambiguous cases to a person who can make the final call.

What happens when a marketplace rejects an image?

Your own validation gates can catch a lot of problems before publication, but they can’t tell you everything a marketplace will do with an image once it gets there. Platform responses are uneven. Some failures generate a clear error or appear in a dashboard. Others may result in an item being unpublished or no longer appearing as expected, without giving you the kind of detailed image-level explanation you’d want.

Amazon, for example, will suppress images for text, logos, graphics, watermarks, non-white backgrounds, additional items included in the image, a cropped product, blurriness, pixelation, and more. It doesn’t offer a timeline, but you can view which products were removed from search under the "Fix" tab in the "Manage Products" view. You can also view the files themselves under "Submission Status" and then the "Content Compliance Issues" report type.

Different platforms handle image rejections differently. For example:

Keep a close eye on marketplace diagnostics, disapprovals, and listing status, as even an image that passed through your internal validation gates can still run into platform-specific enforcement after publication.

How do you govern image compliance across teams?

With several teams and regions publishing to several marketplaces, ownership gets blurry fast. Set up four things:

• One owner per export recipe. One named person approves changes to it.

• A change log. Record when each marketplace's rules were last checked and who checked them.

• A rollout rule. When a spec changes, update the recipe once and regenerate affected files, instead of letting each team patch its own copy.

• Two numbers to track. Track the share of images a marketplace rejects, and how long a fix takes from rejection to republish. Both show whether the workflow is working.

How can Photoroom help enterprises create and check images for every marketplace?

At marketplace scale, the challenge isn't creating one good product image. It's turning one reference image into every file each channel needs, checking those files before they go live, and handling anything that fails. Photoroom's enterprise platform covers that whole loop.

• Channel-ready derivatives. Set your sizes, formats, backgrounds, and brand rules once with Brand Kit, then apply them across the catalog. Process thousands of images in one job, from the web app or the API.

• Visual QA. Every output is scored for fidelity against the criteria you set. Misses are caught and corrected before they reach your customers.

• Visual Fix. Failed outputs are retried with adjusted settings or different models. Cases that need a person are set aside for human review.

• Visual Agents. Agents run the full workflow: analyze each input, generate or edit, score the result, and retry until it passes. Only images that pass reach your catalog.

• Enterprise Guarantee. You pay only for outputs that pass your fidelity criteria. Anything that misses is credited back or regenerated.

It's not only about compliance. From the same reference product image, your team can create ghost mannequin images, on-model imagery with virtual models, and short product videos, without a new shoot. One source image becomes the catalog look, the on-model look, and the video each channel needs.

You set the marketplace rules, and Photoroom applies and checks them at scale. Talk to our team about automating image creation and quality checks across your catalog.