Amazon Product Scraper — Search, Bestsellers & Product Detail in one actor

Scrape Amazon search results, bestseller rankings and full product detail pages from a single actor, into one normalized schema. Built for price monitoring, catalogue building, competitive research, and AI agents.

Most Amazon actors give you one surface and their own field names. This one returns identical records whether a product came from a keyword search, a bestsellers grid, or its own detail page — so your pipeline handles exactly one shape.

Why this actor

• 100% price completeness, and we measure it. Every run reports priceCompleteness in its output. The most common complaint about Amazon scrapers is prices going missing depending on proxy location or delivery country. This actor uses a residential tier by default and treats anything under 95% as a defect, not an acceptable outcome.
• Three surfaces, one schema. Search, bestsellers and detail all produce the same record.
• Deduplicated across surfaces — and billed once. A product found in search, in a bestsellers grid and on its detail page is returned once, with every sighting listed in sources[]. You are not charged three times for one product.
• Sponsored results are excluded by default and always flagged sponsored: true when you include them. Ads are never silently mixed into organic results.
• Prices are parsed, not scraped as strings. { amount, currency, raw }, with the currency recorded as published and no FX guessing. A price that cannot be read is null — never a guess.

Output

{
  "asin": "B0GJTFXNRX",
  "title": "Apple AirTag (2nd Generation): Tracker for Keychain, Wallet, and More",
  "url": "https://www.amazon.com/dp/B0GJTFXNRX",
  "surface": "bestsellers",
  "brand": "Apple",
  "price": { "amount": 24.0, "currency": "USD", "raw": "$24.00" },
  "listPrice": { "amount": 29.0, "currency": "USD", "raw": "List: $29.00" },
  "discountPct": 17,
  "rating": 4.5,
  "reviewCount": 9464,
  "bestsellerRank": 2,
  "category": "electronics",
  "sponsored": false,
  "availability": "In Stock",
  "features": ["..."],
  "sources": [{ "surface": "bestsellers", "url": "https://www.amazon.com/dp/B0GJTFXNRX" }]
}

The run also writes a summary to the key-value store under OUTPUT, including priceCompleteness, degradedSources and creditsSpent. Check degradedSources to tell a genuinely empty result apart from a source that failed.

Input

Field	What it does
searchQueries	Keywords to search. Amazon caps keyword search at ~7 pages.
bestsellerCategories	Category slugs, e.g. electronics. Returns ranked tiles.
asins	Specific ASINs to fetch detail pages for.
scrapeProductDetails	Enrich every result with its detail page (premium event).
includeSponsored	Include ad placements, always flagged. Default off.
domain	www.amazon.com, www.amazon.co.uk, www.amazon.de, …
cheapMode	Cheaper proxy tier. Lowers price completeness to ~84%. Default off.
scrapeDoToken	Optional. Leave empty — access is included. Only for users who prefer their own scrape.do account.

Pricing (pay-per-event)

Event	Fires	Price
product-scraped	per unique product delivered, after deduplication	$0.003 ($3.00 / 1,000 products)
product-detail	per product enriched with a full detail page	$0.005 ($5.00 / 1,000 details)
Actor Start	once per run, charged by the platform	$0.00005 ($0.05 / 1,000 runs)
Result	per record stored, charged by the platform	small, set by the platform

product-detail is charged in addition to product-scraped, not instead of it, and only over records actually delivered. You pay for delivered records, not runtime. Runs respect maxTotalChargeUsd: when the cap is reached the actor stops delivering, reports how much of the result set was delivered, and never charges past it.

Platform usage — compute and network — is bundled. There is no separate usage line on your bill.

Using it from an AI agent (MCP)

The actor is callable as an MCP tool via Apify's MCP server. Example prompts:

"Get the top 50 laptops on Amazon under $600 with their ratings and review counts."

"What are the current top 20 bestsellers in electronics, with rank and price?"

"Fetch full detail for ASIN B0GJTFXNRX including feature bullets and availability."

Every field is documented in the output schema, so an agent can plan against it without scraping the README.

Honest limits

• Customer reviews are not available. Amazon's /product-reviews/ pages require a signed-in session, so this actor returns the rating and review count from public pages but not review text. Any actor offering full review text is doing something this one deliberately does not.
• Amazon caps keyword search at roughly 7 pages. Use bestseller categories or more specific keywords for deeper coverage.
• Bestseller tiles sometimes omit prices. Those rows are excluded from the priceCompleteness metric rather than being allowed to dilute it.
• Availability and delivery text vary by marketplace and region.