How to Scrape Coupang Reviews and Product Q&A
An average rating does not explain whether buyers disliked the product, one specific variant, or an unanswered setup problem. To investigate that, you need review text, the exact option purchased, survey answers, media, and customer questions beside the product's rating distribution.
Our Coupang Reviews Scraper accepts public Coupang product URLs or bare product IDs. It returns one product-summary row, individual review rows, and optional question-and-answer rows. Reviews can include stars, text, date, purchased variant, the source-provided verified-purchase marker, survey answers, photos, videos, and helpful votes.
The text is usually Korean, and some reviews contain only a rating or survey answers. Keep missing text as missing, and decide how translation should work before running sentiment analysis.

Collect twenty recent reviews and the product Q&A
- Open the Actor on Apify and click Try for free.
- Add a Coupang product URL or numeric product ID under Product URLs or IDs.
- Set Max reviews per product to
20. - Choose Newest under Sort reviews by.
- Turn on Include questions and answers, then click Start.
maxReviewsPerProduct applies separately to each product. Five valid products with a limit of 20 can return up to 100 review rows, plus one summary row and any requested Q&A rows for each product.
We used this input:
{
"productUrls": [
"https://www.coupang.com/vp/products/9643918040"
],
"maxReviewsPerProduct": 20,
"sortBy": "newest",
"includeInquiries": true
}
Our example run returned one product summary, 20 reviews, and four Q&A rows in about 15 seconds. The 25-row dataset makes each row type explicit through type.
An abbreviated review row was:
{
"type": "review",
"productId": "9643918040",
"reviewId": 1009640728,
"rating": 5,
"title": "기대 이상이네요",
"reviewAt": "2026-09-17T08:31:27.000Z",
"itemName": "베이직북 16 Ultra5, 12GB, 512GB, 화이트",
"verifiedPurchase": true,
"surveyAnswers": {
"사용 목적": "그래픽/전문가용",
"무게": "적당해요",
"성능": "기대이상뛰어나요"
},
"images": ["https://thumbnail.coupangcdn.com/..."],
"helpfulCount": 1,
"url": "https://www.coupang.com/vp/products/9643918040"
}
One Q&A row showed the separate structure:
{
"type": "inquiry",
"productId": "9643918040",
"inquiryId": 160558513,
"question": "베틀그라운드 같은 게임도 돌아갈까요?",
"askedAt": "2026-09-10T10:20:04+09:00",
"answers": [
{
"answeredBy": "[COUPANG]",
"answeredAt": "2026-09-10T11:40:12+09:00",
"byAi": false
}
]
}
| Field group | How to use it |
|---|---|
productId, reviewId, inquiryId | Keep products, reviews, and questions identifiable across repeated exports. |
rating, title, content, reviewAt | Filter recent low ratings and retain the original text beside any translation or topic label. |
itemName, vendorItemId | Separate feedback about storage, size, color, bundle, or another purchased option. |
surveyAnswers, images, videos | Use structured buyer responses and media as supporting context, not as a replacement for the written review. |
ratingHistogram, ratingAverage | Read the product-level distribution supplied by Coupang before interpreting a capped sample. |
question, answers | Find recurring pre-purchase questions and inspect the answer text, timing, displayed author, and AI flag. |

The Coupang reviews Node.js example runs the same product check with apify-client and retrieves the resulting dataset.
Turn the export into a product-quality review
Start with the product row. Its histogram describes the distribution Coupang returned, while the review rows are the sample you requested. Do not recalculate the whole product rating from 20 recent reviews and present it as the marketplace average.
Then work through the review rows:
- Group by
itemNameorvendorItemIdso different variants are not mixed. - Review low ratings first, but keep the date and any missing text visible.
- Tag repeated subjects such as performance, sizing, packaging, or delivery.
- Retain the Korean source text beside any translated version.
- Count a subject only after defining a consistent tagging rule.
Use inquiry rows for a separate question: what buyers need to know before purchase. Repeated questions about compatibility, setup, dimensions, or included software can reveal information that should be clearer on your own product page. An answer labelled [COUPANG] or byAi: true is source context, not independent proof of who composed it.
For product discovery and offer data, start with the Coupang listings guide and Coupang product details guide. Feed the shortlisted product URLs into this Actor rather than collecting deep review histories for every search result.
Choose newest or most helpful reviews
Use newest for monitoring recent complaints or changes after a product update. Use best when you want Coupang's most-helpful ordering for an initial qualitative review.
Neither order creates a statistically representative sample by itself. A small newest-first export emphasizes recency; a helpful-first export emphasizes votes and Coupang's ranking. Record the selected order with your analysis.
Set maxReviewsPerProduct to 0 when you need only the rating summary. Leave includeInquiries off when customer questions are outside the task. Those choices reduce output and event charges without changing the input product list.
Pricing
Pricing as of September 21, 2026 is additive:
| Event | Price |
|---|---|
| Actor start | $0.00005 per GB of allocated memory, minimum one event |
| Product rating summary | $0.005 per product |
| Review | $0.001 per review |
| Question and answer | $0.001 per Q&A row when enabled |
Our example used 1 GB of memory and returned one product, 20 reviews, and four Q&A rows. Its Actor event charges were $0.00005 + $0.005 + (20 × $0.001) + (4 × $0.001) = $0.02905. Products that are not found are not charged for a product summary. Apify platform usage can be billed separately according to your plan. Check the run cost and live pricing before collecting many products.
Check these cases before scaling up
- A review has no text. Rating-only reviews are valid source rows. Keep
content: nullinstead of treating them as neutral sentiment. - The sample and product average differ. The product summary and capped review sample answer different questions. Keep both and state the selected sort order.
- Several variants appear together. Group by the purchased
itemNameor identifiers before attributing a complaint to the whole product. - A photo or video URL stops working. The files remain hosted by Coupang. Download media you are permitted to retain when it is essential to the analysis.
- Automatic translation changes the meaning. Preserve the original Korean text and review ambiguous or product-specific terms manually.
- A product is missing. Confirm the public URL or numeric ID. Withdrawn and unknown products return a status row rather than fabricated review data.
Frequently asked questions
Can I use a Coupang product ID instead of a full URL?
Yes. Add either a public Coupang product URL containing /vp/products/ or its bare numeric product ID.
Does maxReviewsPerProduct apply to the whole run?
No. It applies separately to each product, while the product rating summary is returned even when the review limit is zero.
Does a Coupang Q&A answer prove that the seller wrote it?
No. The Actor returns the answer text, displayed author label, date, and source-provided AI flag. Treat those fields as Coupang's labels rather than independent identity verification.
