Amazon Reviews Scraper — Review Text, Ratings & Verified Data

Extract Amazon product reviews as clean, structured records: the actual review body, star rating, title, date, country, verified-purchase status, helpful votes and the variant reviewed. Works with ASINs, product URLs or a Best Sellers page, across ten Amazon marketplaces.

You are charged only for complete reviews. Anything that comes back partial is still delivered to your dataset — flagged, with the reason attached — and costs you nothing.

How it works

ASINs · product URLs · Best-Sellers page
                    │
                    ▼
        real browser + Apify proxy      ← Amazon blocks plain HTTP requests
                    │
                    ▼
     parse Amazon's current review markup
                    │
                    ▼
   completeness gate: body + rating + id + date?
          │ yes                    │ no
          ▼                        ▼
   charged `review` row     delivered FREE with
      ($0.002 each)         `validationErrors`
                    │
                    ▼
     free `product-summary` row per product
   (rating histogram, coverage note, charges)

No cookies, no login, no Amazon account. Every row states whether it was billed and why.

What data does it extract?

Two kinds of row, told apart by recordType.

review — one per review:

product-summary — one per product, always free: product title, averageRating, totalRatings, reviewsExtracted, reviewsComplete, reviewsCharged, verifiedPurchaseShare, extractedRatingCounts and a coverageNote.

Use cases

1. Product research & sourcing — pull real review text for candidate products before you commit inventory; the free summary row shows the rating distribution at a glance.
2. Review intelligence & sentiment analysis — feed complete, validated review bodies into an LLM or sentiment model without cleaning accessibility teasers and empty shells first.
3. Brand & listing monitoring — schedule runs on your own ASINs and watch new reviews, ratings and verified-purchase share over time.
4. Competitor analysis — point a run at a Best Sellers or category page and get structured review data for every product on it in one pass.

How to scrape Amazon reviews

1. Paste one or more ASINs or product URLs (any mix — the ASIN is parsed out of URLs for you), or a Best Sellers / category / search URL in discoveryUrl.
2. Optionally pick the marketplace (domain), cap the products (maxProducts) and set a hard cost ceiling (maxCostUsd).
3. Run it. Export the dataset as JSON, CSV or Excel, or read it via the API / MCP.

{
  "asins": ["B0DCH8VDXF", "B0GJTFXNRX"],
  "domain": "amazon.com",
  "maxProducts": 20,
  "maxCostUsd": 5
}

{
  "discoveryUrl": "https://www.amazon.com/Best-Sellers/zgbs/electronics",
  "maxProducts": 50
}

Input parameters

Output example

{
  "recordType": "review",
  "billable": true,
  "complete": true,
  "asin": "B0DCH8VDXF",
  "reviewId": "R2RG0Q50IBXD57",
  "rating": 5,
  "title": "Solid pair of earphones for daily use",
  "text": "Very solid pair of headphones with Apple quality. The material is premium and the fit is comfortable for long sessions…",
  "date": "2026-07-29",
  "country": "United States",
  "author": "BalkShiba",
  "verifiedPurchase": true,
  "helpfulVotes": 3,
  "variant": "Size: One Size Style: USB-C",
  "validationErrors": []
}

How many reviews per product?

Amazon publishes roughly 8–13 full review texts per product to logged-out clients. That is Amazon's limit, not this actor's, and it applies to every scraper on the market:

• ?filterByStar= and ?sortBy= are ignored on the product page — all six variants return the identical set (measured 2026-08-04)
• the reviews-render pagination endpoint returns 403
• /product-reviews/…?pageNumber=2 redirects to Amazon Sign-In

Every product-summary row states this explicitly, alongside the product's true total rating count, so you always know what you have and what you don't. If a tool promises you "all reviews" without an account, check what it actually delivers.

Pricing

Pay per event — you pay for what you get, not for compute time:

A review is complete when it has a body, a 1–5 star rating, a review id and a date. Records missing any required field are delivered free with a validationErrors array naming the field, the rule and the evidence.

One failure mode is worth calling out: Amazon renders an accessibility hint (

"Brief content visible, double tap to read full content."

maxCostUsd caps the run before it starts; if the estimate exceeds it the run exits immediately, uncharged beyond the start fee.

FAQ

How many reviews can I get per product? Roughly 8–13 full texts — Amazon's logged-out limit, stated honestly in every summary row. See the section above.

Do I pay for incomplete reviews, blocked pages or delisted products? No. All of those produce free rows that say exactly what happened. You pay the start fee and $0.002 per complete review, nothing else.

Which Amazon marketplaces are supported? Ten: US, UK, Germany, France, Spain, Italy, Canada, Australia, Japan and India, via the domain input.

Do I need Amazon cookies or an account? No. The actor reads what Amazon publishes to logged-out visitors, through a real browser and Apify proxy.

Can I use it from ChatGPT, Claude or another AI agent? Yes. Output is a flat, typed dataset with an output schema, so it drops straight into MCP tools and agent pipelines without post-processing.

Is scraping Amazon reviews legal? The actor collects only publicly available data. Reviews include author display names, which can be personal data — make sure you have a lawful basis for how you use them (GDPR etc.). This is not legal advice.

Support

Found a bug, a marketplace that misbehaves, or a field you need that isn't here? Open an issue on the actor's Issues tab — reports get fixed, not ignored. More tools: george.the.developer on Apify.