AliExpress Reviews Scraper - reviews with rating, photos, English translation and buyer country, no login

Read the public reviews of any AliExpress product as data: stars, date, the review text with AliExpress's English translation, buyer country, the variant bought, shipping method, helpful votes, photo URLs and the buyer's follow-up comment. Or get one summary row per product with the average rating, star counts and AliExpress's own review topics. Built for dropshipping product research, supplier and quality checks, and review sentiment analysis.

Pay per result, at the rate shown in the pricing panel of this page, plus a very small per-run start fee. You are charged only for the results delivered.

Start it with no input at all and it reads two example products (about 150 reviews in about a minute), so you can see the output before you enter your own products.

What you get

• One row per review (default): rating (1 to 5 stars), date, text and textTranslated (AliExpress's English translation), buyerCountry, variant (the option bought, e.g. Color:Style 3) and skuId (empty on some top sellers, see "Pooled counts" below), images (review photo URLs), helpfulCount and notHelpfulCount, shippingMethod, and the buyer's later comment: followUpText, followUpTextTranslated, followUpDate, followUpDaysAfter, followUpImages.
• Product context on every row: productId, productUrl, reviewId, productAverageRating, productTotalReviews, and the run settings (filter, country, input, scrapedAt).
• Summary mode: one row per product with averageRating, totalReviews, positivePercent, stars5 to stars1, withPhotos, withFollowUp, fromSelectedCountry and topics (AliExpress's own review topics, each with mentions and sentiment, e.g. "poor packaging").
• Pooled counts on some top sellers. For some top sellers AliExpress pools the star votes of several identical listings: stars5 to stars1, averageRating and positivePercent then cover more reviews than totalReviews, which counts the reviews of this product ID (the ones the actor can read). In a check on 2026-10-20, 5 of the 11 products with 47 or more reviews were like this, with 3 to 5 times more star votes than totalReviews (for example 146 reviews and 749 star votes). These are AliExpress's own figures and are passed on unchanged; productAverageRating on review rows is the same pooled average. skuId is empty for such products (AliExpress sends none); variant is still filled in.
• Filters: only reviews with photos, only reviews with a follow-up comment, only buyers from one country, a star range, a start date, text only.
• No browser and no login. Public review pages only.

Input

field	default	notes
products	the two example products	product URLs (https://www.aliexpress.com/item/1005006691583009.html) or item IDs (1005006691583009); duplicates are read once; with no input at all a run reads the two example products (1005006691583009 and 1005010399232494)
outputMode	reviews	reviews (one row per review) or summary (one row per product, one request each; the review filters and limits below do not apply to it)
maxReviewsPerProduct	100	counts delivered rows, 1 to 5000
onlyWithText	true	skips star-only reviews; they are never charged
filter	all	all, image (with photos), additional (with a follow-up comment), local (buyers from country); summary mode ignores it
country	US	2 letters; the country of the local filter and of the summary's fromSelectedCountry count; US also puts American buyers first in the unfiltered list (see "Good to know")
minRating / maxRating	1 / 5	keep only reviews inside this star range
reviewsSince	empty	YYYY-MM-DD; keep reviews on or after this date (AliExpress has no date sort, so the actor reads up to 250 pages per product to find them, see "Good to know")
maxResults	2000	maximum rows per run, up to 100000
proxyConfiguration	Apify datacenter proxy	the setting the actor was tested with

Example output row (review; the text is shortened here)

{
  "type": "review",
  "productId": "1005010399232494",
  "productUrl": "https://www.aliexpress.com/item/1005010399232494.html",
  "reviewId": "60098242905437387",
  "rating": 4,
  "date": "2026-08-09",
  "text": "This is a nice set. The glass container is nicely made and the markings are well printed and easy to read. I like the ...",
  "textTranslated": "This is a nice set. the glass container is nicely made and the markings are well printed and easy to read. i like the ...",
  "buyerCountry": "US",
  "variant": "Color:Style 3",
  "skuId": "12000052282029869",
  "images": ["https://ae-pic-a1.aliexpress-media.com/kf/Aba791f9736234a40b7da56b71967e59bK.jpg"],
  "helpfulCount": 0,
  "notHelpfulCount": 0,
  "shippingMethod": "AliExpress Selection Standard Shipping",
  "followUpText": null,
  "followUpTextTranslated": null,
  "followUpDate": null,
  "followUpDaysAfter": null,
  "followUpImages": [],
  "productAverageRating": 4.6,
  "productTotalReviews": 139,
  "filter": "all",
  "country": "US",
  "input": "1005010399232494",
  "scrapedAt": "2026-10-20T10:06:57.000Z"
}

Example output row (summary mode)

{
  "type": "summary",
  "productId": "1005010399232494",
  "productUrl": "https://www.aliexpress.com/item/1005010399232494.html",
  "averageRating": 4.6,
  "totalReviews": 139,
  "positivePercent": 92.1,
  "stars5": 114, "stars4": 14, "stars3": 3, "stars2": 2, "stars1": 6,
  "withPhotos": 13,
  "withFollowUp": 2,
  "fromSelectedCountry": 18,
  "topics": [
    { "topic": "satisfactory", "mentions": 7, "sentiment": "positive" },
    { "topic": "poor packaging", "mentions": 6, "sentiment": "negative" }
  ],
  "country": "US",
  "input": "1005010399232494",
  "scrapedAt": "2026-10-20T10:06:57.000Z"
}

(topics lists every topic AliExpress returns for the product, six for this one; two are shown here.)

Measured on the platform on 2026-10-20

Runs on the Apify platform with the default datacenter proxy (256 MB of memory unless noted). None was blocked: no 403, 429, redirect or non-data page in any request.

• Default-shaped run (a popular product, an ordinary product with 139 reviews, and one product with no reviews; defaults): 154 reviews from 16 requests in 51 seconds at 512 MB and 59 seconds at 256 MB. reviewId, rating, date, buyerCountry, variant, skuId and shippingMethod were filled in 100% of rows (skuId is empty on products where AliExpress pools several listings, see "What you get"), text and textTranslated in 97.4% (the other rows carry only a follow-up comment), images in 27% (reviews with photos), follow-up comments in 3%. No duplicate reviewId. The product with no reviews gave no row and one no_reviews entry in FAILURES.
• The two example products (what a run with no input reads, everything else at its default): 154 reviews from 15 requests in 56 to 71 seconds (three runs).
• Ordinary products (ten products from search results: nine with 6 to 83 units sold, one new listing): 6 of them had reviews with words, 15 rows in all; 3 had only star-only reviews (12 in total, no words), so they gave no rows and no charge by default; 1 had no reviews. The nine products with reviews had between 1 and 10 each.
• Summary mode on those ten products plus the two products of the first run: 11 summary rows from 12 requests in 35 seconds (the product without reviews gives no row).
• Buyers from one country (filter: local, country: FR, 20 reviews): 20 rows, every buyerCountry is FR.
• Star range: 20 reviews of 1 or 2 stars on a popular product (1,404 reviews) needed 63 requests, about 4 minutes, because only about 5% of its reviews are that low.
• Start date: 30 reviews since 2026-08-01 on a product with 6,294 reviews needed 182 requests, about 13.5 minutes, because AliExpress cannot sort by date. Star-range and date filters read pages until enough reviews match, up to 250 pages (5,000 reviews) per product and up to your per-product limit.
• Speed: about 3 to 5 seconds per request (the 2-second pause plus AliExpress's answer through the proxy): median 3.2 to 3.6 seconds, mean 3.6 to 4.5 seconds, and a few requests take 10 seconds or more (about 1 in 20 in a long run; the slowest we saw: 28 seconds). Each request returns 20 reviews, so 100 reviews take about 15 to 25 seconds; 300 reviews of a popular product took 18 requests and 73 seconds.
• The 5,000-review limit (250 pages, the most AliExpress serves for any list): one run read all 250 pages of a product with 6,294 reviews in 18.7 minutes (4.5 seconds per request) and delivered 4,996 reviews with words; another run read 204 consecutive pages of the same product in 12.6 minutes (3.7 seconds per request). No request was refused. So a full 250-page product takes about 15 to 19 minutes.
• Time budget: with a 45-second timeout and a far too large job, the run stopped early (at 200 reviews in one run, at 100 in another, depending on how fast AliExpress answered) with status SUCCEEDED and RUN_SUMMARY = {"complete": false, "terminationReason": "time_budget", ...}.

Good to know

• Many reviews are stars only. On one ordinary product only 54 of 139 reviews (39%) had any words. That is why onlyWithText is on by default: star-only reviews are skipped and are never charged, and the summary mode still counts every star. A review with a photo but no words counts as star-only; switch onlyWithText off to get those too.
• No "newest first" order. AliExpress's review feed has no working date sort (the one we tested returned the identical order), so reviewsSince cannot skip ahead: the actor reads pages, at most 250 per product (5,000 reviews, all AliExpress serves for any list), until your per-product limit is reached, and keeps the reviews that match. A filter that rarely matches can read all 250 pages and return few rows. The order of the feed is not stable either; it shifts a little between requests, and the actor removes duplicates by reviewId.
• Translations. textTranslated is AliExpress's own English translation. For a review already in English it is a near copy that can differ in capital letters. The original wording is always in text.
• Privacy. Buyer names, avatars and gender are never collected. E-mail addresses and phone numbers that buyers write into review text are replaced with [email redacted] and [phone redacted], best effort. Recognised: e-mail addresses (also written out, like name (at) host (dot) com), numbers with a leading + or 0, US formats, tel: and wa.me links, and any plain run of 9 to 13 digits such as a bare national mobile number. Can still slip through: numbers split into groups without a leading + or 0 (for example 612 345 678), 8-digit local numbers and unusual spellings. A name that a buyer signs into the text cannot be removed.
• Which links work. Use the item ID from an aliexpress.com/item/<id>.html product URL. Short links (a.aliexpress.com/_...), store pages and search pages are not supported; they are listed in the FAILURES record as unsupported_input. A product with no reviews gives no row and is listed as no_reviews; neither is charged.
• Pace. The actor makes one request at a time with a 2-second pause, 20 reviews per request. On the platform that is about 3 to 5 seconds per request (see the measurements above): 100 reviews take about 15 to 25 seconds and the limit of 5,000 reviews per product (250 pages) about 15 to 19 minutes. A very large run stops shortly before the run timeout and saves what it has (raise the timeout for a very large job); the RUN_SUMMARY record says whether the run was complete (complete, terminationReason, rowsPushed).
• If the platform restarts the run. Apify sometimes moves a run to another server and starts it again; this actor then continues where it stopped instead of starting over: it keeps its place in the run's key-value store (RESUME_STATE) and reads back from the dataset which reviews it already delivered, so no review is delivered (or charged) twice, rowsPushed still equals the rows in the dataset, and RUN_SUMMARY gets a restarts count (the key is absent when the run was never restarted).
• If AliExpress blocks the requests (a 403 or 429 answer, a redirect or a non-data page), the actor stops at once, keeps the rows it already has, and does not retry or work around it. RUN_SUMMARY.terminationReason is then blocked and the page is listed in FAILURES. If the run is refused before any row was delivered, it fails with a clear message and nothing is charged.
• Country and order. Reviews are read in English (lang=en_US). The country setting also decides whose reviews come first: with US (the default) AliExpress lists American buyers first, so a capped run can be all US buyers (in one check all 100 rows of a default run, although only 15% of that product's reviews are American). Other codes, for example DE, gave a mixed list in our checks. Use one of them for a general mix, or filter: local to get one country on purpose.

Need product prices, ratings and sold counts from search results? See our AliExpress Search Scraper.

Keywords

aliexpress reviews scraper, aliexpress review scraper, aliexpress product reviews, aliexpress feedback, aliexpress ratings, dropshipping product research, review sentiment analysis, aliexpress reviews api, aliexpress photos reviews