The main image: Amazon's rules, rejection reasons and what's still possible
- You know all hard main-image rules and produce compliant images on the first try.
- You recognize the most common rejection reasons and know each fix.
- You know where AI-created product images stand — and where misleading begins.
- You exploit the legal headroom that lifts your click-through rate.
The main image decides your click in search results — and it's the most strictly regulated element of your listing. Amazon checks main images automatically: violations lead to image rejection or suppression of the whole offer. The rules are strict but clear — and whoever knows them uses the remaining headroom better than the competition.
1The hard rules
- Pure white background: RGB exactly 255, 255, 255 — not “light grey”, not “almost white”. The classic rejection reason.
- The product fills the frame: guideline ~85 % of the frame in the longest dimension. Small products on lots of white look like postage stamps in search results.
- Resolution: Amazon accepts 500 to 10,000 px on the longest side; from 1,000 px the zoom function activates — never go below that. Zoom works properly from 1,600 px+ (common practice recommendation; Amazon itself says “more than 1,000 px”). In practice: 1,600–3,000 px, square, sRGB, JPEG — and zoom sells: customers inspect material and details with it.
- Only the product: exactly what the customer receives — correct quantity, correct color, the complete set (show all parts!). No accessories that aren't included, no props.
- Forbidden on the main image: text, logos, badges (“Bestseller!”), watermarks, graphics, borders, collages/multi-views, people (exception: worn apparel in fashion categories), visible packaging (unless the packaging IS the product).
- The product must be fully in frame (no cropping) and sharp, well lit, noise-free.
Applying the rules to the AURELO spice mill set (€24.99): both mills side by side, slightly offset, square at 2,000 × 2,000 px, background at exactly RGB 255,255,255, the mills filling around 85 % of the frame height. No “set of 2” badge, no little pile of peppercorns as a prop — the image proves the quantity itself, because both mills are fully visible.
2Rejection — and the fix
| Rejection reason | Fix |
|---|---|
| Background not pure white | Cut out cleanly and set to RGB 255,255,255 — don't just “brighten” |
| Text/badge on the image | Move all lettering to the infographic image (image 3–5) |
| Product too small in frame | Crop tighter to ~85 % fill |
| Props/decoration in frame | Show only what's included; decoration goes to lifestyle images |
| Collage/multi-view | One main view; other angles into the gallery |
| Low resolution / blurry | Recreate at 1,600+ px; don't upscale tiny photos |
3Are AI product images allowed?
The honest answer in three sentences: Amazon requires images to show the sold product truthfully — no guideline prescribes HOW the image was made (camera, studio, AI processing). AI-powered cut-outs, background generation and image preparation from real product photos are common practice and unproblematic. The red line is misleading: a fully generated fantasy product that differs from what's delivered (shape, color, details) violates policy — and guarantees returns and 1-star reviews. In short: AI as a tool, yes; AI as an inventor, no. (That's exactly how Listimo works: your real product photos become compliant images — pure-white main image included.)
AI as a tool is like retouching your own passport photo: the light and background get better, but it's still YOU in the picture. AI as an inventor would be handing in someone else's photo. Same with the main image: polishing your real product is allowed — showing a product that isn't what's in the box is misleading.
4The legal headroom: more clicks within the rules
- Maximize frame fill: the difference between 60 % and 85 % fill is dramatic in search results — your product simply looks bigger and more valuable.
- The best angle: slightly turned/from above looks more three-dimensional than a flat front view. Test which angle makes the function obvious fastest.
- Sets: show everything. A set of 2 with two visible mills beats the single-item shot — quantity is a buying argument, and showing it is even required.
- Subtle depth: a fine drop shadow or slight reflection under the product is allowed and lifts it off the white (don't overdo it).
- Make texture visible: set the light so material is recognizable — cotton weave, brushed steel. “Premium” is created by light, not by text.
- Later, with sales data: test main-image variants via experiments (Growth track) — CTR differences of 20–30 % between two compliant main images are not rare.
Before: the AURELO main image shows one mill head-on at only 60 % frame fill — CTR 1.8 %. After: both mills slightly from above, cropped to 85 % fill, a fine drop shadow — CTR 2.2 %. At 10,000 searches a month that's 220 instead of 180 visitors: 22 % more clicks without a cent of ad budget and without breaking a single rule.
“Just a small badge” — a star, a “NEW”, a mini logo on the main image. Amazon's image recognition finds it, rejection or suppression follows, and in the worst case your offer sits without a main image for days (conversion near zero). Every label belongs on gallery images — the main image stays pure.
- Background exactly RGB 255,255,255, cleanly cut out.
- ~85 % frame fill, product complete, correct quantity/color.
- 1,600–3,000 px, square, sharp, sRGB.
- Zero text, logos, badges, props, collages.
- Best angle chosen, material visible in the light.
- Sets: all parts visible.
- The image shows exactly the delivered product (AI as tool, never as inventor).
5Expert insight: Image tests without self-deception
“CTR from 1.8 to 2.2 %” sounds like a clean result — in practice, exactly this measurement is the hardest exercise of this lesson. A before/after comparison has two enemies: confounders and chance. Ignore both and you optimize your main image on noise — and may well swap a winning image for a worse one.
- Confounders: price changes, coupons, PPC budget, season and weekday patterns, a competitor out of stock, holidays — all of it moves clicks and sales more than most image swaps. Rule: what you control (price, ad budget), you freeze during the test window; what you cannot control, you log with a date in a test journal.
- One change per test: swapping image AND title at the same time means never knowing what worked.
- Measure full weeks: Monday through Sunday, at least two weeks per variant — otherwise you compare your weekend audience with your weekday audience.
Click counts fluctuate like dice rolls: the natural noise sits roughly at the square root of the click count. AURELO example with 10,000 impressions per period: 1.8 % CTR is 180 clicks (noise about ±13), 2.2 % is 220 clicks (about ±15). A gap counts as genuinely distinguishable from roughly twice the square root of the combined clicks: the square root of 400 is 20, times two is 40 — so the measured 40-click gap sits exactly on the edge. Meaning: even a 22 percent jump needs 10,000 impressions per variant to be barely reliable. Smaller differences or fewer impressions: a coin flip.
That is why the gold standard is Manage Your Experiments (with Brand Registry): Amazon splits visitors between variant A and B at the same time. Season, price environment and competition hit both variants identically — the confounders drop out as an error source, leaving only sample size. Which is also why experiments run for several weeks and require enough traffic; for low-traffic products, the long, disciplined before/after comparison remains the only route.
The most expensive mistake is the panic swap-back: new main image live, three days of “worse numbers”, back to the old image. Do the math: three days at 10,000 monthly impressions is around 1,000 impressions — at 2 % CTR about 20 clicks, with noise of ±4 to 5. Swings of 20–25 % are pure chance on that basis. Judging by days means judging dice — not images.
The first lessons of every track are open to everyone. From here on you just need a free account — no subscription, no costs.
- 2Rejection — and the fix
- 3Are AI product images allowed?
- 4The legal headroom: more clicks within the rules
- 5Expert insight: Image tests without self-deception
- ✓Quiz: 6 questions with instant feedback
- All 41 lessons in 3 tracks — completely free
- Your progress is saved across devices
- Certificate + 100 free Listimo credits per completed track
Check yourself
6 quick questions — one at a time, instant feedback. With a free account your progress is saved.
Frequently asked
How do I cut my product out onto pure white cleanly?
Three routes: image editing (selection tool + background set to RGB 255,255,255), AI cut-out services — or upload your raw photo to Listimo and get the compliant main image automatically. In all cases: check the edges, especially with hair, bristles or transparent parts.
Why is my main image flagged despite a white background?
Usually the white isn't pure (e.g. 250,250,250 instead of 255,255,255), a shadow gradient sits across the background, or the product fills too little of the frame. Zoom into the image corners: a gray tint there means the background isn't pure white.
How often may I change the main image?
Technically any time — strategically with a system: a swap affects CTR immediately, so test via experiments (with Brand Registry) or measure at least two weeks before/after. A winning image during the Christmas season is left alone.
Check your listing for free against the rules from this lesson — or have Listimo build the whole listing from one product photo: images, copy and A+ content.
Go deeper: the complete guide to optimizing Amazon listings →
Everything in this academy comes from day-to-day selling practice — the same playbook behind Listimo, the tool that turns product photos into complete Amazon listings.