Amazon Ratings Scraper API — star ratings and the full rating histogram by ASIN

Give it ASINs, get back each product's average rating, total rating count, and the complete 1–5 star breakdown — plus estimated counts per star level, which Amazon shows as percentages only.

The distribution covers every rating on the listing, not the handful of reviews visible on the page. On a product with 635,244 ratings you get the shape of all 635,244, not a sample of eight.

Unofficial tool. Not affiliated with, endorsed by, or connected to Amazon.com, Inc.

---

What a product row looks like

{
  "rowType": "product",
  "asin": "B07PXGQC1Q",
  "marketplace": "amazon.com",
  "countryCode": "US",
  "averageRating": 4.7,
  "totalRatings": 635244,
  "ratingHistogram":       { "fiveStar": 87, "fourStar": 8, "threeStar": 3, "twoStar": 0, "oneStar": 2 },
  "estimatedRatingCounts": { "fiveStar": 552662, "fourStar": 50820, "threeStar": 19057, "twoStar": 0, "oneStar": 12705 },
  "positiveSharePct": 95,
  "negativeSharePct": 2,
  "histogramAverage": 4.78
}

Measured coverage: totalRatings and averageRating on 95.7% of product rows, the full histogram on 89.4%.

The fields Amazon does not give you

Amazon publishes the percentages and the grand total but never the number of ratings at each star level. This Actor works them out:

• estimatedRatingCounts — how many 1-star ratings a product actually has, in units rather than percent. Derived as percentage × total and normalised against the real percentage sum, because Amazon's five buckets round and rarely add to exactly 100. They are labelled estimated because they are: treat them as close, not exact.
• positiveSharePct / negativeSharePct — the 4★+5★ and 1★+2★ shares, ready to chart without doing the arithmetic per row.
• histogramAverage — the mean recomputed from the distribution. Amazon's own averageRating is the authority; a gap between the two flags a listing whose displayed score and published distribution disagree, which is worth a look.

Every one of these is null rather than zero when the underlying data is missing. A guessed rating is worse than a missing one.

How to scrape Amazon star ratings by ASIN

{
  "products": ["B07PXGQC1Q", "B09B8V1LZ3", "B0CWXNS552", "B08H93ZRK9"]
}

Bare ASINs, or Amazon URLs pasted straight from the browser — both work, mixed in the same list.

How to compare Amazon ratings across countries

Every marketplace keeps its own rating pool, with its own average, its own total and its own distribution. The same product can be 4.7 in the US and 3.8 in Italy. Add the marketplaces you sell in:

{
  "products": ["B07PXGQC1Q"],
  "marketplaces": ["amazon.com", "amazon.co.uk", "amazon.de", "amazon.it", "amazon.ca"]
}

You get one product row per marketplace, each with its own histogram and rating count — so you can weight them properly instead of averaging averages. Measured on one product: 4.7 across 635,244 US ratings, 4.6 across 21,906 in Canada, 3.8 across 9 in Italy. All 24 Amazon retail marketplaces are supported.

A product not sold in a marketplace returns a free not-listed-in-marketplace row rather than vanishing, so a missing country is never ambiguous.

If Amazon serves a page whose reviews section did not render, you get a free reviews-section-missing row instead of a histogram silently reported as absent — the Actor retries first and never guesses.

How to get the Amazon rating breakdown for many products at once

Paste as many ASINs as you like in one run. Each returns its own rows, and a bad ASIN never stops the run — it gets an invalid-input row and the rest continue.

Use Maximum reviews to put a hard, exact cap on any run's bill.

How to export Amazon ratings to CSV or Excel

Every run's dataset exports as JSON, CSV, Excel, XML or HTML from the Apify UI or the API. A prepared "Product ratings and histograms" view puts the rating fields in a clean table with the review rows out of the way.

How to call the Amazon Ratings API from Python

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("vonsensey/amazon-ratings-scraper-api").call(run_input={
    "products": ["B07PXGQC1Q", "B09B8V1LZ3"],
    "marketplaces": ["amazon.com", "amazon.de"],
})

for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    if row["rowType"] == "product":
        print(row["asin"], row["marketplace"], row["averageRating"],
              row["totalRatings"], row["ratingHistogram"])

You also get the reviews

Every run returns the reviews Amazon publishes on each product page as well — rating, title, full text, date, verified-purchase badge, helpful votes, variant and images — at 100% fill on the core fields. Amazon publishes roughly 8–13 per product per marketplace; that is the public ceiling for every tool, since review-specific pages now require a login.

For ratings work the histogram is usually the better input anyway: it describes every rating on the listing, while any review sample describes only what Amazon chose to show.

What it costs

Charged per review returned. The rating data is not billed at all — histograms, averages, totals, estimated counts and shares are always free.

In practice that means roughly $0.04 per product per marketplace on the free tier (about 8 reviews × $0.005), and you can compute your bill before you run it: reviews returned × your tier price.

• No start fee.
• Failures are free — blocked pages, products not listed in a marketplace, products with no reviews and invalid inputs all cost nothing.
• Product rows are free, always.
• Duplicates are free — the same review found on two marketplaces is charged once.
• Turn on Charge once per review, ever for scheduled runs and previously-paid reviews stay free.

Personal data and GDPR

Reviewer names, profile IDs and profile URLs are personal data and are off by default. Rating data contains none of it. This Actor reads only what Amazon publishes publicly; it never logs in.

Proxy

Required and on by default — Amazon serves a CAPTCHA to unproxied cloud IPs. The default works; no residential proxy add-on needed.

Related

Need to watch products over time? Use Amazon Review Monitoring API, which bills each review only once no matter how often you poll. Need the reviews themselves as the main event? Use Amazon Product Reviews Scraper API.

---

Unofficial tool. Not affiliated with, endorsed by, or connected to Amazon.com, Inc. "Amazon" is a trademark of Amazon.com, Inc., used here only to describe what this tool reads. Publicly available data only.