Amazon Seller & Brand Intel

TL;DR for brand owners, brand-protection teams, distributor vetting, and PL researchers: the seller-intelligence layer of Amazon - who is actually behind a listing, who is selling your brand and at what share, who holds the Buy Box, and how fast a product sells - from public, logged-out pages. Jungle Scout and Helium 10 sell this on subscription; Amazon's own APIs are gated (PA-API needs an approved affiliate account with recent sales, SP-API is seller-only). This is deliberately NOT another review/price scraper - that layer is commoditized. One normalized dataset: seller records with published business identity, offer-by-offer Buy Box maps, brand audits, and BSR-based sales bands with stated confidence.

Run it in 30 seconds

# Via the Apify Python SDK
from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("seibs.co/amazon-seller-intel").call(run_input={
    "mode": "brand_audit",
    "brands": ["Anker"],
    "marketplace": "com",
    "max_products_per_brand": 16
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

Or via curl:

curl -X POST "https://api.apify.com/v2/acts/seibs.co~amazon-seller-intel/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"mode": "asin_offers", "asins": ["B07ZPKBL9V"], "marketplace": "com"}'

Or click "Try for free" on this page if you prefer the no-code UI.

What you get

Each run produces:

• A clean dataset, filterable in the Apify console and downloadable as CSV or JSON
• An OUTPUT.html dashboard preview of your top records
• A sample-output preview at ./.actor/sample-output.json

Live example output

A brand_audit_record - one aggregate per brand:

{
  "record_type": "brand_audit_record",
  "brand": "AcmeBasics",
  "n_products_sampled": 16,
  "distinct_seller_count": 5,
  "distinct_sellers": [
    {"seller_id": "A2R2RITDJNW1Q6", "seller_name": "AcmeBasics Official", "n_products": 11, "est_share_pct": 47.8},
    {"seller_id": null, "seller_name": "Amazon.com", "n_products": 6, "est_share_pct": 26.1}
  ],
  "price_dispersion": {"min": 12.49, "median": 22.99, "max": 39.99, "n": 16},
  "buy_box_brand_or_amz_pct": 68.8,
  "buy_box_third_party_pct": 31.2,
  "total_est_monthly_sales_band": {"low": 4120, "mid": 7480, "high": 12710, "period": "monthly"}
}

A sales_estimate_enrichment - one per ASIN in asin_offers:

{
  "record_type": "sales_estimate_enrichment",
  "asin": "B0ACME1234",
  "bsr": 1450,
  "matched_category": "Home & Kitchen",
  "sales_band": {"low": 578, "mid": 1050, "high": 1785, "period": "monthly"},
  "est_monthly_revenue_band": {"low": 15600.22, "mid": 28339.5, "high": 48177.15, "period": "monthly"},
  "confidence": "medium",
  "methodology": "Modeled as a per-category power law est_daily_sales = A / BSR^B, monthlyized over 30.4 days. ..."
}

What does Amazon Seller & Brand Intel do?

Four things the commodity scrapers skip:

1. Seller profiles. Feed seller ids (or sp?seller= / storefront URLs) and get the seller behind the listing: display name, the published business name and address Amazon shows on the profile, rating percentages across 30d/90d/12m/lifetime windows, ratings count, and a Just-Launched flag. This is the distributor-vetting record.
2. Offer / Buy Box intel. Feed ASINs and get every offer from the All Offers Display: seller name and id, price plus shipping, condition, FBA/FBM/Amazon fulfillment, and which offer holds the Buy Box right now.
3. Brand audits. Feed a brand name and get the seller map: every distinct seller found on a sample of the brand's products, each seller's share, price dispersion across listings (MAP-violation triage), what percent of Buy Boxes the brand or Amazon controls vs third parties, and the brand's estimated monthly sales band.
4. Sales-velocity estimates. Every ASIN with a readable Best Sellers Rank gets a modeled monthly sales and revenue band - the Jungle Scout / Helium 10 product - with confidence, methodology, and caveats stated on the record.

Responsible use / data scope

Public, logged-out pages only. No accounts, no cookies, no login flows, no paywall bypass. The actor includes the published business firmographics Amazon itself displays on a seller profile (legal business name and address) because that is business data; it never harvests emails, phone numbers, or any personal contact data. Outputs describe public marketplace state at fetch time. You are responsible for using the data lawfully and in compliance with the terms that bind you.

AI / RAG / Agent

Seller and brand intel shaped for agent pipelines: every record is flat, typed, and carries record_type, so an agent can route brand audits to a brand-protection workflow and seller records to a vetting workflow without parsing prose. Works with LangChain, LlamaIndex, and MCP-aware runtimes.

from apify_client import ApifyClient

client = ApifyClient("APIFY_TOKEN")
run = client.actor("seibs.co/amazon-seller-intel").call(run_input={
    "mode": "brand_audit",
    "brands": ["YourBrand"],
})
items = client.dataset(run["defaultDatasetId"]).list_items().items
audits = [r for r in items if r.get("record_type") == "brand_audit_record"]
unauthorized = [
    s for a in audits for s in a.get("distinct_sellers", [])
    if s.get("seller_name") not in ("YourBrand Official", "Amazon.com")
]

Features

• Seller records with published business identity and rating windows (30d / 90d / 12m / lifetime).
• All-Offers-Display parsing with Buy Box winner, fulfillment channel (FBA/FBM/AMZ), and shipping cost per offer.
• Brand-to-seller mapping with estimated share per seller and price dispersion.
• BSR-based monthly sales and revenue bands, calibrated per category, with confidence and caveats on every record.
• Block-resilient: Amazon bot walls become labeled fetch_error records with a fix suggestion; the run always finishes.
• Monitor mode: put a seller watchlist on an Apify Schedule and get a change digest (rating and catalog deltas) plus optional Slack webhook.
• Cost-control budget guard: runs abort cleanly if compute outruns revenue; a run that returns nothing costs nothing.

Use cases

• Brand protection - detect unauthorized sellers, watch Buy Box control, triage MAP violations by price dispersion.
• Distributor and supplier vetting - is this seller real, what business is behind it, how do buyers rate it across time windows.
• Private-label research - size a product or brand opportunity with sales bands instead of buying a Jungle Scout seat.
• Agencies - audit a client's (or competitor's) brand presence on a monthly cadence with monitor mode.
• M&A / FBA-aggregator diligence - portfolio-level seller and velocity snapshots.

Modes

Mode	Input	Emits
seller_profile (default)	seller_ids	one seller_record per seller
asin_offers	asins	product_summary + product_offer_record per offer + sales_estimate_enrichment per ASIN
brand_audit	brands	sampled products + offers + one brand_audit_record per brand
seller_monitor	seller_ids (watchlist)	seller records + monitor digest (for Schedules)

Sales estimates - how, and how honest

Amazon never publishes unit sales. The estimator models them from the Best Sellers Rank using the same per-category power-law shape the paid tools describe: est_daily_sales = A / BSR^B, calibrated per top-level category against published anchor points, then converted to a monthly band. Three honesty rules:

1. Bands are wide on purpose (about -45% / +70% around the mid). Treat the band, not the midpoint, as the answer.
2. Confidence is capped at medium even on a clean category match - never high from rank alone. Unmatched categories fall back to a generic curve at low.
3. Every estimate carries its methodology sentence, the curve constants used, and caveats. See ./SPEC.md for the calibration table.

Input

See the input schema for full reference. Example - brand audit:

{
  "mode": "brand_audit",
  "brands": ["Anker"],
  "marketplace": "com",
  "max_products_per_brand": 16,
  "max_offers_per_asin": 10,
  "apify_proxy_groups": ["RESIDENTIAL"],
  "concurrency": 2
}

Hard caps: 25 seller ids, 100 ASINs, 10 brands, 48 products per brand, 20 offers per ASIN, concurrency 3. Marketplaces: com, co.uk, de, ca.

Output

Every record carries record_type, mode, available, reason, and scraped_at. Blocked or missing pages come back as record_type: "fetch_error" with a human-readable reason and are never charged. See ./.actor/sample-output.json for one realistic record of each type.

Pricing

Pay per event - you pay for what the run actually emits:

Event	Price	When
seller_record	$0.012	One parsed seller profile
product_offer_record	$0.005	One offer from the All Offers Display
brand_audit_record	$0.015	One aggregate brand roll-up
sales_estimate_enrichment	$0.010	One BSR-based sales/revenue band per ASIN
scheduled_delta_run	$0.050	One scheduled monitor-mode digest

A run that returns nothing costs nothing. fetch_error records are free.

FAQ

Q: Why does it default to RESIDENTIAL proxies? A: Amazon is among the most bot-protected sites on the web and blocks datacenter IPs aggressively. RESIDENTIAL keeps the success rate high; the trade-off is that residential bandwidth is the actor's main cost driver. The client requests HTML only, compresses transfers, caches URLs per run, and keeps concurrency low (default 2) to spend as little of it as possible.

Q: How accurate are the sales estimates? A: Order-of-magnitude. They are modeled from BSR with per-category calibrated curves, the same approach the subscription tools describe, and every record states its confidence and caveats. Use the band for sizing decisions; do not read the midpoint as a measurement.

Q: Why no reviews or review sentiment? A: Review scraping is commoditized on the Apify Store at around $0.10 per thousand. This actor is the seller-intelligence layer instead: who sells, who controls the Buy Box, and how fast it moves.

Q: What does "Buy Box winner" mean here? A: The featured offer Amazon was showing at fetch time. The Buy Box rotates, so treat it as a snapshot; run on a schedule to see control over time.

Q: What happens when Amazon blocks a page? A: The client rotates its session and retries once; if still blocked you get a fetch_error record naming the page and suggesting apify_proxy_groups=["RESIDENTIAL"] with concurrency 1-2. The run continues with the remaining inputs and finishes SUCCEEDED.

Q: Which marketplaces are supported? A: amazon.com, amazon.co.uk, amazon.de, and amazon.ca in v0.1. URLs and currency are built per marketplace.

Save your input as an Apify Task

Click "Save as task" on the actor page, name it (e.g. brand-audit-yourbrand-monthly), and re-run the exact configuration in one click - or trigger it via API with the task id.

Run this weekly with Apify Schedules

Schedules + seller_monitor is the intended production setup: console.apify.com/schedules -> New schedule -> pick your saved Task -> cron 0 6 * * 1 (Mondays 06:00). Each scheduled run emits the change digest described below.

Monitor mode (beta)

When the actor runs under a saved Task (Schedules included), it compares the current records against the previous scheduled run and emits a monitor_digest record: new / changed / removed sellers and rating deltas, plus a markdown digest. Set monitor_webhook_url to also receive the digest in Slack (or any Slack-compatible webhook). One scheduled_delta_run event ($0.050) is charged per digest.

Related Actors

• shopify-store-discovery - find and qualify Shopify merchants (many also sell on Amazon).
• app-store-revenue-estimator - the same modeled-estimate discipline, pointed at iOS / Google Play apps.
• b2b-sales-triggers - convert seller and brand events into outbound sales triggers.

Support

Issues tab on this actor, or the contact form on my Apify profile. Bug reports with a run URL get fixed fastest.

Changelog

See ./CHANGELOG.md.

Found this useful?

If this actor saved you time or money, please consider leaving a quick review on the Apify Store - it is the single biggest signal for which features get built next. Leave a review: https://apify.com/seibs.co/amazon-seller-intel#reviews