The Amazon comparison image — the us-vs-them graphic sitting in slot 5 or 6 of your carousel — is the most copied and least understood image in the stack. I've reviewed 50,000+ listings, and the comparison slot shows up on maybe 40% of them. Most of those graphics are doing one of two things: nothing, or damage.
Here's the tension nobody talks about. A comparison image is the only slot in your carousel that exists because of other products. Every other image sells your product. This one sells your product against the alternative the shopper is already holding in a second tab. Done right, it's the image that closes a comparison shopper who was about to leave. Done wrong, it's a green-checkmark-red-X table that reminds the shopper there ARE alternatives — and hands them the exact spec list to go evaluate them with.
This is the full deep-dive: when the comparison slot earns its place, when it should be cut, the design rules that survive a 280-pixel thumbnail and a phone screen, the TOS landmines that get listings suppressed, and the audit I run on every comparison image that crosses my desk.
Note the scope: this is about the carousel comparison image, not the A+ comparison chart module. I wrote about the A+ module separately — it's a different tool with a different job (it links real ASINs and lives below the fold). The carousel version is a flat graphic in your image stack, it appears in the first scroll of a mobile PDP, and it plays by image rules, not module rules.
What the comparison image is actually for
Strip away the template and the comparison image has exactly one job: answering the objection "why this one and not the cheaper/more popular/more familiar one?" without the shopper leaving your page.
That framing matters because it tells you who the image is for. It is not for the impulse buyer — they never get past slot 3. It is not for the brand-loyal repeat buyer — they're skipping straight to checkout. It's for the considered shopper mid-comparison: the person with three tabs open, or the person who searched a generic keyword, clicked you AND two competitors, and is now toggling.
On listings where I've been able to isolate behavior, that mid-comparison segment is where the slot earns its keep. The pattern I see repeatedly: comparison images don't move CVR much on listings that dominate their niche (there's nothing to compare against), and they don't save listings that lose on the fundamentals (price gap too wide, rating gap too wide — no graphic fixes a 3.9 against a 4.6). The band where they work is the middle: competitive categories, comparable price points, a real differentiator that's invisible in a photo. In that band, adding or fixing a comparison image is routinely worth 3-8% relative CVR — which on a $100K/month ASIN is real money for one graphic.
When the comparison slot pays — and when to cut it
Four conditions. You want at least three before the slot earns carousel real estate:
1. The category has a default alternative. Garlic press, phone mount, dog harness — categories where the shopper's mental model is "they're all basically the same." A comparison image exists to break that assumption. If your category has no default (novel products, gifts, one-of-one designs), there's nothing to compare against and the slot is wasted.
2. Your differentiator is invisible in a photo. Thicker gauge steel, double-stitched seams, 2x the active ingredient, a sealed bearing where competitors use a bushing. If the difference photographs (a visible feature, a size advantage), show it directly — a photo beats a chart every time. The comparison image is for differences a photo can't carry.
3. You're losing the toggle war. Pull your Search Query Performance data. If your impressions and clicks are healthy but purchases leak to two or three specific competitors on the same queries, shoppers are comparing and choosing them. That's the exact behavior this slot interrupts.
4. You can win the comparison honestly. This one gets skipped constantly. If your honest grid is 4-for-6 against the category default, fine. If you have to dilute it with filler rows — "premium quality ✓," "satisfaction guarantee ✓" — to manufacture a win, the shopper smells it, and you've spent a carousel slot announcing that your real advantages are thin.
When fewer than three conditions hold, cut the slot and give the space to something that works harder — a second lifestyle context, a dimensions/contents image, social proof. The carousel is seven to nine slots and every one is contested real estate. "Everyone in my category has a comparison image" is not a reason. It's usually the opposite: in a category where every listing runs the same green-check template, the listing that doesn't look like a spreadsheet stands out.
The three formats, ranked
Format 1: The two-column us-vs-them. Your product, one generic alternative ("ours" vs "other brands"). Photo of your product on your side; a desaturated, generic representation on theirs. Three to five rows, each row a benefit, not a spec. This is the workhorse, and when I A/B comparison formats, the two-column wins most often — because it matches how the shopper is actually thinking: this one, or that one.
Format 2: The feature-callout hybrid. Not a table at all — your product photographed large, with two or three callouts that each make an implicit comparison: "Sealed bearings — most use plastic bushings." "4mm gauge — 2x thicker than standard." This is my preferred format when only one or two differentiators matter. It reads as product education, not as defensiveness, and it survives mobile shrinkage far better than any grid.
Format 3: The multi-column matrix. You versus three or four unnamed competitors, five-plus rows. I'll be blunt: in the carousel, this format loses. It's an A+ module pattern crammed into an image slot. At thumbnail size it's an unreadable grid of marks; on a phone it demands pinch-zoom, and pinch-zoom is where conversions go to die. If you genuinely need a multi-way matrix, that's your signal it belongs in A+ — not the stack.
Design rules that survive 280 pixels
Every rule here comes from the same constraint: 70%+ of your traffic is on a phone, and the carousel image renders small before it renders big.
One winning row is worth five rows. The shopper gives this image two seconds. The question isn't "how many advantages can I list," it's "which single advantage closes the toggle war?" Lead with it — top row, biggest type. Rows two through four exist to make row one credible, not to compete with it.
Benefits in the rows, not specs. "Won't crack in cold weather" beats "UV-stabilized ABS polymer." The us-vs-them format already provides the proof structure; the row language should be the outcome. Specs go in the bullets and the attribute fields where the AI shopping surfaces can read them — a flat image is invisible to Rufus, which is one more reason this graphic should persuade humans, not catalog features.
Your side gets the photography, their side gets the silhouette. The fastest visual shorthand for "us vs them" isn't checkmarks — it's color and fidelity. Your column: full color, real product photo, brand accent. Their column: grayscale, generic outline. The shopper should know which side is winning before they read a word. That's not a trick; that's hierarchy doing its job.
Minimum text size: legible at 50% zoom. My desk test: shrink the image to half size on your monitor. If any row label is hard to read, it's invisible on a phone. In practice this caps you at about 12-15 words of row text total. The matrix format dies right here, every time.
Don't lead with red X's. A grid that's visually dominated by the competitor's failures reads as negative campaigning, and shoppers discount it the same way voters discount attack ads. The eye should land on what yours does. I've watched check-dominant layouts beat X-dominant layouts repeatedly in preference polls — same data, same rows, different emotional read.
The TOS landmines
This slot generates more compliance trouble than any other image in the stack. The rules that matter in 2026:
Never name a competitor brand. Not in text, not in a logo, not in a recognizable trade-dress silhouette. "vs. Brand X" with the actual brand inferable gets you a listing suppression and, increasingly, a counter-complaint from the brand itself. "Other brands" / "standard versions" / "typical models" is the ceiling.
Never use a competitor's product photo. Their images are their copyright. Even photographing a competitor unit you bought is a gray zone if the product is identifiable — generic render or stylized silhouette is the safe lane.
Every claim in the grid is a claim you must be able to support. "2x thicker" needs a measurement. "Lasts longer" needs substantiation. Amazon's enforcement on image claims has tightened every year, and category teams pull listings first and ask questions later — supplements, topicals, and anything safety-adjacent get the strictest read. The honest-grid rule from earlier isn't just ethics; it's account hygiene.
Watch the superlatives. "#1," "best," "top-rated" in a comparison graphic without an Amazon-issued badge to back it is a takedown waiting for a competitor with a reporting habit. Your competitors are auditing your images. In contested categories, assume every claim gets adversarial review.
Where the comparison image sits in the stack
Slot placement is about matching the shopper's objection sequence. By slot 2-3 they understand what the product is; by slot 4-5 they understand how it works and what it includes. The comparison question — but is it better than the other one? — arrives late. That's why the comparison image belongs in slot 5-7, after identification and education, before (or alongside) social proof.
Putting it early is the common mistake, and it's worse than wasted — a slot-2 comparison image plants "there are alternatives" in the mind of a shopper who hadn't been comparing at all. You introduced the objection you were trying to answer. The carousel's front half builds desire for this product; the back half closes doubts. Comparison is a back-half image, full stop.
One sequencing note: comparison and social proof are a natural pair. The us-vs-them makes a claim; a review-derived proof point ("4.7 from 12,000 owners") backs it. On stacks where I run them adjacent, the pairing consistently outperforms either image isolated in a different order.
Anti-patterns: the nine ways this image goes wrong
- The template grid. Same green checks, same red X's, same five rows as every competitor — sometimes from the same Fiverr template. In a category where four listings run it, the comparison image stops comparing.
- The dishonest sweep. Six rows, six wins, including rows nobody loses ("easy to use ✓"). Shoppers aren't innumerate. A 6-0 grid against the whole category reads as fiction.
- The spec dump. Eight rows of material codes and certifications. That's an A+ chart's job. In the carousel it's wallpaper.
- The pinch-zoom matrix. Covered above — if it needs zoom, it's already lost.
- The early placement. Slot 2 comparison = introducing competitors to a shopper who wasn't thinking about them.
- The brand-name callout. Compliance time bomb, even when it's accurate. Especially when it's accurate.
- The X-heavy attack ad. Visual negativity that drags the listing's whole feel down. You're selling yours, not prosecuting theirs.
- The orphan differentiator. The grid claims "2x stronger" but no other image, bullet, or A+ module ever mentions strength again. Unsupported claims read as decoration. Your stack is one argument; the comparison image is a beat in it, not a standalone flyer.
- The stale grid. Built in 2024, never revisited; meanwhile the category default added the feature you "win" on. A wrong comparison is worse than none. Re-verify the rows quarterly — it's a 15-minute job.
The 6-step comparison slot audit
Run this on your own listing this week:
- Earn the slot. Score the four conditions (default alternative, invisible differentiator, toggle-war evidence in SQP, honest win). Fewer than three? Cut the image and re-purpose the slot.
- The two-second test. Show the image to someone for two seconds. Ask what the product wins on. If they can't name the top-row advantage, the hierarchy failed.
- The 50%-zoom test. Half-size on desktop. Every row label legible, or the format is wrong for the slot.
- The honesty pass. Delete every row you couldn't defend to an Amazon category reviewer with data in hand. If fewer than three rows survive, switch to the feature-callout hybrid — two strong callouts beat four weak rows.
- The compliance sweep. No brand names, no competitor photos, no unbackable superlatives, claims substantiated. Check it like a competitor would, because one will.
- The stack-coherence check. Does the winning row echo somewhere else — a bullet, the A+ chart, a review quote image? One claim, three surfaces. That's when it starts compounding.
FAQ
Do I need both a carousel comparison image and an A+ comparison chart? They're different tools. The carousel image is a persuasion asset for the first-scroll mobile shopper: one winner, three to five benefit rows, two-second read. The A+ chart is a navigation-and-detail asset for the researcher: real ASINs, real specs, cross-sell within your own catalog. Strong listings in considered categories often run both — but the carousel version is the one that earns its slot against tougher competition, so audit it harder.
Can I show a real competitor product if I blur the logo? Don't. If the product is identifiable by shape or trade dress, blurring the logo doesn't make it generic — it makes it look like you're hiding something, and it doesn't protect you from a complaint. Stylized silhouette or generic render, always.
What's the right number of rows? Three to five for a two-column. The real answer is "however many you can win honestly while staying legible at 50% zoom" — which in practice means the row count is set by your weakest defensible claim and your smallest screen, not by how much you'd like to say.
My product is more expensive than the category default. Should the grid include price? Yes — reframed, never hidden. The shopper sees both prices anyway; a grid that ignores a 40% premium answers every question except the one being asked. A "cost per use" or "what's included" row that contextualizes the premium is the strongest play; silence on price is the weakest.
Does the comparison image matter for Rufus and the AI shopping surfaces? Not directly — flat-image text is not a reliable input, and I treat it as invisible to retrieval. The grid persuades the human after the AI surfaces deliver them. But the claims in the grid should also live as structured attributes and bullet text, where the assistants can read them. If your differentiator only exists inside a JPEG, it doesn't exist for half your future traffic.
If your comparison image came off a template — or your carousel doesn't have one and your SQP data says shoppers are toggling away — that's exactly the kind of stack problem I work on. The comparison slot is one image, but it's the one that fights your competitors hand-to-hand on your own listing. Make it earn the space.