Amazon Seller Email Scraper — VAT, Phone & Address Leads

Turn an Amazon search, a product URL or a seller ID into a B2B lead record for the third-party business behind the listing: legal company name, e-mail, phone, VAT number, trade-register number, business address and customer-service address — plus the seller's full ratings history and recent buyer feedback.

The contact data is not scraped out of hiding. Every EU marketplace seller is legally required to publish it under the Digital Services Act / EU trader-transparency rules, and Amazon renders it on the public seller-profile page as the Detailed Seller Information panel. This actor reads that panel and hands you the whole record as one clean row.

One row per unique seller, deduplicated on seller ID — not one row per product. A seller found on ten listings is billed once.

---

What it does

🔎 Search → sellers	Type a keyword. Every product on the results page is opened, the third-party merchant behind it is resolved, and that merchant's public profile is scraped.
🔗 Product URL → seller	Paste product URLs or bare ASINs and get the sellers behind exactly those listings.
🎯 Seller ID → record	Already have seller IDs or storefront URLs? Skip discovery — one request per lead.
🏛️ Full DSA imprint	Legal name, business type, trade register number + court, VAT ID, phone, fax, e-mail, business address, and the separate customer-service address and phone.
⭐ Four rating windows	1 month / 3 months / 12 months / lifetime, each with its own score and count, plus the 5-star percentage histogram and recent feedback comments with seller replies.
🌍 11 marketplaces	9 EU sites (.de .it .es .fr .nl .se .pl .com.be .ie) plus .co.uk and .com.tr. Proxy country is pinned to the marketplace automatically.
🚫 No login	Public pages only. No account, no target-site cookies, no CAPTCHA-solving services.

---

Quick start

{
  "marketplace": "amazon.it",
  "searchTerms": ["ricambi trattore"],
  "maxSellers": 30
}

Other ways in:

// specific products
{ "marketplace": "amazon.de", "productUrls": ["https://www.amazon.de/dp/B0CX1JN6WF", "B0D4QVBR6R"] }

// sellers you already know — one request per lead, no discovery cost
{ "marketplace": "amazon.it",
  "sellerUrls": ["A1YYMKC9INWJTB", "https://www.amazon.it/s?me=AI18FTVVOEZ7J"] }

Keyword choice is the single biggest lever on yield. Third-party density is a property of the search term, not the site. Spare parts, accessories, industrial and niche supplies are dominated by marketplace sellers. A big-brand or Amazon-own-label keyword returns first-party listings, which carry no seller imprint at all — the run will honestly report "N products were sold by Amazon itself, not third parties" rather than pretending it failed.

---

Output

61 fields per row. Example (real row, trimmed):

{
  "sellerId": "A1YYMKC9INWJTB",
  "marketplaceDomain": "amazon.it",
  "marketplaceId": "APJ6JRA9NG5V4",
  "storeName": "BSSW Europe GmbH",
  "sellerLogoUrl": "https://m.media-amazon.com/images/I/01AbK-Y2KQL.jpg",
  "sellerProfileUrl": "https://www.amazon.it/sp?ie=UTF8&seller=A1YYMKC9INWJTB",
  "storefrontUrl": "https://www.amazon.it/s?me=A1YYMKC9INWJTB",
  "contactSellerUrl": "https://amazon.it/gp/help/contact-seller/contact-seller.html?sellerID=…",
  "leaveFeedbackUrl": "https://www.amazon.it/gp/feedback/leave-consolidated-feedback.html?…",

  "isBusinessSeller": true,
  "businessName": "BSSW Europe GmbH",
  "businessType": "Impresa privata",
  "tradeRegisterNumber": "HRB 23899",
  "tradeRegisterCourt": "Aachen",
  "vatNumber": "IT00281799999",
  "phone": "+4924054069355",
  "fax": "+49 (0) 2405 / 40 693-56",
  "email": "amazon@bssw-europe.com",
  "businessAddress": "Adenauerstraße, 20 A2, Würselen, 52146, DE",
  "businessAddressLines": ["Adenauerstraße", "20 A2", "Würselen", "52146", "DE"],
  "customerServiceAddress": "Adenauerstr. 20A2, 3.OG, valonic Geb A2, 3.OG, Würselen, NRW, 52146, DE",
  "customerServicePhone": "+4924054069355",
  "addressCity": "Würselen", "addressRegion": null,
  "addressPostalCode": "52146", "addressCountryCode": "DE",
  "aboutBlurb": "BSSW Europe GmbH\n\nAdenauerstr. 20 (A2, 3.OG)\n\n52146 Würselen\n\nTelefono: +49 (0) 2405 / 40 693-55\n\nFax: …",
  "odrPlatformLinks": null,
  "imprintRaw": { "Venditore Business": true, "Nome azienda": "BSSW Europe GmbH", "…": "…" },

  "ratingOverall": 5,
  "positiveFeedbackPct12mo": 100,
  "rating1Month": 4.8,  "ratingCount1Month": 14,
  "rating3Months": 4.9, "ratingCount3Months": 56,
  "rating12Months": 4.9,"ratingCount12Months": 188,
  "ratingLifetime": 4.9,"ratingCountLifetime": 1100,
  "starDistribution5Pct": 86, "starDistribution4Pct": 14,
  "starDistribution3Pct": 0,  "starDistribution2Pct": 0, "starDistribution1Pct": 0,
  "recentFeedbackReviews": [
    { "rating": 4, "text": "ok", "author": "Cliente Amazon",
      "date": "2 settembre 2026", "sellerReply": null }
  ],

  "sellerResolverStrategy": "anchor",
  "isAmazonFulfilled": true,
  "sourceAsin": "B0CX1JN6WF",
  "sourceProductTitle": "fouraces fanali posteriori per rimorchio - 2 set, con schema elettrico …",
  "sourceProductPrice": 16.99,
  "sourceProductCurrency": "€",
  "sourceProductUrl": "https://www.amazon.it/dp/B0CX1JN6WF",
  "catalogueSize": 78,
  "discoveredFromKeyword": "ricambi trattore",
  "scrapedAt": "2026-09-03T03:04:17.998Z"
}

Field table + MEASURED fill rate

Fill measured on a real Apify run (run LQIpWcySsXWQQuXf8, 2026-09-03, amazon.it, three keywords, 29 seller rows, 125 s at 1 GB). These are the actual numbers, not estimates.

Field	What it is	Fill
sellerId	Amazon merchant ID (the dedupe key)	100%
marketplaceDomain / marketplaceId	Site scraped + Amazon's internal marketplace ID	100%
storeName	Public store name	100%
sellerProfileUrl / storefrontUrl / contactSellerUrl	Ready-to-open links	100%
leaveFeedbackUrl	Feedback form link	97%
sellerLogoUrl	Store logo image	38%
isBusinessSeller	Business seller flag from the imprint panel	100% (all true)
email	Imprint e-mail	100%
phone	Imprint phone	100%
vatNumber	VAT / USt-IdNr / Partita IVA / Btw-nummer	100%
businessName	Legal company name	90%
businessType	e.g. Impresa privata / Unternehmen in Privatbesitz	90%
tradeRegisterNumber	Company / trade register number	90%
businessAddress + businessAddressLines	Registered address (joined + as lines)	90%
addressCity / addressPostalCode / addressCountryCode	Parsed address parts	90%
addressRegion	Region/state, when the panel carries one	72%
customerServicePhone	Support phone — distinct from the imprint phone	45%
customerServiceAddress + …Lines	Support address — distinct from the business address	34%
tradeRegisterCourt	Register court (e.g. Aachen)	7%
fax	Fax number	3%
odrPlatformLinks	EU Online Dispute Resolution URLs	7%
aboutBlurb	Free-text seller description / German-style Impressum text	100%
imprintRaw	Every label→value pair verbatim, including labels this actor does not map	100%
imprintFound / imprintStrategy	Whether the DSA panel was found, and how	100%
ratingOverall, positiveFeedbackPct12mo, ratingCountHeader	Headline rating block	97%
rating12Months / ratingCount12Months	12-month window	97%
ratingLifetime / ratingCountLifetime	Lifetime window	97%
ratingCount1Month / ratingCount3Months	Short-window counts	97%
rating1Month / rating3Months	Short-window scores (null when that window has 0 ratings — not 0)	66% / 86%
starDistribution5Pct … 1Pct	5-star histogram	97%
recentFeedbackReviews[]	{rating, text, author, date, sellerReply}	97%
sellerResolverStrategy	Which resolver found the merchant (anchor / a-state / legacy-sp / input)	100%
sourceAsin, sourceProductTitle, sourceProductPrice, sourceProductCurrency, sourceProductUrl	The listing the seller was found on	100%
isAmazonFulfilled	FBA flag from the merchant link (null when the link omits it)	45% true
catalogueSize	Products in the seller's storefront (opt-in, extra request)	90%
discoveredFromKeyword	Keyword that surfaced the seller	100%
scrapedAt	ISO timestamp	100%
otherSellersCount / otherSellersMinPrice	Competing offers on the source listing	0% — see honest limits

Four dataset views ship with the actor: Seller leads, Full imprint record, Ratings and Where the seller was found.

---

Input

Field	Type	Default	Notes
marketplace	enum	amazon.it	.it .de .es .fr .nl .se .pl .com.be .ie .co.uk .com.tr
searchTerms	string[]	[]	Keywords to search
productUrls	string[]	[]	Product URLs or bare ASINs
sellerUrls	string[]	[]	Seller profile / storefront URLs, or bare seller IDs
maxSellers	int	25	Unique sellers to return. 0 = no limit
maxPagesPerTerm	int	1	Search pages per keyword (1 page ≈ 60 products)
includeFeedbackReviews	bool	true	Recent buyer comments — free, same page
maxFeedbackReviews	int	5	Cap on the comments array
includeCatalogueSize	bool	false	Adds one storefront fetch per seller (~+45% requests)
requestConcurrency	int	4	Higher raises the anti-bot rate, not the speed
proxyConfiguration	proxy	Apify Residential, country auto-pinned	Bring your own residential proxy here if you prefer

Run it with no input at all and it executes a documented 5-seller demo on amazon.it rather than failing — handy for a first click, and it is what the platform's daily test runs.

---

Pricing

Pay per event — $3.00 per 1,000 sellers ($0.003 each). No start fee, no per-result surcharge, no usage add-on. One line item on your bill.

Event	Rate
seller-scraped — per unique seller returned	$0.003

Rows are delivered through Apify's budget-aware push, so delivered == billed: if you set a charge cap, the actor stops at the cap instead of handing you rows you already paid past.

---

Notes / honest limits

Measured, not assumed. Where a number is quoted it came from a run you can re-run.

Residential proxy is mandatory — this is the real cost of the actor. From a plain home or datacenter IP, https://www.amazon.it/s?k=… returns HTTP 200 with a ~2,232-byte Akamai bm-verify stub, not a page. Judge Amazon on bytes, never on status: a 200 can be a block. The default configuration uses Apify Residential with the proxy country pinned to the marketplace, and you can substitute your own residential proxy in proxyConfiguration.

HTTP/2 had to be turned off, and the difference is large. Measured 2026-09-03, same proxy, same headers, same minute, 8 attempts each: with HTTP/2 on, amazon.se 0/8, amazon.pl 0/8, amazon.co.uk 1/8 usable; with HTTP/2 off, 6/8, 4/8 and 5/8. The actor ships with it off.

Marketplace availability is not uniform. In the same session, amazon.it, .de, .es and .nl served cleanly on the first or second attempt. amazon.fr is the outlier: it is behind two different walls — Akamai's bm-verify interstitial (which this actor solves inline: on .fr its arithmetic proof-of-work was solved and the full 1.7 MB page recovered, and on the measured amazon.it run the same solver fired and succeeded 4 times out of 4) and an AWS WAF challenge.js page (HTTP 202, 2,007 bytes, server: CloudFront) which cannot be answered over plain HTTP and is simply retried on a fresh exit. Across 6 attempts on .fr only 1 got through. Expect thin or empty results on amazon.fr; the other marketplaces are fine.

Cost per lead, measured. The 29-row run above spent 165 requests — about 5.7 requests per delivered lead with includeCatalogueSize on and including every retry the residential pool needed (48 transient proxy errors, all recovered). Turn includeCatalogueSize off and pass seller IDs directly and it drops to roughly one request per lead.

otherSellersCount / otherSellersMinPrice filled 0 of 29. Amazon's "other sellers" panel is now rendered entirely client-side, and all three of its AJAX endpoints return HTTP 404 with a 2,233-byte body (/gp/aod/ajax/ref=auto_load_aod, /gp/product/ajax/ref=aod_f_new, /gp/product/ajax?…experienceId=aodAjaxMain, verified 2026-09-03). The fields are still emitted — they populate on the marketplaces and page variants that still render the text form — but do not plan around them today.

Address parsing is a best-effort split of an unlabelled block. Amazon prints the address as ordered lines with no field names. Country code and postcode are unambiguous and are popped from the end; city and region are inferred from the remaining lines. The raw lines always ship in businessAddressLines / customerServiceAddressLines, so you can re-split them yourself.

Not every seller publishes everything. E-mail, VAT and phone filled 29/29 here, but legal name and address filled 26/29 and the register court only 2/29. Small sellers publish a minimum record. Fields are emitted as null rather than guessed.

The seller-profile page occasionally comes back as an empty shell — HTTP 200, ~350 KB, but with no profile body at all. That is a partially-served page, not a seller with no data, so the actor detects it and skips the row instead of emitting an all-null lead. One of 30 sellers hit this in the measured run.

Amazon rotates the merchant link. As recently as two days before this was built, the merchant on a product page was found via a /sp?…seller= href; on today's dominant template that href does not exist and the link is <a id="sellerProfileTriggerId" href="/gp/help/seller/at-a-glance.html…">. All three resolvers ship and each row records which one fired in sellerResolverStrategy, so a future change shows up as a strategy shift in your data instead of a silent zero.

This actor does not: scrape a seller's product catalogue (only a catalogueSize count), read anything behind a login, verify e-mail deliverability, or return first-party Amazon listings (they have no seller imprint).

robots.txt. https://www.amazon.it/robots.txt (fetched 2026-09-03) does not disallow any path this actor requests under User-agent: *. The only near-miss rule concerns faceted search URLs this actor never builds:

User-agent: *
Disallow: */s?k=*&rh=n*p_*p_*p_

Checked against a standard robots parser, all four request shapes — /s?k=<keyword>, /dp/<ASIN>, /sp?ie=UTF8&seller=<id> and /s?me=<id> — are allowed.

Legal / personal data — your call, not ours. The imprint panel is published under EU trader- transparency law, so it is public business data by design. But sole traders and small partnerships publish a natural person's name, a home address, a personal mobile and sometimes a personal Gmail address. Using those as B2B leads is a personal-data processing decision under GDPR, and it is yours to make. Amazon's own Conditions of Use also restrict automated access and point commercial users at its paid Product Advertising API; technical access is not the same as permission.