If you read any popular Amazon optimization guide in the last three years, you'd come away thinking image quality is one of the top three things to fix on a listing. It isn't. We're going to make a case for caring about images less, while caring more about three specific image decisions.
The disconnect comes from two things. First, image production is the most visible part of listing work — designers can show before/after; copywriters can't. Second, the people selling image services have an incentive to inflate the importance of image services. The actual A10 signal is small. The CVR signal is real but concentrated in specific places.
We pulled the image-related variables off 6,847 listings on our panel and ran them through the same regression machinery we used on title length and bullet patterns. Here's what shook out.
§ 01What we measured
For each listing we logged 11 image variables: total image count, presence of an infographic, presence of a comparison image, presence of a lifestyle shot, presence of a packaging/unboxing image, average image resolution, white-background compliance on the main image, presence of a size-reference cue, video presence, main-image product fill percentage, and main-image text presence.
Then we ran the same regression we run on every dispatch: rank delta as a function of the variables, listing-level fixed effects, controls for product category and price tier. The image-quality composite came out at 0.06 inferred weight. That's about a third of title relevance (0.22) and a fifth of CVR-14 (0.32).
Worth noting before we go further: this is the directweight. Images definitely affect rank indirectly via CVR. We'll cover that separately.
§ 02What actually moves CTR
The main image is the only image a shopper sees before clicking your listing. So if there's an image lever that matters for CTR (and therefore for downstream rank via CVR-14), it's the main.
We split-tested 280 main-image variants over the past three months. The variables that moved CTR meaningfully:
| Variable | Tests | CTR lift | Verdict |
|---|---|---|---|
| Product fills 70%+ of frame (vs <50%) | 82 | +19% | Real lift |
| White background (vs colored) | 44 | +14% | TOS + lift |
| No text overlay (vs text) | 62 | +11% | TOS + lift |
| Single product (vs multi) | 38 | +4% | Marginal |
| 3D render (vs photo) | 26 | +2% | Noise |
| Drop shadow added | 28 | +1% | Noise |
Three real wins. The first is the simplest: scale the product up in the frame. Most listings have the product at 30-50% fill, surrounded by white space. Cropping tight pushes the product into the "I can see what this is" zone faster, and shoppers scanning a results page click faster.
The second and third are TOS-compliance: Amazon requires pure white (RGB 255/255/255) backgrounds and zero text on main images. The enforcement is inconsistent, so some listings get away with violations. Our data says the violations don't even help: the listings that played it safe also won the CTR test. Two birds.
We zoomed the product from 35% frame to 78% frame on our hero listing. Same product, same lighting. CTR went up 23% in three weeks. The fix took our designer 20 minutes.
— Panel seller, kitchen brand
§ 03The image-slot economics
Once a shopper clicks through, the secondary images do the work of conversion. We tracked CVR by total image count to find the diminishing-returns point.
| Total images | Median CVR-14 | % of 7-image baseline | Marginal lift |
|---|---|---|---|
| 1 | 5.4% | 60% | — |
| 2 | 6.4% | 71% | +18% |
| 3 | 7.5% | 83% | +17% |
| 4 | 8.0% | 89% | +7% |
| 5 | 8.3% | 92% | +4% |
| 6 | 8.7% | 97% | +5% |
| 7 | 9.0% | 100% | +3% |
The 1→2 image jump is huge (+18% relative CVR). The 2→3 and 3→4 jumps are still significant. Then the curve flattens. Slot 5 onwards adds 3-5% each, mostly defensive (preventing the listing from looking sparse).
CVR a 4-image listing achieves vs a 7-image listing. The marginal CVR gain from images 5, 6, and 7 totals roughly 11 percentage points — meaningful, but small enough that for sellers with limited photography budget, fixing image 1 and getting to image 4 is the right priority.
ListFocal panel
§ 04What should be in each slot
Based on the CVR contribution of each image type when present versus absent (controlled for image count), here's a slot-by-slot recommendation that should hold across most consumer categories:
Image 1 (main): Hero shot, product fills 70%+ of frame, pure white background, no text. This is the only image used in search results.
Image 2 (infographic): An infographic that answers the top return-driving objection. For phone accessories: compatibility chart. For kitchen: dimensions. For supplements: serving size and key ingredient. The infographic should be readable on mobile (most listings are viewed on phones).
Image 3 (lifestyle): Product in context with a human or familiar object that establishes scale. Don't make it abstract; make it specific. A travel coffee mug being used in a car, not a coffee mug floating on a beach.
Image 4 (feature highlight): One specific feature shown clearly. Material close-up, control panel detail, fit demonstration. This is where you make the unique-value proposition visual.
Image 5 (size/scale): Dimensional reference, ideally next to a common object the buyer can mentally calibrate against.
Image 6 (use case): Different use case or compatibility example. Or a second lifestyle in a different setting.
Image 7 (brand/trust): Brand logo, certifications, packaging, or warranty graphic. This is the "we're a real brand" closing image.
Optional: video. We tested video presence vs absence on 88 listings. CVR lift averaged 4.2% — smaller than expected. Video's real benefit is time-on-page (and arguably session quality, which feeds into CVR-14 indirectly). For sellers with budget, video is worth doing. For sellers without, skip it and put the same money into a better infographic.
§ 05What we couldn't replicate
A few popular image "best practices" we couldn't reproduce in our data:
3D renders vs photos. Some consultants insist 3D renders are now winning over photos. Our data showed essentially zero CTR difference (within noise). Use whichever your team can produce cheaper and faster.
Drop shadows.Adding subtle drop shadows for "depth" — we couldn't find a CTR lift in 28 tests. Some designers swear by it. We'd call it cosmetic.
Carousel-style first image with multiple angles.The compositions that pack three product angles into the main image consistently underperformed simple single-angle shots in our split tests. The shopper's eye gets confused.
§ 06The image investment that actually pays back
If we ranked image investments by ROI for the average panel seller:
1. Re-shoot or re-crop the main image.Make the product fill 70%+ of the frame. If you're currently at 40%, this is the highest single-listing CVR move available.
2. Add an infographic that answers the top return reason. Pull your returns data, find the #1 reason (fit, compatibility, size, expectations), build an infographic that pre-empts it. Lower returns has its own A10 weight, and the CVR lift is sticky.
3. Add a single lifestyle image with human scale.Not abstract. Concrete. Show somebody actually using the product in a context that matches your target buyer's reality.
4. Get to 4 images minimum.If you're below 4, every additional image to image 4 gives a real CVR lift. After 4, diminishing returns.
5. Add video, if budget permits. Small CVR lift, decent time-on-page lift. Probably worth it long-term.
6. Add a 5th, 6th, 7th image. Defensive. Each gives 3-5% relative CVR. Worth doing for top-revenue SKUs, low priority for tail SKUs.
§ 07What we didn't measure
Two big gaps in our analysis you should know about.
First, we don't measure image quality at the pixel level. Some sellers obsess over color calibration, micro-contrast, and lighting consistency. Those things probably matter for premium brands competing on perceived quality. Our data couldn't isolate the effect because there's no clean way to quantify "visual polish" in a regression.
Second, we didn't isolate Amazon's newer image-recognition behaviors. Amazon's search increasingly uses image content (not just keyword text) to match listings to queries. We suspect this matters more for visual-discovery categories (apparel, home decor) than for our consumer-electronics-heavy panel. If you're in apparel, expect image content to matter more than the numbers here suggest.
§ 08A 7-day image audit
For sellers acting on this:
Day 1. Open your top 5 listings. Screenshot the main image. Estimate product-fill percentage. Anything under 60% is a candidate for re-crop.
Day 2. Pull your returns data. Identify the top return reason for each listing. Sketch an infographic that addresses it.
Day 3-4.Re-crop main images or commission new shots for the under-filled ones. Re-build infographics where the existing one isn't answering the return reason.
Day 5. Upload changes to one listing first as a control. Hold for 14 days before applying to the rest. Image changes settle in about 7-10 days; you want to see the CVR signal before scaling.
Day 6-7. Plan the lifestyle shoot. One human, one product, one context. Specific not abstract.
That's the meaningful image work. Everything else is decoration.
This closes out our six-part series on the post-May-2026 A10 weights. Next week we're starting on category-specific deep dives — first up, vlogging microphones (it's the panel category we have the most data on and the one where post-rewrite competition is reshuffling fastest).
Cite this work. Figures licensed CC-BY-4.0. Quote any passage with attribution to ListFocal.
Field notes by email, Saturdays.
Algorithm shifts and category dynamics. Free. Unsubscribe in one click.