I have reviewed 50,000+ Amazon listings and optimized 14,000+ hero images, and I can usually tell within thirty seconds whether an image stack was built from a real creative brief or from a Slack message that said "we need new images, something premium."
The tell isn't polish. Un-briefed stacks are often more polished — the designer had nothing to work from except their own taste, so taste is what you got. Eight beautiful images that answer none of the questions the shopper actually came in with.
Here's the thing nobody says out loud: most bad Amazon image stacks are not a design failure. They're a briefing failure. The designer executed exactly what they were asked to do. The problem is that nobody did the merchandising work before the design work started, so the most expensive decisions — what each image says, in what order, to answer which objection — got made by default, by the person furthest from the customer.
The fix is a document. The Amazon image stack creative brief is the single highest-leverage artifact in the whole creative process, and almost nobody writes one. This post is the exact structure I use — what goes in it, section by section, plus the shot list format, the specs that prevent rework, and the mistakes that quietly ruin briefs that look thorough.
Why "make it look premium" is the most expensive sentence in ecommerce
When a brand skips the brief, the designer has to invent answers to merchandising questions on the fly:
- Which objection does slot 3 exist to kill?
- What does the shopper need to believe before the price feels fair?
- What's the one claim a competitor can't copy?
- What has to be legible at 280 pixels on a phone?
Designers are not equipped to answer these — not because they aren't smart, but because the answers live in your data: your reviews, your return reasons, your search term reports, your customer questions. A designer who has never read your 1-star reviews cannot design the image that prevents them.
So the un-briefed designer defaults to the universal template: hero, benefits-icon row, lifestyle, dimensions, another lifestyle, maybe a badge collage. It looks professional. It merchandises nothing. And the revision cycle that follows — "can we make it pop more," round after round — is the brand paying hourly to do the strategy work they skipped, in the worst possible medium: pixel feedback.
I've watched brands burn $3,000-6,000 and six weeks on revision loops that a two-page brief would have prevented. The brief costs an afternoon. Write the brief.
What a real image stack brief contains (the 6 sections)
My briefs run two to four pages. Longer than that and the designer skims; shorter and you're back to vibes. Six sections, in this order:
1. The buyer, in one paragraph
Not a persona deck. One paragraph: who actually buys this (from your Brand Analytics demographics, not your aspiration), what problem state they're in when they search, and the phrase they'd use to describe the product to a friend. Pull that phrase from your reviews verbatim — buyer vocabulary belongs in image copy, and the brief is where it enters the pipeline.
One line matters more than the rest: what the buyer is afraid of. Every product has a dominant purchase fear — it's too small, it's flimsy, it won't fit my model, it'll taste bad, it'll look cheap in person. Name it. The whole stack is organized around killing it.
2. The objection map
This is the core of the document and the section that separates a brief from a mood board. List the top 5-7 objections in the order the shopper hits them, each sourced from evidence:
- Reviews (yours and competitors'): 3-star reviews are the gold seam — they bought, they were disappointed, and they tell you exactly which expectation the images set wrong.
- Return reasons: every clustered return reason is an image that failed or is missing.
- Customer questions on the listing: every question asked twice is an objection your stack doesn't answer.
- Competitor review mining: what buyers of the alternative complain about is your differentiation slide, written for you.
Format each objection as a row: the objection, the evidence for it, the proof that kills it. "Worried it's too small — 14 of the last 50 reviews mention size — show it in a hand next to a coffee cup" is a design instruction a designer can execute. "Communicate quality" is not.
3. Slot assignments: one job per image
Now assign each objection to a slot. This is where sequencing strategy becomes a build order:
- Slot 1 (hero): identification and click. The hero's job is CTR, full stop — instant category recognition, contrast against the white grid, the single most differentiating attribute visible at thumbnail. The brief should state what the hero must communicate in under one second, and what it must not try to do (everything else).
- Slot 2: the #1 objection, always. This is the most scrolled-to image and the one shoppers flick back to. If your category's dominant question is size, slot 2 is scale-in-context. If it's trust, slot 2 is the proof shot.
- Slots 3-5: objections 2-4, one each. Mechanism, material, compatibility, whatever your map says — in descending order of how often the evidence shows shoppers care.
- Slot 6: in-use or lifestyle with the actual buyer demographic — the shopper needs to project themselves in, and a model twenty years younger than your Brand Analytics data breaks that.
- Slot 7: the close — the strongest remaining proof (review-derived claim, guarantee, what's-in-the-box for kit products).
The rule that makes this section work: if you can't name the exact question an image answers, the image doesn't go in the stack. Seven slots is a maximum, not a quota. A brief that assigns real jobs to six images beats one that pads to nine.
4. The copy layer, written by you, not the designer
Every word that appears on an image goes in the brief, final, before design starts. Three reasons:
First, image copy is merchandising, and merchandising is your job. The designer will write "PREMIUM QUALITY MATERIALS" because that's what template images say. You will write "survives the dishwasher — top rack or bottom" because you read the reviews.
Second, the copy layer is now machine-read. Rufus and the AI shopping layer OCR the text baked into your images, so on-image claims feed retrieval, not just persuasion. That text should carry your buyer-vocabulary phrases and your differentiating attributes — a decision made in the brief, not in Photoshop.
Third, legibility requirements have to travel with the copy: minimum effective font weight, high contrast against background, no text in the outer 5% of frame, everything readable at 280px wide. Put the constraint next to the words it governs.
5. Reference and anti-reference
Give the designer two short lists. Three listings (or specific images) that do something right, each annotated with what specifically to take from it — "steal the scale device, not the palette." Then three anti-references: things you explicitly do not want. Mine almost always includes: no four-icon benefit rows, no floating product on gradient backgrounds, no lifestyle stock photos with models who never touch the product, no collage heroes.
The anti-reference list is the cheapest revision-prevention tool that exists. Every item on it is a revision round you just skipped.
6. Specs and success criteria
The boring section that saves the project:
- Deliverables: count, dimensions (3000px longest side working files; know that Amazon requires 1600px+ for zoom), formats, layered source files included.
- Compliance: hero on pure white (RGB 255/255/255), product filling ~85% of frame, no text/logos/badges on the main image, no depictions that violate category rules.
- Mobile-first review standard: every image is approved at phone size before it's approved at all. Write this into the brief so "it looks great on my monitor" is out of bounds by contract.
- Success criteria: what metric this project moves and how you'll read it — CTR from the search results (expect signal in 2-3 weeks), CVR on the detail page (30-60 days to stabilize). Naming the metric in the brief changes how everyone works, because it reframes the project from "make new images" to "move this number."
The shot list: translating the brief for a photographer
If real photography is involved (it should be — at minimum for the hero base, material macros, and true-demographic lifestyle), the brief compresses into a shot list. Mine is a table, one row per shot:
Shot # | Slot it feeds | Angle/framing | Distance | Props/context | The question this shot answers
Two rules from the field:
Vary camera distance deliberately. You need at least one far shot (context and scale), several working-distance shots (mechanism, use), and true macros (material, texture, build quality). A photographer left alone will shoot everything at the same flattering middle distance, and you'll have gorgeous images that teach the buyer nothing they couldn't guess.
Shoot 3x what the stack needs. Extra angles, extra prop variations, hands in/hands out, both orientations. Marginal cost of an extra frame on set: near zero. Cost of a reshoot because the infographic needed a left-facing angle you didn't capture: two weeks and a setup fee. The overage also becomes your testing inventory — variant heroes for A/B tests, seasonal swaps, ad creative.
Send the photographer the actual brief too, not just the list. A photographer who knows the size objection is the whole battle will find you three scale devices you didn't think of.
Where AI fits in the briefing process (and where it doesn't)
My team uses AI at the front of this process, never the end. Three places it earns its keep:
Review mining at scale. Feeding 500 reviews (yours plus two competitors') through Claude with a prompt asking for clustered objections, verbatim buyer phrases, and expectation gaps turns four hours of reading into twenty minutes of verification. The objection map practically writes its first draft.
Concept mockups inside the brief. A rough AI-generated composition taped into a slot assignment communicates intent faster than three paragraphs. The designer knows it's direction, not art. It kills the most common briefing failure — the designer and the strategist holding different pictures in their heads for two weeks.
Copy stress-testing. Asking a model "you're a shopper worried about X — does this on-image claim resolve it?" catches lazy copy before it ships.
What AI doesn't do in my shop: final production assets. The brief can be AI-assisted; the images that carry your brand at $80K/month cannot be AI-final. Generated finals still fail on product accuracy, and one review that says "the product looks nothing like the pictures" costs more than the photographer did.
The five briefing mistakes I see most
Briefing execution instead of outcome. "Blue background, product on the left, bold font" hires the designer as a pair of hands and wastes the one thing you paid for — their craft. Brief the job ("this image must make a 55-year-old believe it survives a drop"), constrain the guardrails, and let them solve it.
No evidence, all opinion. If the objection map came from a brainstorm instead of reviews and returns, you've written down your team's guesses in a formal font. The brief is only as good as the mining behind it.
The committee brief. Five stakeholders each adding a must-have produces the collage stack — every image trying to say three things, no image saying one thing well. One person owns the brief. Everyone else comments once.
Briefing the stack without the hero strategy. The hero is a different discipline with a different metric (CTR vs CVR) and a different reader (the search grid vs the detail page). Give it its own section with its own success criterion, or it becomes "slot 1" and gets designed like an infographic.
Writing the brief and then not enforcing it. The brief is the acceptance test. Round-one review isn't "do we like it" — it's "does slot 2 kill the size objection, yes or no." The moment feedback drifts to taste ("can we try a warmer tone"), you've abandoned the document and re-entered the vibes economy. Taste feedback is allowed exactly once, at the end, after every slot passes its job check.
FAQ
How long should an image stack creative brief take to write? An afternoon, if the inputs exist — 2-3 hours of review/return/search-term mining, an hour of writing. If it takes fifteen minutes, you skipped the mining and wrote a wishlist. If it takes a week, you're writing a strategy deck, not a brief; cut it to the six sections.
Do I need a brief if I'm hiring a "full service" Amazon creative agency? Especially then. A good agency will build the objection map with you — that's what you're paying for, ask to see it. A bad one will quote you a per-image price without asking for review access, which tells you the strategy layer doesn't exist. The brief is also your quality-control document: if you can't check delivered images against written slot jobs, you can only check them against your mood.
What if my designer pushes back on the brief as too restrictive? Good designers love real briefs — constraints with evidence behind them beat "make it pop" every time. Pushback usually means the designer is used to being the de facto strategist. Clarify the split: you own what each image says and why, they own how it looks. If they want to challenge a slot job, they're welcome to — with evidence, in writing, before design starts.
How often should the brief be redone? Re-mine the inputs quarterly; rewrite the brief when the evidence shifts — new clustered return reason, a competitor repositioning, a demographic drift in Brand Analytics, or a variant taking over sales. The brief for a listing is a living document tied to the review stream, not a one-time launch artifact.
Does this apply to A+ content too? The structure ports directly — A+ gets its own objection assignments (education, comparison, brand trust) so the stack and A+ divide labor instead of repeating each other. Duplicated messaging between carousel and A+ is one of the most common audit findings in my work: two surfaces, one set of questions answered, the other set answered nowhere.
The gap between a stack that looks professional and a stack that sells is almost never the designer. It's whether anyone did the merchandising thinking before the design started — and wrote it down where the designer could execute it. Two to four pages. One afternoon. It's the highest-ROI document in your creative process, and it's the one most brands have never written.