The most common structural failure I find in an Amazon image stack isn't a bad hero. It isn't a weak A+ module either. It's that both surfaces are answering the same four questions, and six real buyer questions are answered nowhere on the page.
I've reviewed north of 50,000 listings. When I put a brand's carousel claims in one column and their A+ Content claims in another, the overlap is routinely 60-70%. Same four benefits. Same icon set, sometimes literally the same icon file. Same "premium materials" line, once in image five and once in module four.
Meanwhile the question that's actually costing them the sale — how it installs, whether it fits a face-frame cabinet, what happens after 200 washes — appears on neither surface.
Nobody decided this. That's what makes it worth writing about. Two vendors did competent work from the same brief, three months apart, and the result is a listing that says one thing loudly and stays silent on everything else.
Two Surfaces, Two Readers, Two Jobs
The starting mistake is treating the carousel and A+ as "the images" and "the other images." They're not the same surface at a lower position. They are consumed by different people at different moments with different amounts of intent.
The carousel is a qualification surface. It works on a stranger. It renders at 280px in search results, it gets a half-second glance, and most shoppers see two or three frames before deciding whether this listing deserves more of their afternoon. Roughly 70% of that traffic is on a phone, flicking. The carousel's job is to survive contact with someone who has no reason to care.
A+ Content is a resolution surface. Nobody reaches it by accident. To see module one, a shopper had to click into your listing, not bounce, scroll past the buy box, past the bullets, past the sponsored carousel, and keep going. That's a self-selected, high-intent reader. They have already decided you're a candidate. They're now looking for a reason to stop looking.
The economic consequence: the carousel decides whether you get considered. A+ decides whether the consideration closes. Those are different jobs, and they need different content.
There's a second layer people miss. Scroll depth into A+ on mobile averages around 62% in the session data I've looked at — meaning roughly a third of the shoppers who reach A+ at all never see modules five, six, or seven. So A+ isn't just lower-traffic than the carousel, it's front-loaded internally too. Anything you place in the back half of A+ is reaching a fraction of a fraction.
The Machine Reads Them Differently Too
This part changed the calculus in the last two years and most creative plans haven't caught up.
The carousel is read by OCR. Text baked into your images is machine-legible, and it feeds the systems deciding whether you belong in a result set. That argues for short, high-contrast, literal phrases on your images — the kind of thing a person scans and a machine parses cleanly.
A+ Content is not indexed by A9. Amazon has been consistent about this: A+ text does not help you rank in Amazon search. If you're keyword-stuffing your A+ modules, you're doing it for nothing, and you're making the copy worse for the human who does read it.
But two other things do read it. Google indexes A+ Content, including alt text, which is why A+ can pull external traffic to your listing. And Rufus reads it as context when deciding whether your product answers a shopper's conversational question. Those systems reward natural language, not comma-salad.
So the copy logic inverts across the two surfaces. Carousel: compressed phrases, high contrast, designed to be parsed at a glance by a person and by OCR. A+: readable prose, real sentences, alt text written like a description rather than a keyword dump.
The Redundancy Audit: Find It in Ten Minutes
Before you plan anything, measure what you've already got. This is the fastest diagnostic in creative work and almost nobody runs it.
- Open your listing on a phone.
- Write down every claim on every carousel image — every headline, callout, icon label, badge. One line each.
- Do the same for every A+ module.
- Put the two lists side by side.
- Count exact and near-exact repeats.
Then do the harder half: pull your top 30 recent reviews, your return reason codes, and your customer questions, and write down every distinct thing a buyer wanted to know. Cross off each one that appears somewhere on your two lists.
What's left uncrossed is your gap. In most audits I run it's four to eight real questions, and at least two of them are directly implicated in the return rate.
The typical finding reads like this: eleven claims across the carousel, nine across A+, twelve of the twenty are duplicates, and the two most common return drivers are addressed on neither.
Why It Happens: Everyone Did Their Job
This isn't incompetence, which is why telling people to "try harder" doesn't fix it.
The carousel gets built first, usually from a brand deck with a "key benefits" slide. Four benefits, because decks have four. The photographer and designer execute against it.
Six weeks or six months later, somebody decides the listing needs A+. A different vendor — often a different agency entirely — gets briefed. They ask for the source material. They're handed the same brand deck with the same four-benefit slide.
Both vendors executed the brief accurately. Nobody was ever given an allocation map, because nobody knew one was a deliverable. The output is a listing that repeats itself, and the brand experiences it as "our creative is done" when what they have is one surface's worth of information spread across two.
The fix is upstream. It's a document, not a design skill.
The Allocation Rule
Here's the sorting rule I use, and it's simple enough to apply in a meeting:
If the answer changes whether they click, it belongs in the carousel. If it changes whether they buy after they've already clicked, it belongs in A+.
Click-stage questions are fast, visual, and disqualifying. They're the ones a shopper resolves with their thumb still moving. Buy-stage questions are slower, more specific, and they belong to someone who's already interested and is now hunting for a reason to be sure.
Run every question from your inventory through that filter. Most sort cleanly. The ones that don't are usually your differentiator, and I'll get to those.
Carousel Territory
"What is this and what does it do?" — the hero's entire job. Function legible at thumbnail.
"How big is it, really?" — scale is the highest-leverage post-hero frame in almost every category, and it's a click-stage question because a shopper who suspects it's smaller than it looks won't click at all. Slot two, with a real reference object or a dimension callout.
"Which version do I need?" — variant, size, count. If a shopper can't tell which one they're looking at, they'll go find a listing where they can.
"Is it built well?" — a material macro. Texture, stitching, weld, weave. This is a click-stage question because perceived quality sets whether your price looks reasonable before anyone reads a word.
"Will it actually show up looking like this?" — the packaging shot. A verification signal, not a brand signal, and it's answering live counterfeit and gray-market anxiety.
"What's the one reason to pick this over the neighbors?" — your single differentiator, stated once, plainly.
A+ Territory
"How does this compare to the three others in my open tabs?" — the comparison chart. This is a buy-stage question by definition: nobody comparison-shops a product they haven't clicked into. It's also consistently the highest-CVR-impact module I see, which is why it should sit early in the A+ stack rather than in slot six where a third of readers never arrive.
"How do I install / assemble / use / care for it?" — process content. Sequential, multi-step, needs room. It cannot compress into a 280px frame and it shouldn't try.
"What about my specific situation?" — the edge cases. Does it work with a face-frame cabinet, a 3/4" post, a 40-inch inseam, a septic system. This is the FAQ module's actual job, and it's the module most often wasted on marketing copy.
"Who makes this and should I trust them?" — Brand Story. A stranger scrolling search results does not care about your founding. A shopper 60% of the way down your page, deciding between you and one competitor, sometimes does.
"What's the full spec?" — the technical detail that would clutter a carousel frame and is exactly what a buy-stage reader wants.
"What are other people's outcomes?" — designed proof, review-language callouts, before-and-afters where the category allows them.
The One Thing You Should Repeat On Purpose
There's exactly one item that should appear on both surfaces: your single differentiator.
Not your four benefits. The one thing that is the reason to buy this instead of the cheaper lookalike. That should land in the carousel as a compressed visual claim and again in A+ as the expanded, evidenced version.
The distinction matters. Repeating your spine is reinforcement. Repeating your feature grid is waste. A shopper who sees the same differentiator stated twice, once fast and once with proof, experiences consistency. A shopper who sees the same four icons twice experiences nothing, because they already processed it and their brain files the second instance as "seen."
Kill the Duplicate Feature Grid
If you take one action item from this piece, make it this one.
The four-icon feature grid is the most duplicated asset in Amazon creative. It appears as carousel image five, and it appears again as A+ module four, frequently with identical icons and identical eight-word labels. Two of your most expensive slots, one message.
Delete one of them. Which one depends on the category:
Keep it in the carousel, cut it from A+ when the buying decision is fast and mostly visual — consumables, accessories, low-consideration home goods, anything under about $30. The shopper may never reach A+, and if they do they're close to converting and need proof, not a recap.
Keep it in A+, cut it from the carousel when the product is technical or high-consideration — electronics, tools, equipment, anything over roughly $150. Those shoppers reliably scroll, and the carousel slot is better spent on scale, build quality, or a use-case frame that earns the click in the first place.
Then fill the freed slot from your uncrossed question list. That's where the actual CVR movement lives — not in improving the duplicate, but in answering something that was previously answered nowhere.
When You Can Only Afford One
Brands ask this constantly, usually when a budget got cut mid-project.
Build the carousel. Every time. It isn't close.
The reasoning is structural, not aesthetic. A+ Content cannot be seen by a shopper who never clicked, and the carousel is what determines whether they click. A perfect A+ page behind a weak hero is a beautifully argued case presented to an empty room. The reverse — a strong carousel with no A+ at all — still transacts. It transacts worse than it should, but it transacts.
There's a sequencing implication too. If you're rebuilding both, don't rebuild them simultaneously. Change the carousel, let CTR read for two to three weeks and CVR settle over 30-60 days, then change A+. Move both at once and you'll have a number that changed and no idea which surface changed it. I've watched brands spend a full quarter re-litigating that internally.
The Six-Step Allocation Audit
Run this per hero SKU. It takes about 45 minutes the first time and 20 after that.
- Build the question inventory. Reviews, return reason codes, customer questions, and your PPC search terms. Every distinct thing a buyer wanted to know. Aim for 12-20 items.
- Rank by cost. Which unanswered questions are producing returns, and which are producing bounces? Return-driver questions get priority — they're costing you twice, once at the sale and once at the refund.
- Sort by the click/buy rule. Every question gets exactly one surface. Write it in a two-column table. Force the assignment; "both" is only allowed for your one differentiator.
- Map to slots. Carousel questions go to specific image numbers, highest-cost question in slot two. A+ questions go to specific module numbers, comparison chart early, never in the back third.
- Find the orphans and the duplicates. Any question with no slot is a gap. Any slot answering an already-answered question is a duplicate to reassign.
- Check the machine layer. Carousel: is the text large and high-contrast enough to OCR cleanly? A+: is the copy in real sentences, and is the alt text descriptive rather than a keyword list?
The output is a one-page allocation map. Hand that to whoever builds either surface next, and the duplication problem doesn't come back.
Frequently Asked Questions
Does A+ Content help my Amazon search ranking? Not directly. A+ text is not indexed by A9, so keywords placed there don't help you rank on Amazon. What it does do is improve conversion, and conversion velocity feeds ranking indirectly. Separately, Google does index A+ Content including alt text, and Rufus reads it as context. So write it for humans and machines that read language — not for A9.
Should my A+ repeat the hero image? No. The shopper scrolled past your hero to get there; showing it again spends a full-width module on something they've already processed. If you want continuity, keep the visual system consistent — same color logic, same type, same photographic treatment — without reusing the frame.
How many carousel images versus how many A+ modules? Fill the carousel — use every slot Amazon gives you, because they're the highest-traffic frames you own. For A+, five to seven modules is the working range, front-loaded. Adding modules eight and nine mostly adds content that a third of your readers won't reach.
Does this change if I have Premium A+? The allocation logic is identical; you just have better tools on the A+ side. Premium's interactive modules — hover hotspots, video, the swappable-image carousel — are genuinely good at edge-case and process questions, which are already A+ territory. Premium doesn't move any question from the carousel column to the A+ column. It executes the A+ column better.
Can I test the allocation itself? Partly. Manage Your Experiments will A/B test A+ Content variants, so you can test two different A+ builds against each other. It won't test "carousel versus A+" as a split, because those aren't competing variants of one surface. The honest approach is sequential: fix the carousel gap first, read it, then fix A+ and read that.
What if the same question genuinely matters at both stages? That's your differentiator, and it's the one exception. State it compressed in the carousel and expanded with evidence in A+. If you think three questions qualify, you haven't picked a differentiator yet — you have a feature list.
The listings that convert well aren't the ones with the prettiest carousel or the most elaborate A+ page. They're the ones where every buyer question has exactly one home and no question has two.
Most brands have already paid for enough creative to cover their whole question inventory. It's just distributed badly — two surfaces shouting the same four things at a shopper who wanted to know six others.
Run the audit. The gap list is usually shorter than you fear, and it's almost always already sitting in your return reason codes waiting to be read.