The six marketplace image rejection reasons that account for almost everything
Sellers who process large catalogs across several channels tend to see the same distribution of failures. The exact percentages shift with category, but the ranking is stable: background problems dominate, followed by framing, then text and graphic overlays.
The numbers are indicative rather than a published benchmark, but the shape matters: roughly three quarters of rejections come from background, framing and overlays, and all three are things you control at the retouching stage, not at the shoot. That is good news. It means a single pre-upload pass can clear most of the queue.
The rest of this article walks through each cause in that order.
Reason 1: the background is not actually white
"Pure white" means RGB 255,255,255. Amazon states this explicitly for main images, Walmart and Target Plus follow the same rule, and Google Merchant Center's "promotional overlay" and "generic image" policies are enforced most strictly on cluttered backgrounds. The checkers sample pixels around the edges and corners of the image; if those pixels read as 248,248,248 or a warm 255,252,245, the image is not white as far as the machine is concerned, even though it looks white to you.
Common ways an off-white background sneaks through:
- Camera white balance. A sweep lit with tungsten or mixed light renders as cream or grey. Auto white balance rarely lands on neutral.
- Falloff. The centre of the sweep is blown to white but the corners are 10–15 points darker. Edge sampling catches the corners.
- JPEG compression. Saving at low quality introduces blocking around the product edge that pulls neighbouring background pixels below 255.
- "Levels" instead of masking. Pushing the white point in Lightroom until the background clips also clips the highlights on a white product, which then fails a different check for lost detail.
Open the image and read the pixel value in all four corners and along each edge. If any sample is below 250 on any channel, the background will not pass an automated white check. The reliable fix is a cutout mask and a painted 255,255,255 plate, not a global brightness adjustment.
Note the opposite failure too: sellers sometimes over-correct by dropping a product on a flat white plate with no shadow at all, producing a floating, pasted look. Marketplaces do not reject that, but shoppers respond worse to it. Keep a soft contact shadow and ensure the shadow pixels are inside the product's footprint rather than smeared across the plate edges.
Reason 2: the product fills too little (or too much) of the frame
Framing is the second biggest source of rejections and the one that varies most between platforms. Each marketplace defines a minimum fill ratio for the main image, and several also enforce a maximum so the product is not cropped at the edges.
| Marketplace | Main image fill rule | Failure symptom |
|---|---|---|
| Amazon | Product fills 85%+ of the longest side | Listing suppressed; "image does not meet standards" |
| Walmart | Product fills 70–90% of the frame | Item stays unpublished; image flagged in Item Spec report |
| Google Merchant Center | Product should occupy 75–90% | Offer disapproved or demoted: "image quality" |
| eBay | No hard ratio; 500px minimum; no borders | Reduced visibility rather than rejection |
| Etsy | No hard ratio; 2000px on shortest side recommended | Lower ranking in search; no rejection |
A single photo cannot satisfy all of these if it is framed for one channel and cropped by hand for the rest. The 85% Amazon rule in particular is unforgiving on tall, thin items like a serum bottle or a pair of trousers: a bottle framed to fill 85% of the height leaves huge empty margins on either side, and a crop tight enough to fix that then clips the cap.
The fix is to separate the product cutout from the framing decision. Once the product is isolated on a transparent layer, you can place it inside a channel-specific canvas with the right fill ratio, padding and aspect ratio programmatically, rather than re-cropping the source photograph five ways. This is also the only way to keep scale consistent across a category: a 30 ml bottle and a 200 ml bottle should not both fill 85% of their frames, or shoppers lose any sense of relative size. Our guide to Amazon image compliance and suppression covers the Amazon end of this in detail.
Reason 3: text, badges, watermarks and borders
Anything that is not the product is a candidate for rejection on a main image. Amazon prohibits text, logos (other than on the product itself), watermarks, badges like "Best Seller" or "Free Shipping", borders, and colour swatches. Google Merchant Center's promotional overlay policy blocks the same set, and since 2024 its automated image checks will also flag images where a large portion of the frame is a graphic rather than a photograph. Walmart's rules mirror Amazon's.
The cases that catch experienced sellers are subtle:
- Photographer watermarks in a corner at 10% opacity. Invisible to you, obvious to a detector trained on them.
- Packaging text that is actually on the product. This is allowed, but a badge sticker applied for the shoot ("New formula!") is not, because it is not part of the shipped product.
- Thin borders added by a template or an old export preset. A one-pixel grey line around the image is enough to fail both a border check and a white-background check at once.
- Size charts and colour swatches composited into the main image for convenience. These belong in secondary images.
Amazon and Google often report only the first rule an image fails. An image with a watermark, a grey border and 82% fill will come back with one reason. Fix that one, re-upload, and you get the next. Run the full checklist once rather than iterating through the marketplace's queue.
Reason 4: resolution, file size and format
Resolution rejections are the easiest to fix and the most embarrassing to receive, because they are entirely deterministic. The thresholds:
| Marketplace | Minimum | Recommended | Formats |
|---|---|---|---|
| Amazon | 1000px longest side (for zoom) | 1600px+ longest side | JPEG, PNG, TIFF, GIF (JPEG preferred) |
| Walmart | 1000 × 1000px | 2000 × 2000px | JPEG, PNG |
| Google Merchant Center | 100 × 100px (250 for apparel) | 800 × 800px+ | JPEG, PNG, WebP, GIF, BMP, TIFF |
| eBay | 500px longest side | 1600px+ longest side | JPEG, PNG, GIF, BMP, TIFF, HEIC |
| Etsy | No hard minimum | 2000px shortest side | JPEG, PNG, GIF |
Where sellers go wrong is not the source file but the export. A 6000px camera original gets downscaled for a website, that 800px web copy is later re-used for a marketplace feed, and the upload fails the zoom threshold. The safer rule is to store one master at full resolution and derive every channel version from that master at upload time, never from a previous derivative.
Two other format traps: PNG files with transparency are accepted on most platforms but rendered on a background you do not control (Amazon flattens to white, some eBay views flatten to grey), so flatten to white yourself. And CMYK JPEGs exported from print workflows are rejected outright or render with shifted colours — always export sRGB. Our compression and quality loss guide covers how much you can compress before edge artefacts start failing the white-background check.
Reason 5: the main image shows something other than the product
The main image must show only the item for sale, exactly as it ships, with nothing else in the frame. That rule generates rejections for:
- Props and styling. A serum bottle beside a towel and a sprig of eucalyptus is a lifestyle shot. It belongs in slot 2 or later.
- Models, in categories that ban them. Amazon allows models for adult apparel but not for accessories, children's clothing (except on a child model, not a mannequin), or most non-apparel categories. A watch shown on a wrist fails as a main image in some categories and passes in others.
- Mannequins. Visible mannequins are prohibited for apparel main images on Amazon; invisible (ghost) mannequin shots are fine.
- Multiple units or variants. Showing all five colourways in one image when the listing is for a single colour is a "does not match the product" rejection. This is also the most common cause of variant-level image problems on Shopify and Google feeds.
- Packaging alone. Boxed products are accepted in some categories (electronics, cosmetics) but a box with no product visible often fails apparel and home rules.
The practical implication is that a catalogue needs two kinds of imagery per SKU: one strict, product-only, white-plate image for slot 1, and then the lifestyle, on-model and detail images that actually sell. Sellers who shoot only lifestyle and then try to crop a main image out of it are the ones who see this rejection repeatedly.
Shoot-then-crop
- One lifestyle set per SKU
- Main image cropped from a styled shot
- Props and shadows bleed into the frame
- Reshoot needed when the crop fails
Cutout-first
- Product isolated once, on a transparent layer
- Main image generated on a 255 white plate with channel-specific framing
- Same cutout reused for lifestyle backgrounds and colour variants
- No reshoot; re-render the frame instead
This is the workflow most catalogue teams have moved to, and it is the one AI background-removal tools such as Retouchable are built around: a clean cutout, a deterministic white plate, and a framing step that applies the fill ratio per channel rather than per photo.
Reason 6: images that do not match the listing, or each other
The last group is less about a single image and more about the set. Marketplaces increasingly compare images to the product data and to one another:
- Colour mismatch. The listing says "navy", the image reads as black because the shot was underexposed. Amazon's and Walmart's return-reason data flags "item not as described" against the ASIN and it can be suppressed for high return rates rather than for the image itself.
- Duplicate images across variants. Using the red image for every size and colour of a variant group triggers "image does not represent the variation" on Amazon and a generic-image disapproval on Google.
- Inconsistent framing across a set. Not a hard rejection anywhere, but Google's image quality signals and Amazon's search ranking both favour sets where the product is presented at a consistent scale and angle.
- Stale images after a product change. The packaging was redesigned, the listing still shows the old box. Customer complaints, not automated checks, drive these rejections, and they are slower to resolve.
The defence here is a colour-managed pipeline (a grey card or colour target in the master shot, correction applied before cutout) and a naming convention that ties every derived image back to its SKU and variant. When an image is regenerated, every channel picks up the new version rather than a stale copy.
A pre-upload checklist that clears the queue
Working through the six causes in order, here is a checklist you can apply to every main image before it goes to any marketplace. It takes under a minute per image by hand, and can be fully automated.
- Sample the background at all four corners and mid-edges. All channels ≥ 250, ideally 255.
- Measure the product's bounding box. Confirm the fill ratio for the strictest channel you sell on.
- Look for anything that is not the shipped product: overlays, borders, props, stray shadows past the frame edge.
- Check dimensions and colour space. Export from the master, not from a web derivative.
- Confirm the image matches its variant: colour, size where visible, packaging version.
- Check the set: same scale, same angle, same plate colour across the family.
Teams that build this into the retouching step rather than the upload step see rejection rates drop to near zero, because every image is generated to spec rather than checked after the fact. For the exact numbers per platform, keep the cross-marketplace requirements guide open alongside this one.