2,500 free credits — no credit card required. Start building
Give the API an ASIN and get the reviews back as flat rows — rating, text, date, reviewer, permalink. One call, no browser extension, no paging logic to write.
2 endpoints · 7 credits per review · up to 100 per product
Amazon shows reviews ten to a page behind a lazy-loading widget. Copying them by hand stops being viable somewhere around the second product, and browser extensions break whenever the page markup shifts.
The export path is an API call instead: pass an ASIN, get every review the endpoint can reach as a JSON array where the rating is a number and the date is an ISO timestamp. Write it to CSV, load it into a sheet, or pipe it straight into a sentiment model.
If you only want the complaints, there is a second endpoint that filters to 1- and 2-star reviews server-side, so you are not paying to page through the praise to find them.
One GET, one header. Everything else is a query parameter, so you can test it in a browser tab before you write any code.
curl "https://sensecollect.com/v1/amazon/reviews?asin=B09G9FPHY6&limit=100" \ -H "x-api-key: $SENSECOLLECT_KEY"The same record shape every SenseCollect source answers with. Fields Amazon cannot fill come back empty rather than missing, so your parser never has to branch.
The ASIN is in every Amazon product URL, after /dp/. Pass one, or several separated by commas to export a whole product line in a single call.
GET /v1/amazon/reviews with your key in the x-api-key header. Set limit up to 100 reviews per product; the default is 10 so a test call stays cheap.
data is a flat array — hand it to csv.DictWriter, a dataframe, or your warehouse loader. Nothing is nested more than one level except raw.
Every review carries a stable id. Keep the ids you have already stored and a weekly re-run only adds what is new.
Reviews are billed per row returned, so a product with twelve reviews costs twelve rows — not a per-call minimum rounded up.
The questions that come up most often before someone runs their first export.
Call GET /v1/amazon/reviews with the product's ASIN and a limit, then write the data array to a CSV file. Every record is flat — rating, content, published_at, author.name, url — so a standard CSV writer handles it without any reshaping. There is no separate export step and no file to wait for; the response is the export.
Amazon's own Product Advertising API does not return review text — it returns a rating summary and an iframe. SenseCollect's /v1/amazon/reviews endpoint returns the review bodies themselves as structured records, which is what most people are actually looking for when they search for an Amazon reviews API.
Up to 100 reviews per ASIN per call, which is the ceiling Amazon's public review pages expose. Pass several ASINs in one call to cover a product line, and re-run on a schedule to accumulate more over time — the stable review id makes deduplication a set difference.
Yes. GET /v1/amazon/complaints runs the same collection but filters to 1- and 2-star reviews before the response is built, so you get the complaints without paying to page past the five-star ones.
Reviews on a public product page are public data, and collecting public data for analysis is broadly accepted. That is not legal advice: how you use the text matters more than how you got it, republishing wholesale is a different question from analysing it, and you should read Amazon's terms and your own jurisdiction's rules before you build a product on it.
You can browse and run endpoints in the explorer without a key. To pull real rows you need an account, which starts with 2,500 free credits — around 830 review rows — and no card.
2,500 free credits on signup, no card. That is roughly 830 review rows to test the shape against your own pipeline.