Amazon US · read-only · Public beta

A technical audit of an Amazon listing, with the evidence attached

Paste an ASIN. It reads what Amazon publishes about the listing — attributes, the product type’s own rules, the variation family, images, price and fees — and reports what is wrong with it. Every finding carries the place in Amazon’s response it came from.

Accounts are created by an administrator. Ask yours, or write to us for access.

Listing Checkup is in public beta and is used daily on Launch Hub’s own listings. General availability is planned for Q1 2028.

Nothing about the listing is changed
Every finding cites Amazon's own response
Undecided is never counted as a pass
No product page is ever read

Catalog
Product type schema
Variation family
Images
Offers & fees
Review themes
Your submission

Public beta — free by invitation

Access is by invitation and costs nothing during the public beta. Analyses are metered in points so the cost of an Amazon call is visible from the start — today those points are granted, not sold. A paid subscription tier of about $99 per month is planned for general availability in Q1 2028. Anybody already using the product will be told before pricing changes.

Questions we get

Is Listing Checkup in beta?

Yes. Listing Checkup is in public beta. Every check — catalog attributes, product-type schema, variation family, images, offers and fees, review themes — works today, and every finding cites the place in Amazon’s response it came from. Access is free by invitation during the beta. General availability with a paid tier is planned for Q1 2028.

What it promises

Three things it will not do to you

Nothing is changed

It reads Amazon's API and writes nothing back. No listing, no price and no inventory is touched by an analysis — the worst it can do is spend a few API calls.

Every claim is checkable

A finding names the exact place in Amazon's response it came from, down to the field. You can disagree with the conclusion and still see the fact it rests on.

No data is not a pass

A check that could not be decided says so and is never counted as passing. Where Amazon publishes nothing, the report says Amazon publishes nothing.

The checks

What it looks at

Required attributes, against the product type's own schema from Amazon
Values outside the vocabulary Amazon's filters match on
Barcodes: missing, duplicated, or failing their own check digit
The variation family: themes that match no axis, gaps in the grid
Dimensions and weights that contradict each other
Title length and style against the policy that is actually enforced
A letter from another alphabet hiding inside a Latin word
Images: how many, and whether they are large enough for the zoom viewer
Price against the list price, and what Amazon takes in fees
Up to eight comparable listings, with the median of each figure
The themes Amazon extracts from reviews, against how often the category raises them
Amazon's own complaints about a listing you sell
Whether your connected account may list the ASIN, and on what terms
The same listing over time, from the analyses already run on it

And what one of them produces

One finding

A letter from another alphabet is hiding in one word

High

A7 v1.0.0 · B0H8QT5W4J

Bullet 3 — “Сomplete”: U+0421 С where C belongs
Fields read for disguised letters: 11

catalog /data/attributes/bullet_point/2/value

A real finding from a real run. The last line is the field in Amazon’s own response — follow it and check us.

Authorization

What this application accesses on Amazon

Authorizing is optional and only affects one part of a report. Everything else is read from Amazon's public catalogue and needs no connection at all.

Read, never write

The application holds read scopes only. It submits no feeds, changes no attribute, sets no price and adjusts no inventory. There is no code path in it that writes to Amazon, authorized or not.

Your own listings, when you connect one

With a connection it reads the attributes you submitted and the issues Amazon reports about your listings — the two things nobody else can see and the only reason to authorize.

Public catalogue data, always

Product attributes, images, dimensions, identifiers, the product type’s schema, offers and Amazon’s own fee estimate. All of it is what Amazon publishes about a listing to any developer.

Nothing about buyers

No orders, no customers, no addresses, no personally identifiable information of any kind. The application neither requests those roles nor has anywhere to put them.

Your data

What happens to what it reads

Read from Amazon
Public catalogue, plus your own submission when connected
Kept with its report
So a report opened in six months shows the evidence it rested on
Gone with it
Delete the analysis and the responses go too — no copy elsewhere

Stored with the report it produced

An analysis keeps the Amazon responses it was built from, so that a report opened in six months shows the evidence it actually rested on rather than today’s data. They belong to your organization and are visible to nobody outside it.

Deleted when you delete the analysis

Removing an analysis removes it from your history; ask us and the stored responses go with it. There is no copy anywhere else, and nothing is sold, shared or used to train anything.

The authorization itself

Your refresh token is encrypted at rest with AES-256-GCM under a key held outside the database, and is read by exactly one function, which exchanges it for a short-lived access token.

Withdrawing access

Disconnect here, and remove the application under Apps and Services in Seller Central — that second step is the one that revokes the grant on Amazon’s side, and only you can take it.

The limits

What it is not

It does not read product pages, guess at search volume, estimate how much a fix is worth per day without a sales figure you supply, or promise what Amazon’s ranking will do. Where a number would have to be invented to fill a column, the column says so instead.