Amazon Reviews API: Collect Review Data by ASIN

An Amazon reviews API lets software retrieve product review data or feedback insights without manually copying individual reviews. The first decision is whether you need summarized themes or individual review records. Those are different datasets, and the word “reviews” in an API name does not guarantee both.

For product teams, individual reviews can reveal recurring problems with assembly, fit, packaging, or durability. For analysts, the challenge is keeping the product, marketplace, collection time, and sample size attached to every conclusion.

Which Amazon review API do you need?

Amazon offers a Customer Feedback API within its Selling Partner API. It provides customer feedback insights at ASIN and browse-node levels, with return insights at browse-node level. Its documentation describes weekly refreshes and English insights. This is a useful official route for supported feedback analysis, but it is not an unrestricted feed of every individual review.

SenseCollect's Amazon reviews API collects product reviews by ASIN and offers a separate complaints endpoint. It does not provide Amazon keyword rankings or seller feedback. If your question is “where does this product rank?”, use the Amazon SERP data guide to choose the correct source.

Start with a small ASIN sample

An ASIN identifies an Amazon catalog item. Build a small input list with the ASIN, the product you expect to receive, and the marketplace you are investigating. Check variants carefully: reviews can be presented across related product variations, so the displayed product and the reviewed variant may differ.

For SenseCollect, a starter request looks like this:

curl "https://sensecollect.com/v1/amazon/reviews?asin=B09G9FPHY6&limit=25" \  -H "x-api-key: $SENSECOLLECT_KEY"

Replace the example ASIN with your target product. The endpoint also accepts comma-separated ASINs. Its configured limit is currently up to 100 reviews per product; a requested limit is not a promise that the source will return that many records. Follow the response and job-handling instructions in the API documentation before treating a request as complete.

For a focused sample of one- and two-star reviews:

curl "https://sensecollect.com/v1/amazon/complaints?asin=B09G9FPHY6&limit=25" \  -H "x-api-key: $SENSECOLLECT_KEY"

This produces a deliberately negative sample. It can help discover failure modes, but it cannot establish the overall proportion of customers experiencing a problem.

Keep an analysis table you can audit

Preserve the original response, then create a separate analysis table. The following are recommended analysis columns, not a guarantee that every source populates every field:

• ASIN and marketplace: which product and storefront the observation concerns.
• Review identifier or source URL: a reference for deduplication and inspection.
• Review text and rating: the evidence behind each assigned theme.
• Review date and collection time: separate the customer's experience from your data collection.
• Variant, language, and verification status: retain these when available; leave missing values empty.
• Theme and analyst notes: your interpretation, kept separate from source facts.

Deduplicate before counting themes. Keep a review that mentions three problems as one review with three labels, rather than silently turning it into three dissatisfied customers. A missing review date should remain missing; replacing it with today's date will distort trend analysis.

Turn review text into useful product decisions

Start with a narrow question, such as “what makes this item difficult to assemble?” Read a sample manually and define labels before applying them across the dataset. Useful labels might include unclear instructions, missing hardware, tool requirements, and alignment problems.

Report the numerator and denominator together: “12 of 80 collected reviews mention unclear instructions” describes the sample. “15% of customers cannot assemble the product” does not follow from that evidence. A review sample excludes customers who did not review and may reflect a particular sort order or collection window.

Compare similar products using the same collection method. A recent negative-review sample for one ASIN and an all-time top-review sample for another will produce misleading comparisons, even if both files contain 100 rows.

Questions to ask before scaling

Confirm which marketplaces are covered, whether variant information is retained, what happens when fewer reviews are available, and how retries are charged. Test a popular product, a low-review product, and an unavailable ASIN before scheduling larger runs.

Check whether the provider returns collection metadata and a stable reference to the source. A large file without provenance is harder to maintain than a smaller sample you can explain and reproduce.

Frequently asked questions

Does Amazon have an official reviews API?

Amazon has an official Customer Feedback API for supported feedback insights. That is different from downloading an unrestricted collection of individual review bodies. Choose based on the actual fields and access your project needs.

Can an API return every review for a product?

Do not assume complete coverage. Provider limits, source availability, variants, sorting, and collection settings can constrain a sample. SenseCollect's configured per-product cap is not an Amazon-wide limit.

Are product reviews the same as seller feedback?

No. Product reviews describe catalog items; seller feedback concerns a seller's service. Keep these datasets separate when evaluating quality or performance.

Can I use this API to get Amazon search rankings?

SenseCollect's review endpoints do not return keyword SERP rankings. For that task, collect search result observations with query and marketplace context. Read Amazon SERP data explained, or start collecting review records from the Amazon API page.