Amazon Image Brief & Conversion Associate

Amazon images are where most listings win or lose. The main image decides whether anyone clicks. The six behind it decide whether they buy — and whether the parcel comes back.

We are hiring one person to own that. Take a live listing, work out from its reviews and returns why it is not converting, and write the image brief that fixes it — a brief a remote designer can execute without a single follow-up question.

This is an operational and analytical role, not a design role. You will not be making the artwork. You decide what it has to say, in what order, and you hold the returned files to that standard.

The split is 75 / 25.

• 75% — the briefs. The conversion thinking, the ranked USPs, the seven slots, the QA of what comes back.

• 25% — analysis and reporting. Part of it is the work that feeds the brief: pulling the reviews, reading them properly, turning them into a persona and a pain-point list. The rest is the follow-up reporting we owe clients — sales and traffic, return reasons, and what actually moved once new images went live.

You will work across markets. Mostly European and US listings, across many markets and languages, in categories you have never shopped in yourself.

What you will do

Image briefs and conversion thinking — 75% of the role

• Diagnose why a listing is not converting: what the gallery fails to answer, which objection nobody addressed, where the buyer loses confidence and leaves.

• Know what each slot is for. The main image earns the click. Images two to seven answer objections in the order a buyer actually raises them. The size and fit slots decide whether the sale stays sold.

• Rank seven USPs by what wins the click and what prevents a return. They are not the same list, and the order is the decision.

• Write the brief for each slot in our template: one message per slot, one headline, an optional subline, a maximum of three to four callouts, every line adding something new.

• Apply the two-second rule. One idea per image, legible at thumbnail size on a phone. If a slot needs a paragraph to work, it is not a slot.

• Work inside the client’s brand book — colours, fonts, existing assets — naming exactly which asset to reuse and where the file lives.

• Keep sizing accurate and internally consistent: backend size chart against image size chart, US against EU conventions.

• Know Amazon image policy cold: main-image rules, no price, ratings, review counts or badges, and how the gallery renders on a phone.

• QA the returned files against the brief and say precisely why something is not working — hierarchy, contrast, whitespace, text density, whether a close-up box earns its place.

The analysis behind the briefs — part of the 25%

• Download the reviews yourself and read them properly: themes, frequency, and the exact words customers use.

• Pull Voice of the Customer and the return report, and weight return reasons against units actually sold.

• Turn that into a buyer persona and a ranked pain-point list — for categories you did not grow up in.

• Separate “we can fix this on the listing” from “the product is the problem”.

• Evidence only. Never invent a return reason, a measurement or a number, and ask when something is missing.

Reporting that follows up sales — the rest of the 25%

• Find and download the right reports without being walked through it: Sales & Traffic, Search Query Performance, Returns, Voice of the Customer.

• Build the recurring follow-ups: sales and traffic trend, return rate by size, conversion before and after an image update.

• Merge two exports on ASIN or SKU and produce a clean pivot, without hand-counting rows.

• Deliver a one-page summary that says what matters, why it matters and what to do about it. Not a data dump.

Making the workflow better with AI — across all of the above

• Use Claude and ChatGPT as part of the work, not as a novelty: review analysis, first-draft copy, translation, summarising exports.

• Work inside our Claude projects, with the context and templates already there, and keep them current as the accounts change.

• Build repeatable workflows — project instructions, skills, prompt templates, SOPs — so the next product of a kind is faster than the last.

• Know where AI is unreliable and check it. Invented numbers, invented quotes and plausible-sounding measurements are the failure mode, and they are your responsibility, not the tool’s.

• Have an opinion about the tooling itself. We write our briefs in PowerPoint today and we are open to something faster — if you know a better way to brief a designer, we want to hear it and we will let you build it.