Marketplace channel · Rufus + COSMO

Rufus doesn't read your listing the way A9 did. Score it the way COSMO does.

On Amazon you don't own the robots.txt or the JSON-LD. You own four fields in Seller Central: Title, Bullets, A+ Content, Backend Keywords. eCommerce Insights scores every ASIN against Rufus citations and the 15 intent relations in Amazon's COSMO knowledge graph, then hands you the edits for exactly those four fields.

Any public ASINNo credentialsWeekly re-score
Why Amazon listings are exposed

Amazon is a closed system. The levers are different.

No schema to patch

"Add Product JSON-LD" isn't an action here.

Amazon emits its own product schema and you can't change it. Generic AI visibility tools whose advice resolves to schema patches have nothing to say on Amazon. The editable surface is listing content — so that's what gets scored and rewritten.

COSMO

Rufus reads relations, not keyword density.

Amazon Science's COSMO paper (SIGMOD 2024) describes a commonsense graph of 15 relation types — capableOf, usedFor, partOf, locatedAt, mannerOf, and others — mapping products to intent. A listing heavy on "madeOf" facts but missing the "usedFor" scenario sits outside the graph that routes Rufus answers, as of mid-2026.

Off-Amazon citations

ChatGPT often cites your Amazon listing, not your site.

When a shopper asks ChatGPT or Perplexity for a product, the cited source is frequently the Amazon listing. eCommerce Insights tracks those external citations too, so you see what your Amazon copy says on your behalf everywhere.


What the scan checks on Amazon

Scored against the surfaces Amazon actually leaves you.

CheckScoreSeller Central surface
COSMO relation coverage — which of the 15 relations the listing expressesCitationTitle · Bullets · A+
Rufus citation record against a category query bankCitationtracked weekly
Title intent coverage in the first 80 charactersCitationProduct Title
Bullet answerability — capableOf + usedFor sentence structureCitationBullet Points
A+ Content modules scored on answerability, module by moduleCitationA+ Content
Search Terms field vs. COSMO intent clusters for the categoryCitationBackend Keywords
Off-Amazon citations — ChatGPT, Perplexity, Google AI Overviews citing the ASINBothexternal scans

Source for COSMO: Amazon Science, SIGMOD 2024. Scoring detail: Rufus Score docs.

From finding to fix

Paste-ready edits for the four fields you own.

There is no push to Amazon — Seller Central doesn't work that way, and eCommerce Insights doesn't pretend it does. Each ASIN gets a diff: the Title reordered to put the missing capableOf relation up front, Bullets rewritten as scenario sentences Rufus can lift directly, A+ modules flagged for rewrite, and the missing COSMO intent phrases for the 250-character Search Terms field. You copy the edits into Seller Central, ship, and the weekly re-score tracks the Rufus citation count from there. CSV export covers bulk workflows and flat-file uploads.

Hybrid DTC-plus-Amazon brand? The Shopify side of the catalog gets the metafield workflow; the ASINs get this one. Same dashboard, channel-aware output for each.

ASIN diff · Bullet 2 · illustrative
— Made with 100% organic cotton, pre-shrunk.
+ Wicks sweat through 45-minute workouts; the 100% organic cotton is pre-shrunk so the fit holds after washing.
COSMO relations added: usedFor · capableOf · hasProperty
Worked example

One ASIN, two scan cycles.

COSMO relations present6 → 11 / 15
Rufus citations per 14-query scan2 → 6
Title score54 → 82
Fields editedTitle · 5 Bullets · Search Terms

Illustrative: a health-and-wellness ASIN whose top bullets were all "madeOf" and "partOf" facts — material, ingredients, package contents — with no usedFor scenario anywhere in the listing. The diff rewrote five bullets as scenario sentences and moved the primary use case into the first 80 characters of the Title. Rufus citations on the category query bank rose over the following two weekly scans.

Numbers are illustrative; no customer data is published without permission.


Frequently asked questions

Do I need to connect Seller Central to use eCommerce Insights on Amazon?

No. eCommerce Insights audits any listing from its public ASIN URL. Paste a storefront or a list of ASINs, get scores against Rufus and the COSMO graph, and copy the recommended edits into Seller Central yourself. A Seller Central API connection is on the roadmap; today the workflow is read-only and out-of-band.

What is COSMO and why does it matter for Rufus visibility?

COSMO is the commonsense knowledge graph Amazon Science published at SIGMOD 2024. It defines 15 relation types — capableOf, usedFor, partOf, locatedAt, mannerOf, and others — that map products to shopper intent. Rufus reads listings through that graph, so a listing that expresses what the product does and the scenario it serves fits the graph; a listing that only lists features sits outside it.

What does eCommerce Insights tell me to change in Seller Central?

Specific edits to the four fields you own: Product Title (which COSMO relations are missing from the first 80 characters), Bullet Points (feature bullets rewritten as capableOf-plus-usedFor sentences Rufus can lift), A+ Content (which modules pair the image with an answerable spec), and Backend Keywords (the intent phrases missing from the Search Terms field). Every edit ships as a diff against your current copy.

Does this optimize for Amazon search or for Rufus?

Both, because they share listing infrastructure. Title and Bullet rewrites that lift Rufus visibility also satisfy A9 indexing, since both systems read the same fields. Where the two diverge — Rufus weights answerability and usedFor scenarios over keyword density — the diff flags the trade-off and the default rewrite favors Rufus without hurting A9.

Can eCommerce Insights track classic Amazon search rank too?

No, deliberately. Helium 10 and Jungle Scout already track search rank well. eCommerce Insights covers the layer above: Rufus queries on Amazon, plus the off-Amazon AI engines — ChatGPT, Perplexity, Google AI Overviews — when they cite your Amazon listing as the source.

Ask AI about eCommerce Insights for Amazon

Have your preferred AI engine summarize this page for your ASINs.

The four fields Amazon leaves you

Title. Bullets. A+. Backend Keywords. Scored.

Audit any public ASIN. No Seller Central credentials. Weekly Rufus tracking on every plan.

15 COSMO relations · 4 editable fields · weekly