Audit an Amazon listing with Helium 10 and AI

The question is what on this page is costing the shopper's click or purchase, and what do the winning competitors do instead? This job judges the page. It does not pick search terms to target or analyse ad spend; those belong to the keyword and ads jobs.

1. Confirm the target before pulling anything

Read-only tools are safe to run, but the audit is only useful against the right product. Confirm with the user:

• the ASIN and the marketplace (the same ASIN can differ by country);
• the Helium 10 account if the user manages several, and that it is the one holding this product.

If the user has not said which marketplace, ask once. Do not guess.

2. Pull the listing and its score

Call get_listing_details for the ASIN, then get_listing_score. Record the current title length, bullet count, whether each bullet leads with a benefit or a feature, description presence and length, backend terms if returned, and images or A+ content if the response includes them. Compare what you read against what the score says. Where they disagree, trust the page content and say so.

Treat the score as a guide to which areas are weak, not as the verdict. A low score with a strong product and good reviews is usually a copy problem; a high score with poor conversion usually points to price, reviews, images or the offer.

3. Compare against the competitors who win

A listing is only weak relative to something. Run compare_listings with the target ASIN and the competitors the user names or the top organic sellers you can identify. If the user gives no competitors, use get_competitor_overview to find the strongest sellers in the same niche and say which ones you picked and why.

Look for the gaps that repeat across winners:

• Shared claims the target lacks (a certification, a material, a size range, a use case that every top listing states in its first bullet).
• Shared structure: winners open with the outcome for the buyer, then specifics, then reassurance.
• Vocabulary: words in winners' titles and bullets that the target never uses. Note them as candidate terms, but do not stuff them in; flag that the keyword job should confirm which ones are worth ranking for.
• Length and density: a title that is far shorter than every winner's may be underselling; one that is crammed with repeated terms reads as spam.

4. Judge each section by what it does for the shopper

Work top to bottom, in the order the shopper sees the page:

1. Title. Does it name the product type in plain words, carry the main differentiator, and read cleanly on a phone? Keyword stuffing, repeated words and ALL-CAPS are traps: they lower conversion and can trigger suppression.
2. Main image and gallery. Visible from the listing details, judge only what the data shows. Do not claim you have seen an image you cannot inspect.
3. Bullets. Each bullet should answer one buyer question. Check that the first two bullets carry the strongest benefit, that claims are specific (numbers, materials, dimensions), and that none promises what the product cannot deliver.
4. Description and A+ content. Is there one? Does it expand on the bullets rather than repeat them? Long unbroken text gets skipped.
5. Backend search terms. If returned, check for duplication of the title and for terms that are not relevant. Do not state what Amazon indexes beyond what the docs or tool output says.

5. Find what is missing, not just what is wrong

The most valuable findings are absent: no use case stated, no care or sizing information, no answer to the obvious objection ("is it dishwasher safe", "will it fit my model"), no comparison with the category norm. Build a short list of the buyer questions a shopper would have, then mark which the listing answers. Each unanswered question is a fix.

6. Prioritise by impact and effort

Rank every fix:

• High impact, low effort: rewrite the title opening, move the strongest benefit into bullet 1, fix a factual error or a contradiction between bullets.
• High impact, higher effort: add a comparison or use-case section to A+ content, produce missing images.
• Lower impact: tighten wording, reorder minor bullets, add a closing reassurance line.

Never recommend a change that makes a claim you cannot verify from the listing data or the user's product facts. Phrase such items as "confirm this, then add it".

Common patterns and traps

• Title stuffed with every keyword: reads badly and converts worse. Keep the title for the product, the differentiator and the main keyword, and move the rest to bullets.
• Bullets that restate the title: wasted space. Each bullet must add a new fact or benefit.
• Feature lists with no benefit: "BPA-free stainless steel" is a feature; say what it means for the buyer in the same line.
• Competitor copied word for word: a compliance and differentiation risk. Recommend the shared claim, expressed in the user's own words.
• Low score, good sales: do not rewrite a listing that converts well just to raise a score. Ask what is actually failing.
• Blaming the copy for a pricing or review problem: check the offer, rating and review count before concluding the page is at fault.
• Changing the wrong ASIN or marketplace: the most common expensive error. Repeat the ASIN and marketplace back in the report.

Output

ASIN / marketplace: <asin> / <marketplace>
Listing score: <value or not returned> — <what it measures, in one line>
Winners compared: <asin list and why chosen>

Top fixes (ranked)
1. <section> — current: "<short quote>" — change to: "<proposed>" — why: <buyer question or gap>
2. ...

Missing buyer questions: <list>
Shared competitor claims not made: <list>
Candidate terms for keyword review (not yet confirmed): <list>
Not established: <what could not be checked, e.g. images, A+ content, reviews>

Rules

• Do not make any change on Amazon or in Helium 10. This skill only reads and drafts. Any later write, such as updating a listing or saving a note, must wait for the user's explicit yes.
• Do not invent product facts (materials, certifications, dimensions, warranty). Only use what the listing data or the user supplies; otherwise write "confirm before adding".
• Do not copy competitor wording into recommendations.
• Do not claim a listing score or rank that the tool did not return.
• Do not take a switch of account, workspace or marketplace as a formality: say which one is active and wait for the user's explicit yes before switching.

Done when

• The ASIN and marketplace are confirmed and repeated in the output.
• Each top fix names a section, the current wording or gap, and the buyer reason for the change.
• At least one competitor comparison is shown, or the report says none could be found.
• The "Not established" line lists what was not checked.