Amazon Seller Data Extractor — Ratings, Contacts and Products

Amazon Seller Data Extractor — Ratings, Contacts and Products

Amazon Seller Data Extractor Plus pulls structured seller records — storefront name, seller ID, ratings, feedback history, business address, contact phone/email, fulfillment type, and (optionally) the seller's own product catalog — from Amazon seller pages, product pages, category URLs, and keyword search, and pushes them to your dataset as typed JSON. It resolves the seller behind any asinInputs, usernameInputs, startUrls, keywordInputs, or categoryUrls you paste in, then optionally paginates each seller's storefront (includeProducts) and feedback history (maxFeedbackPages) for deeper coverage. Sourcing teams, marketplace analysts, and dropshipping/wholesale buyers use it to build supplier shortlists, vet fulfillment method before reaching out, and track seller-level rating trends over time. The sections below cover every input, every output field, and how to narrow a run down to exactly the sellers you want.

What is Amazon Seller Data Extractor Plus?

Amazon Seller Data Extractor Plus is an Apify Actor that scrapes public Amazon seller-profile pages (/sp?seller=…) and the product pages that link to them, and returns one structured JSON record per seller — plus, if you switch it on, one additional record per product in that seller's storefront. It runs entirely over HTTP (no browser), authenticating each request with a rotating Apify residential-proxy session rather than an Amazon account, and needs no Amazon login or API key of any kind. Results stream to the dataset in real time as each section finishes, so a run with dozens of inputs starts producing rows immediately rather than waiting for the whole batch.

Key capabilities, all present in the current source:

• Five ways to point at a seller — a direct asinInputs product code, a usernameInputs seller ID, a startUrls seller-profile link, or discovery via keywordInputs search terms or categoryUrls category/search-result pages.
• Optional child entity: product listings. With includeProducts on, the Actor paginates the seller's own storefront (/s?me=<sellerId>) and emits each product as its own row (type: "product", isChild: true) linked back to the seller via parentSellerId.
• Feedback beyond page 1. maxFeedbackPages walks the seller's feedback pagination, and sortOrder controls whether the merged feedback list comes back newest-first, in Amazon's on-page order, or highest-rated first.
• Structured contact and address fields. businessName, a parsed businessAddress (line1/city/state/postalCode/country), contactPhone, and contactEmail are extracted where the seller has published them.
• Quality and fulfillment filters applied before billing. minRatingPositive, minLifetimeRatings, and fulfillmentType drop sellers that don't clear your bar — those rows are never pushed and never charged.
• Locale and marketplace control. amazonDomain targets any Amazon storefront (amazon.com, amazon.co.uk, amazon.de, amazon.co.jp, …) and language sets the Accept-Language header Amazon replies to.

What data can I extract with Amazon Seller Data Extractor Plus?

Every run returns two related record types — seller (the parent record) and, when includeProducts is enabled, product (a child of that seller) — both pushed to the same dataset and both billed under the same charged event.

Field	Example value	Use case
type	"seller" or "product"	Split seller vs. product rows in downstream processing
isChild	false / true	Filter parent seller rows from child product rows
section_index	3	Trace a row back to its position in the input list
section	"B00P8XQPY4" or "keyword:wireless earbuds"	Trace a row back to the input value that produced it
sellerId	"A33W53J5GVPZ8K"	Primary key for joining seller and product rows
sellerName	"Peckomatic"	Human-readable storefront name
url	"https://www.amazon.com/sp?seller=A33W53J5GVPZ8K"	The exact page scraped
canonicalSellerUrl	"https://www.amazon.com/sp?seller=A33W53J5GVPZ8K"	Stable link back to the seller profile
description	"Peckomatic is committed to providing…"	Storefront "About" copy
businessName	"Francis Kunnumpurath"	Legal/registered business name behind the storefront
businessAddress	{ "line1": "2612 State Route 80", "city": "Lafayette", "state": "NY", "postalCode": "13084", "country": "US" }	Structured mailing address for outreach or compliance checks
contactPhone	"+1 555-0134"	Direct phone line, when the seller has published one
contactEmail	"support@example.com"	Direct email, when the seller has published one
stars / ratingStarsNumeric	"4.5 out of 5 stars" / 4.5	Raw and numeric overall star rating
rating_positive / ratingPositivePercent	"88% positive" / 88	Raw and numeric positive-feedback percentage
rating_count_lifetime	42	Total lifetime ratings — feeds minLifetimeRatings filtering
ratingCountM1 / ratingCountM3 / ratingCountM12	2 / 7 / 30	Rating volume over the last 30 / 90 / 365 days, for trend detection
feedbacks	[{ "stars": "5 out of 5 stars", "starsNumeric": 5.0, "text": "Bird arrived in great condition.", "date": "By Kao Y. on November 16, 2021." }]	Individual buyer feedback rows, ordered per sortOrder
feedbackCount	1	Number of feedback rows attached to this seller record
feedbacks_percentages	{ "star_5": "79%", "star_4": "10%", "star_1": "5%" }	Star-rating distribution as published on the profile
fulfillment	"FBA" or "FBM"	Detected fulfillment method — feeds fulfillmentType filtering
productCount	30	Number of product rows collected for this seller (when includeProducts is on)
products_link	"https://www.amazon.com/s?ie=UTF8&me=A33W53J5GVPZ8K"	Direct link to the seller's storefront search page
shipping_policies	"Unless noted otherwise…ships all items within two days…"	Seller-published shipping terms
privacy_policy / privacy_security	"Peckomatic values the privacy of your personal data."	Seller-published privacy copy
country	"US"	Marketplace country, derived from amazonDomain
scrapedAt	"2026-07-01T10:11:16Z"	ISO-8601 timestamp of extraction
parentSellerId / parentSellerName (product rows)	"A33W53J5GVPZ8K" / "Peckomatic"	Joins a product row back to its parent seller
asin (product rows)	"B00P8XQPY4"	Amazon product identifier
price / priceRaw / currency (product rows)	29.99 / "$29.99" / "USD"	Parsed numeric price plus the original on-page string
rating / reviewCount (product rows)	4.3 / 1024	Product-level rating and review volume

seller_id and seller_name also ship as legacy lower-snake-case aliases of sellerId/sellerName for backward compatibility with earlier versions of this Actor.

Ratings and feedback data

The rating block is the decision-driving data on every seller record: ratingStarsNumeric and ratingPositivePercent give you two directly comparable scores, rating_count_lifetime plus the ratingCountM1/ratingCountM3/ratingCountM12 trio let you see whether a seller's feedback volume is accelerating or has gone quiet, and feedbacks carries the actual buyer text, per-review star rating, and posting date — pulled from as many pages as maxFeedbackPages allows and capped per seller by maxComments. A wholesale buyer screening 200 candidate sellers might sort on ratingPositivePercent and rating_count_lifetime first, then spot-check feedbacks text on the top 20 before reaching out.

Targeting and fulfillment data

fulfillment (FBA vs. FBM, detected from the product offer when a run starts from asinInputs), country (derived from amazonDomain), and productCount (populated only when includeProducts is on) are the fields you segment on. Combined with the three quality filters — minRatingPositive, minLifetimeRatings, fulfillmentType — they let a run return only, say, FBA sellers on amazon.co.uk with at least 100 lifetime ratings and 90%+ positive feedback, without any post-processing.

Why not build this yourself?

Amazon does not publish a general-purpose API for looking up arbitrary third-party sellers — the Selling Partner API is scoped to a seller's own account and requires Amazon's developer approval, so it cannot answer "who sells this ASIN" or "what's this competitor's rating history" for a business you don't control. Building that lookup by hand means solving three problems this Actor's source code already handles: seller-ID extraction has to fall back across several DOM selectors and a regex scan because Amazon doesn't expose one stable "seller ID" element on every page layout; blocked or CAPTCHA'd responses have to be detected (the Actor checks page length and a list of known block phrases such as "robot check" and "enter the characters you see below") and retried with a fresh residential-proxy session and rotated browser headers up to several times before giving up; and a cold request without a prior homepage visit is itself a bot signal, so the Actor warms up each session against the Amazon homepage before requesting the target page. None of that is exposed as a documented, rate-limited public endpoint anywhere — it's scraping-specific plumbing, which is exactly what this Actor packages up.

How to use data extracted from Amazon?

Sourcing and procurement teams

Paste candidate usernameInputs or startUrls for sellers you're evaluating, set minRatingPositive and minLifetimeRatings to your minimum bar, and turn on includeProducts with a maxProductsPerSeller cap to pull each seller's current catalog alongside their rating history. The output lands as one row per qualifying seller plus one row per product, ready to filter on fulfillment, businessAddress, and contactEmail/contactPhone before a first outreach email goes out.

Agencies and marketplace consultants

Running the same seller list on a schedule — weekly or monthly — turns ratingCountM1/ratingCountM3/ratingCountM12 into a trend line per client seller: a sudden drop in ratingCountM1 relative to the ratingCountM12 average flags a seller worth a closer look before a quarterly review call. keywordInputs runs against a client's core search terms also surface newly-appearing competitor sellers in that category between reports.

Market research and intelligence

categoryUrls or keywordInputs runs resolve the sellers behind a whole category or search result page, and fulfillmentType splits the resulting set into FBA vs. FBM to gauge how a niche is fulfilled. Pairing ratingPositivePercent distribution across that seller set with country (from amazonDomain) supports cross-marketplace comparisons — e.g., is FBA share higher on amazon.de than amazon.com for a given keyword.

AI agents and automated pipelines

Because output is typed JSON keyed by stable fields (sellerId, ratingPositivePercent, fulfillment, businessAddress), an agent can use this Actor as a tool call — "look up the seller behind ASIN B0XXXX and tell me if they're FBA with 90%+ positive feedback" — and reason directly over the returned record without a parsing step, or feed a batch of sellerId/ratingPositivePercent pairs into a lead-scoring pipeline.

🔼 Input sample

All 18 input parameters are optional — the schema sets required: [] — but a run needs at least one non-empty source array (asinInputs, usernameInputs, keywordInputs, categoryUrls, or startUrls) or it logs an error and exits without pushing anything.

Parameter	Required	Type	Description	Example value
asinInputs	No	array	One Amazon ASIN per row; the Actor opens the product page and resolves the seller behind the offer.	["B00P8XQPY4"]
usernameInputs	No	array	Amazon seller IDs (e.g. A33W53J5GVPZ8K) when you already know exactly who to look up.	["A33W53J5GVPZ8K"]
keywordInputs	No	array	Search phrases; the Actor runs each search, scans up to the first 8 products, and resolves the seller behind each one.	["wireless earbuds"]
categoryUrls	No	array	Full Amazon search or category result URLs, expanded into products and then sellers.	["https://www.amazon.com/s?k=coffee+grinder"]
startUrls	No	array	Direct Amazon seller-profile URLs (/sp?seller=…).	["https://www.amazon.com/sp?seller=A33W53J5GVPZ8K"]
includeProducts	No	boolean	Default false. Scrape each seller's storefront and emit each product as its own child row.	true
maxProductsPerSeller	No	integer	Default 20, range 0–500. Caps product rows collected per seller when includeProducts is on; 0 collects everything the storefront paginates.	30
maxFeedbackPages	No	integer	Default 1, range 1–50. How many seller-feedback pages to paginate; 1 reads only the first page.	3
minRatingPositive	No	integer	Default 0, range 0–100. Drops sellers whose positive-feedback percentage is below this; 0 disables the filter.	90
minLifetimeRatings	No	integer	Default 0, range 0–1,000,000. Drops sellers with fewer lifetime ratings than this; 0 disables the filter.	100
fulfillmentType	No	string (any/FBA/FBM)	Default "any". Keeps only sellers with the given fulfillment method — only known for sellers resolved from asinInputs.	"FBA"
amazonDomain	No	string	Default "amazon.com". The Amazon storefront domain to scrape.	"amazon.co.uk"
language	No	string	Default "" (Amazon's default). Optional locale sent as the Accept-Language header.	"en-GB"
sortOrder	No	string (recent/helpful/rating)	Default "recent". How the merged feedback list is ordered in each seller row.	"rating"
maxComments	No	integer	Default 10, range 0–1000. Caps feedback rows kept per seller; 0 keeps everything collected.	25
proxyCountry	No	string	Default "AUTO" (derives the country from amazonDomain). ISO-style country hint for proxy routing.	"GB"
maxConcurrency	No	integer	Default 10, range 1–20. Maximum sections processed in parallel; lower values are gentler on fragile storefronts.	5
proxyConfiguration	No	object	Apify Proxy configuration. This Actor always runs on RESIDENTIAL proxies regardless of what's set here.	{ "useApifyProxy": false }

{
  "asinInputs": ["B00P8XQPY4"],
  "usernameInputs": [],
  "keywordInputs": ["wireless earbuds"],
  "includeProducts": true,
  "maxProductsPerSeller": 30,
  "maxFeedbackPages": 2,
  "minRatingPositive": 85,
  "minLifetimeRatings": 50,
  "fulfillmentType": "any",
  "amazonDomain": "amazon.com",
  "sortOrder": "recent",
  "maxComments": 15,
  "proxyCountry": "AUTO",
  "maxConcurrency": 10
}

Common pitfall: asinInputs applies a heuristic — any value that starts with "A" and is longer than 10 characters is treated as a seller ID, not a product ASIN, and routed straight to a seller-profile fetch instead of a product-page fetch. Real Amazon ASINs are exactly 10 alphanumeric characters, so this only misfires if you paste a seller ID into asinInputs by mistake; put seller IDs in usernameInputs instead to avoid ambiguity.

🔽 Output sample

Each row is typed, normalized JSON with a consistent schema per record type — there is no CSV- or Excel-specific formatting step in the source, but Apify's dataset export supports downloading any run's results as CSV, Excel, JSON, XML, or RSS regardless of how the Actor wrote them.

Seller row (type: "seller"):

{
  "section_index": 1,
  "section": "B00P8XQPY4",
  "type": "seller",
  "isChild": false,
  "url": "https://www.amazon.com/sp?seller=A33W53J5GVPZ8K",
  "canonicalSellerUrl": "https://www.amazon.com/sp?seller=A33W53J5GVPZ8K",
  "sellerId": "A33W53J5GVPZ8K",
  "sellerName": "Peckomatic",
  "description": "Peckomatic is committed to providing each customer with the highest standard of customer service.",
  "businessName": "Francis Kunnumpurath",
  "businessAddress": {
    "raw": "2612 State Route 80, Lafayette, NY, 13084, US",
    "line1": "2612 State Route 80",
    "city": "Lafayette",
    "state": "NY",
    "postalCode": "13084",
    "country": "US"
  },
  "contactPhone": null,
  "contactEmail": null,
  "feedbacks": [
    { "stars": "5 out of 5 stars", "starsNumeric": 5.0, "text": "Bird arrived in great condition.", "date": "By Kao Y. on November 16, 2021." }
  ],
  "feedbackCount": 1,
  "stars": "4.5 out of 5 stars",
  "ratingStarsNumeric": 4.5,
  "rating_positive": "88% positive",
  "ratingPositivePercent": 88,
  "rating_count_lifetime": 42,
  "ratingCountM1": 2,
  "ratingCountM3": 7,
  "ratingCountM12": 30,
  "feedbacks_percentages": { "star_5": "79%", "star_4": "10%", "star_3": "7%", "star_2": "0%", "star_1": "5%" },
  "productCount": 30,
  "fulfillment": "FBM",
  "products_link": "https://www.amazon.com/s?ie=UTF8&me=A33W53J5GVPZ8K",
  "shipping_policies": "Unless noted otherwise in the ordering pipeline, Peckomatic ships all items within two days of receiving an order.",
  "country": "US",
  "scrapedAt": "2026-07-01T10:11:16Z"
}

Product row (type: "product", child of the seller):

{
  "section_index": 1,
  "section": "B00P8XQPY4",
  "type": "product",
  "isChild": true,
  "parentSellerId": "A33W53J5GVPZ8K",
  "parentSellerName": "Peckomatic",
  "asin": "B00P8XQPY4",
  "title": "Peckomatic Bird Feeder",
  "price": 29.99,
  "priceRaw": "$29.99",
  "currency": "USD",
  "rating": 4.3,
  "reviewCount": 1024,
  "url": "https://www.amazon.com/dp/B00P8XQPY4",
  "scrapedAt": "2026-07-01T10:11:18Z"
}

How do you filter and target specific sellers?

This Actor exposes two independent ways to find sellers, and three thresholds to keep only the ones worth paying for.

Direct lookup vs. discovery. When you already know the seller — an ASIN, a seller ID, or a seller profile URL — feed asinInputs, usernameInputs, or startUrls, which cost one fetch (plus one product-page fetch for ASINs) per seller. When you don't, keywordInputs or categoryUrls run a search or category page first, scan its first products, and resolve each one's seller — more requests per result, but the only way to prospect a category you haven't already scoped down to known sellers.

Quality thresholds. minRatingPositive and minLifetimeRatings filter on the seller's feedback record; fulfillmentType filters on FBA vs. FBM. All three are applied to the fully-scraped seller record before it's pushed — a seller that fails any active filter is dropped silently (logged, not pushed, not charged), so a narrow filter set reduces both dataset size and cost, not just dataset size.

Volume controls. maxProductsPerSeller and maxFeedbackPages control how deep each individual seller gets scraped, not how many sellers you can target — the source arrays (asinInputs, usernameInputs, keywordInputs, categoryUrls, startUrls) are plain lists with no schema-enforced maximum length, so run size scales with how many rows you paste in.

Three real examples:

Cast a wide FBA-only net across one Amazon domain, no volume limits on feedback or products:

{ "keywordInputs": ["ceramic mugs"], "fulfillmentType": "FBA", "minRatingPositive": 0 }

Vet a known shortlist of sellers with a strict quality bar before outreach:

{ "usernameInputs": ["A33W53J5GVPZ8K", "A2N5Q13T20FFT2"], "minRatingPositive": 90, "minLifetimeRatings": 200 }

Pull one seller's full current catalog plus three pages of feedback history for a deep-dive audit:

{ "startUrls": ["https://www.amazon.com/sp?seller=A33W53J5GVPZ8K"], "includeProducts": true, "maxProductsPerSeller": 0, "maxFeedbackPages": 3 }

▶️ Want to try other Amazon scrapers?

Scraper name	What it extracts
Amazon Best Sellers Scraper (amazon-bestsellers-scraper)	Amazon Best Sellers, New Releases, and Most Wished For rankings, with BSR rank, ASIN, price, and rating

Comparable contact/directory scrapers	What it extracts
LinkedIn Profile Contact Lookup (linkedin-profile-contact-lookup)	Contact details resolved from LinkedIn profile URLs, public IDs, or name + company
Instagram Profile Contact Enricher (instagram-profile-contact-enricher)	Contact details enriched from Instagram profile usernames, URLs, or IDs
Facebook Groups Search Scraper Plus (facebook-groups-search-scraper-plus)	Facebook groups resolved from keyword or group-link search

How to extract Amazon data programmatically

This Actor runs as a standard Apify Actor: one authenticated call to the Apify API starts the run, and results come back as structured JSON from the run's dataset — no separate REST endpoint to design or maintain.

Python example

from apify_client import ApifyClient

client = ApifyClient("<APIFY_API_TOKEN>")

run = client.actor("SimpleAPI/amazon-seller-data-extractor-plus").call(run_input={
    "usernameInputs": ["A33W53J5GVPZ8K"],
    "minRatingPositive": 85,
    "fulfillmentType": "FBA",
})

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["sellerName"], item["ratingPositivePercent"], item["fulfillment"])

MCP for AI agents

Apify exposes every Actor, including this one, through its Actors MCP Server at mcp.apify.com, so MCP-compatible clients — Claude, Cursor, and other agent frameworks that speak the Model Context Protocol — can call it as a tool without custom integration code:

$npx -y @apify/actors-mcp-server --actors SimpleAPI/amazon-seller-data-extractor-plus

A sourcing agent can then call this Actor mid-conversation — "find FBA sellers for ceramic mugs with at least 90% positive feedback" — and get back structured sellerId/ratingPositivePercent/fulfillment rows to reason over directly.

Export to spreadsheets or CRM

Every Apify dataset can be downloaded as CSV or Excel directly from the run's Storage tab or via the API's dataset export endpoint. Map sellerName → company/contact name, contactEmail/contactPhone → your CRM's contact fields, and businessAddress.line1/city/state/postalCode/country → address columns for a direct import into a spreadsheet or CRM outreach list.

Is it legal to scrape Amazon?

Scraping publicly accessible Amazon seller pages is generally legal — these pages are published by Amazon for any visitor to view without logging in, which is what this Actor's HTTP-only requests do. Storefront descriptions, ratings, and feedback counts are business data with no personal-data component. Business address, contactPhone, and contactEmail, however, can identify an individual when the seller is a sole trader rather than a registered company, which brings GDPR (for EU/UK data subjects) or CCPA (for California residents) into play for how you store and use that data. Always review Amazon's Conditions of Use for your account and consult legal counsel for commercial applications involving bulk storage of personal data.

❓ FAQ

How does the Actor know if a seller listing is still active?

The Actor scrapes each seller page live at request time; it does not track or flag delisted or suspended sellers between runs. A seller profile that returns a blocked/empty response (Amazon's "robot check" or CAPTCHA pages, or a response under roughly 2,000 characters) is treated the same as a genuinely missing seller and simply produces no row — re-run the same input later to distinguish a temporary block from a real removal.

Can I get the seller's product catalog along with the seller record?

Yes — set includeProducts to true and the Actor scrapes the seller's storefront (/s?me=<sellerId>), paginating up to maxProductsPerSeller product rows (or every page the storefront exposes if maxProductsPerSeller is 0). Each product is pushed as its own type: "product" row linked to the seller via parentSellerId, and product rows are also mirrored to a separate per-run dataset named products-<runId>.

How accurate is the contact and rating data?

The Actor returns the seller's rating, feedback, and contact fields exactly as published on their Amazon profile at request time — accuracy depends on how current the seller keeps that profile. Ratings and lifetime counts reflect Amazon's own displayed figures; contact fields (contactPhone, contactEmail) are only populated when the seller has published them publicly. Verify any contact detail before using it for outreach at scale.

How many records can I get per run?

There is no schema-enforced cap on how many sellers a single run can process — asinInputs, usernameInputs, keywordInputs, categoryUrls, and startUrls are all open-ended arrays. What is capped per seller is depth: maxProductsPerSeller (0–500, default 20) for product rows, maxComments (0–1000, default 10) for feedback rows, and maxFeedbackPages (1–50, default 1) for how many feedback pages get paginated. maxConcurrency (1–20, default 10) controls how many sellers are processed in parallel, not the run's total ceiling.

How does the fulfillment filter work?

fulfillmentType (any/FBA/FBM) keeps only sellers whose detected fulfillment method matches. Fulfillment is detected from the buy-box/merchant-info block on the product page, which means it is only known for sellers resolved via asinInputs — sellers resolved from usernameInputs, startUrls, keywordInputs, or categoryUrls won't have a reliable fulfillment value unless that same seller was also reached through an ASIN in the same run.

Does Amazon Seller Data Extractor Plus work with Claude, ChatGPT, and AI agent frameworks?

Yes — it's reachable through Apify's Actors MCP Server (mcp.apify.com) for MCP-native clients like Claude, and callable as a standard Apify API run by any agent framework that can make an authenticated HTTP request, including custom ChatGPT tool integrations.

How does this compare to other Amazon seller scrapers?

As observed on the Apify Store on 2026-07-26, both pratikdani/amazon-seller-extractor and api-empire/amazon-seller-data-extractor return single seller records (name, rating, feedback, contact fields) without a linked product-catalog child entity; this Actor adds that as an optional includeProducts toggle, plus feedback pagination (maxFeedbackPages) and pre-billing quality filters (minRatingPositive, minLifetimeRatings, fulfillmentType) that neither competitor's listing describes. Field-level completeness on any individual seller still depends on what that seller has chosen to publish on their Amazon profile, regardless of which scraper is used.

Can I use this Actor without an Amazon seller account or API key?

Yes. The Actor needs only your Apify account to run — it authenticates to Amazon's public pages over rotating residential proxy sessions, not an Amazon login, and requires no Amazon Selling Partner API credentials or developer approval.

Conclusion

Amazon Seller Data Extractor Plus turns a list of ASINs, seller IDs, seller URLs, keywords, or category pages into structured seller intelligence — ratings, feedback trends, business contact details, fulfillment type, and an optional linked product catalog — without an Amazon account, an API key, or hand-built anti-bot handling. It's built for sourcing teams vetting suppliers, agencies tracking client sellers over time, and researchers mapping a category's seller base, and every field it returns traces directly to what Amazon publishes on the seller's own profile. Start a run from the Apify Console with a handful of usernameInputs or asinInputs to see the seller and product schema on real data before scaling up.