Amazon Product Data — Prices, Ratings & Specs

Scrape Amazon products by keyword or ASIN. Feed in search terms, ASINs, or product URLs and get the full record: price, list price and discount, star rating, review count, seller, Prime status, stock, every hi-res image, the complete specifications table, category breadcrumbs and best-seller rank.

No API key. No third-party Actor in the middle.

---

The detail most Amazon scrapers get wrong

Amazon prices in the currency of whichever IP requests the page. The same laptop stand returns MAD149.35 from a Moroccan IP and $15.99 from a US one — and because cloud scrapers rotate through proxy exit nodes in different countries, the currency can change between rows of the same run. That silently corrupts every price comparison built on the data.

This Actor pins the currency explicitly and returns it on every row as currency, so a price is always interpretable. Set currency to override it.

What you get per product

Group	Fields
Identity	asin, title, brand, url, domain
Pricing	price, list_price, discount_pct, currency, has_coupon
Reception	rating, ratings_count
Availability	in_stock, availability, seller, ships_from, is_prime
Content	features[] (bullet points), images[] (all hi-res), main_image, images_count
Placement	categories[] (breadcrumbs), best_seller_ranks[] (rank + category)
Technical	specifications{} — the full product details table, as key/value
Meta	scraped_at

Search-only mode (fullDetail: false) returns the lighter subset visible on the results page — asin, title, price, rating, ratings_count, image, is_sponsored — which is much faster and cheaper when you only need to shortlist.

Input

Field	Type	Notes
searchTerms	array	Keywords, e.g. ["laptop stand", "standing desk"]
asins	array	Bare ASINs (B077B9W343) or any Amazon product URL; mixes freely with keywords
domain	select	com is the verified marketplace. 16 others are selectable but returned a bot-check page from shared cloud IPs when measured — see limits
currency	string	Pins prices. Defaults to the marketplace's own currency
fullDetail	boolean	One request per product for the complete record; off = search-page fields only
maxItems	integer	Hard cap per run
minRating / maxPrice	string	0 = no filter
onlyPrime / onlyInStock / skipSponsored	boolean	onlyInStock needs fullDetail on
delaySeconds	string	Pause between requests; lower = higher chance of an early stop
proxyConfiguration	object	Required in practice — see below

Example

{
  "searchTerms": ["mechanical keyboard"],
  "domain": "com",
  "currency": "USD",
  "fullDetail": true,
  "minRating": "4.0",
  "skipSponsored": true,
  "maxItems": 100
}

Honest limits

• A proxy is required. Apify blocks direct connections to Amazon from Actors (refused in milliseconds), and Amazon rate-limits repeat IPs. This Actor draws a fresh proxy IP per request; the standard datacenter group is enough — you do not need RESIDENTIAL.
• Only amazon.com is verified. Measured 2026-08-03 from Apify's datacenter proxies: amazon.co.uk and amazon.de both returned a 2 kb bot-check page rather than results, so the other 16 marketplaces are offered but unverified — they may work behind your own residential proxy. The Actor now reports that case as a bot check rather than as "0 products", so you will never get a silent empty run mistaken for an empty catalogue.
• One search page per keyword (~16–48 products). Deeper pagination triggers Amazon's bot check, and this Actor does not attempt to defeat it. To go wider, pass more keywords or narrower ones rather than expecting page 2.
• best_seller_ranks and brand are not on every listing. Amazon omits them for many products; the fields are null/[] rather than missing, so filtering stays safe.
• If Amazon's bot check stops a run, the Actor stops and tells you how many products went unscraped instead of returning a short list that looks complete. You are only charged for products that returned data.
• Prices and availability are marketplace-specific: the same ASIN legitimately differs between amazon.com and amazon.de.

Use cases

• Price monitoring and repricing — track price vs list_price and discount_pct over time.
• Competitor and catalogue research — pull a category's products with ratings and review counts to find gaps.
• Product sourcing — filter by rating and price band to shortlist winners before buying.
• Review-volume analysis — ratings_count and rating together separate proven sellers from new listings.
• Feeding an AI agent — the specifications table and features array give a model clean, structured product context.

Pricing

Pay per result: charged once per product delivered. Products removed by your filters, not found, or blocked are not charged.

FAQ

Do I need an Amazon API key or account? No. This reads public product pages.

Why do I need a proxy? Apify's own network blocks direct egress to Amazon, and Amazon throttles repeated requests from one IP. The datacenter group is sufficient.

Can I get more than the first page of results? Not reliably — Amazon presents a bot check on deeper pages, which this Actor deliberately does not try to bypass. Use more, narrower keywords instead.

Why is brand sometimes empty? Amazon does not render a brand byline on every product. Where it is absent, the specifications table often carries Brand or Manufacturer.