Amazon Reviews Scraper — public review cards and verified badges

Read available individual Amazon reviews by ASIN or product URL: native review IDs, text, ratings, dates, verified-purchase badges, helpful votes and variants. Filter stars and sort the available sample. Export JSON, CSV or Excel; free KVS reports summarize delivered reviews.

Coverage is a public window, not a full archive. Product pages may hide reviews behind sign-in or anti-bot pages. Dedicated review pages currently redirect to sign-in even through a US residential proxy. maxPages is best effort and stops at that boundary. Each result carries coverageComplete=false and freshnessVerified=false. Full history, globally newest reviews, replies and exact cross-run deduplication are unavailable.

First run

{"asins":["B09G9FPHY6"],"maxResults":5,"maxPages":1,"outputPreset":"compact"}

Omitting input uses this ASIN, a 25-review cap and full output. Explicit asins: [] collects no products. Always inspect SUMMARY and ERRORS, even after a SUCCEEDED platform run.

Source

Paid rows require an identified individual-review card, native R… ID, displayed rating and title or body. Aggregate ratings, Customers Say summaries and errors never enter the Dataset. Missing fields remain null. A verified badge becomes true; no badge means unknown. Reviews can describe variants of the requested product. Native IDs replace pre-1.3 content hashes; migrate downstream comparison keys.

Direct HTTP is tried first, then up to three regional Apify RESIDENTIAL attempts if no reviews appear. Country follows the marketplace. This uses your Apify proxy access without an external provider key. Proxy and platform usage are additional user-paid costs; residential bandwidth can dominate the review fee. Disable useResidentialFallback to avoid residential costs, which can yield zero rows. Diagnostics include mode, status, bytes and row counts without credentials.

Inputs

Input	Default	Behavior
asins	["B09G9FPHY6"]	Up to 25 exact ASINs or product/review URLs; deduplicate product + marketplace.
domain	amazon.com	Marketplace for bare ASINs; URLs select their own market.
maxResults	25	Hard run-wide paid cap, 1–10,000. Raising it does not expand source coverage.
maxPages	1	Best effort 1–5: product page followed by deeper review pages until empty, blocked or login required.
filterByStar	all_stars	Local filter before billing: individual 1–5; critical 1–2; positive 4–5.
sortBy	recent	Sort each product window by parsed English date or helpful count. Unknown values follow known ones. Other-language dates remain text with null parsed date.
workflow	auto	reviews standard; research guarantees rating, title, text, verified and helpful analysis fields.
outputPreset	full	compact fewer fields for AI; research normalized fields and provenance; full displayed + normalized fields.
extract*	true	Hide selected output fields; source evidence still validates the card. Title and text cannot both be disabled. Research overrides relevant toggles.
useResidentialFallback	true	Up to three regional residential attempts with user-paid proxy usage.
maxConcurrency	2	Concurrent products 1–5; output follows input product order.
timeout	120	HTTP timeout per attempt, 15–300 seconds.

Recognized domains: amazon.com, amazon.co.uk, amazon.de, amazon.fr, amazon.it, amazon.es, amazon.ca, amazon.com.au, amazon.co.jp, amazon.in, amazon.com.mx and amazon.com.br. Recognition does not guarantee live reviews on all twelve. Spoofed hosts, credentials, unsupported paths and arbitrary strings containing ASIN-like tokens are rejected. Products differ by marketplace.

{"asins":["B09G9FPHY6"],"filterByStar":"four_star","sortBy":"helpful","maxResults":5,"workflow":"research"}

Output and reports

Full output includes success, asin, domain, productUrl, reviewId, reviewUrl, sourceMode, collectedAt, reviewerName, rating, ratingValue, title, date, reviewDate, text, verifiedPurchase, verifiedPurchaseValue, helpfulCount, helpfulCountValue, country, productVariant and coverage flags. Parsed reviewDate is an ISO date when supported; collectedAt is collection time and does not prove freshness.

• SUMMARY: delivered/source/filter counts, cap, source attempts, coverage flags and outcome.
• ERRORS: stable INVALID_PRODUCT_INPUT, LOGIN_REQUIRED, SOURCE_BLOCKED, PRODUCT_NOT_FOUND, PROXY_UNAVAILABLE and identity diagnostics. Rejected raw URLs and credentials are omitted.
• FAILED_INPUTS and FAILED_PRODUCTS: invalid inputs and products with no source reviews.
• PRODUCT_SUMMARIES: histogram, average rating, verified coverage and English keyword signals from delivered rows before output-field suppression.

Summary distinguishes INVALID_INPUT, SOURCE_UNAVAILABLE, EMPTY_FILTERED, PARTIAL, BUDGET_LIMIT and SUCCEEDED. Partial rows can coexist with a deeper-page login error. Invalid and empty runs are platform SUCCEEDED with zero review charges.

Sentiment uses english_keyword_heuristic: word matching, not a language model, customer intent or a product-quality verdict. Negation, small samples and other languages can mislead it.

Pricing

$0.003 per delivered review ($3/1,000) plus $0.00005 per start event. Platform and proxy usage are paid by the user separately. Start events depend on memory (one per GB, minimum one). Invalid inputs, failed sources, duplicates, filtered reviews and reports incur no review event. maxResults and Apify's charge budget cap Actor events; they do not cap separate platform/proxy usage. Current SDK guards apply before source work and each push; only confirmed delivery enters summaries. All requested products may be fetched before the cap fills, so limit the product list too.

MCP and AI agents

Start with maxResults 1, maxPages 1 and outputPreset compact. Read Dataset, SUMMARY and ERRORS through Apify MCP storage tools. Tool success alone does not prove source availability. Research reports remain in free KVS. No login cookies or third-party key are required.

Integrations

Run Actor ID KXlDi1LT7eCGlCmRs through the Apify API or existing n8n clients with these inputs. Existing Tasks are preserved. Reviews contain public personal data and copyrighted content; collect and reuse only what you need and have rights to use. Amazon is a trademark of Amazon.com, Inc.; this independent Actor is not affiliated with Amazon.

Amazon review research from a bounded public sample

Use this Amazon review scraper to organize individual source review cards for product-feedback analysis, rating-specific research and a cited review file. It collects the public window available on the requested source, not a complete review archive. Separate source access, applied filters and delivered review counts when presenting findings.

Research task	Small configuration	Evidence to retain
First review sample	One ASIN, one page, low cap	Native review ID and URL
Low-rating feedback review	critical filter	Actual displayed rating and text
Helpful-review research	helpful ordering	Available helpful count and sample scope
Cross-product comparison	Few products, global cap	ASIN+marketplace and delivered counts
Lighter AI input	compact preset	Coverage flags and diagnostics

A two-review compact request

{
  "asins": [
    "B09G9FPHY6"
  ],
  "maxResults": 2,
  "maxPages": 1,
  "outputPreset": "compact",
  "maxConcurrency": 1
}

A two-row cap does not guarantee two accessible review cards. Read SUMMARY and ERRORS even when the platform reports SUCCEEDED. Validity requires native review identity and substantive source evidence, not a summary of overall ratings.

Inspect critical feedback with research fields

{
  "asins": [
    "B09G9FPHY6"
  ],
  "workflow": "research",
  "filterByStar": "critical",
  "maxResults": 3,
  "maxPages": 1,
  "outputPreset": "research",
  "maxConcurrency": 1
}

Critical means one- or two-star cards in the available window. It is not all critical feedback, a defect rate or a representative complaint distribution. Research ensures fields needed for summaries; it does not create missing text.

Examine displayed helpful-review evidence

{
  "asins": [
    "B09G9FPHY6"
  ],
  "sortBy": "helpful",
  "maxResults": 3,
  "maxPages": 1,
  "outputPreset": "research",
  "maxConcurrency": 1
}

Ordering is local to the collected window. It does not establish the most helpful reviews across Amazon's complete history, and absent vote values remain unknown.

Request direct-only collection

{
  "asins": [
    "B09G9FPHY6"
  ],
  "maxResults": 2,
  "maxPages": 1,
  "outputPreset": "compact",
  "useResidentialFallback": false,
  "maxConcurrency": 1
}

This disables residential fallback and its bandwidth costs, but can return zero cards. It does not remove other platform resources or start-event costs.

Interpret ratings, verification and sentiment

A verified-purchase badge is a displayed Amazon signal, not independent proof of authenticity. No badge is unknown rather than a confirmed unverified purchase. Rating histograms and averages in PRODUCT_SUMMARIES describe delivered reviews after relevant filters, not the product's complete customer population.

English keyword sentiment is a disclosed heuristic. Negation, sarcasm and other languages can be misclassified. Quote or summarize only what the actual review text supports and keep the source URL. Do not turn review topics into verified safety, quality or purchasing-intent conclusions.

ReviewDate is parsed only when supported; collectedAt is collection time. Neither makes freshnessVerified true. Retain domain with ASIN because marketplace review windows can differ. Native review IDs help comparison, but this Actor does not claim exact historical cross-run deduplication, edits or deleted reviews.

Cost and handoff examples

Two delivered reviews have $0.006 in result events, plus the start event and user-paid platform/proxy resources. Residential requests can cost more than the review fee itself. maxResults caps paid output while source requests for multiple products can still occur; limit the product list as well as the row count.

Use apivault_labs/amazon-reviews-scraper and its existing clients with the same input. MCP clients should discover apivault_labs--amazon-reviews-scraper, inspect tools/list, start small and read Dataset, SUMMARY and ERRORS from the returned run storage. Keep credentials in private configuration. JSON preserves normalized fields and coverage flags; use stored Dataset CSV/Excel export without a new run.

Amazon reviews FAQ

Can a larger maxPages unlock sign-in content? No. It is best-effort public pagination and stops at access boundaries. No login cookies are supplied.

Does zero output mean no reviews? No. Inspect source-unavailable, login, filter and budget diagnostics. Do not invent rows to complete a comparison.

Can I compare two products fairly? Report separate sample counts, filters, marketplace and scope. A shared run cap can leave uneven product coverage.

n8n and developer integrations

Use this Actor in your automation or application with the existing community integrations below. An Apify account and API token are required; Actor and platform charges follow the current pricing.

• n8n community node: n8n-nodes-apivault-amazon-reviews. Install package name n8n-nodes-apivault-amazon-reviews in a self-hosted n8n instance that permits community packages.
• Python client and examples: GitHub.
• n8n source and usage: GitHub.

These are community packages. Publication on npm does not establish n8n verification or availability in n8n Cloud. Review the linked package README for supported inputs and setup.