An Amazon scraper API is an endpoint that takes a search keyword and returns Amazon's result cards as structured rows — ASIN, title, brand, current and list price, star rating, review count, sponsored flag, image and a clean /dp/ URL — without you opening a browser, holding a session or maintaining a parser.
The hard part of scraping Amazon has never been the HTML. It is staying unblocked: Amazon fronts suspicious traffic with a bot-check interstitial, varies its layout per marketplace and per test bucket, and treats a datacenter IP that requests fifty pages in a minute exactly the way you would expect. This collector absorbs all of that — it reads the server-rendered page over our TLS tier without a browser, and when Amazon answers with the interstitial instead of results it retries the page on a fresh residential exit until it gets a real one.
$0.001 per delivered product, up to 200 products per run. Nothing delivered means nothing charged.
One POST carries the whole job. You send a keyword, optionally a country and how many products you want; the collector resolves the country to the right Amazon marketplace and to a proxy exit in that country, walks the result pages in order, parses each card into a typed row and stops when it reaches your max_results.
Three details matter more than they sound:
rank across the whole run and the page it came from, so you can reconstruct pagination or ignore it entirely.The input is a search, not a URL list. You never build /s?k=… query strings, never track the page parameter and never discover that Amazon moved the pagination token — that is the collector's problem. country does double duty: it picks the marketplace (amazon.co.uk, amazon.de, amazon.it…) and pins the proxy exit to the same country, which is what makes prices and delivery promises match what a local shopper sees.
Every delivered product is one flat row with the fields below. Nullable means Amazon did not show it on that card — the field comes back empty rather than guessed, so a missing brand is information, not a parser failure. Two fields do quiet work: price_value is the numeric twin of the displayed price, so you can sort and diff without parsing currency strings, and sponsored lets you drop paid placements before they pollute a rank tracker.
$0.001 per delivered product ($1 per 1,000). Nothing delivered means nothing charged, and the $2 monthly free credit covers roughly 2,000 products here. Volume tiers take up to 30% off.
A worked example beats a price list. Tracking 500 keywords daily at 40 products each is 20,000 delivered products, so $20 a day at list price — and less on a volume tier, which takes up to 30% off. The same job on a self-hosted stack is 20,000 page fetches through residential bandwidth plus the retries Amazon forces on you, plus a parser somebody owns.
What you are not billed for matters just as much: a blocked page, a bot-check interstitial, a retry on a fresh exit and a run that comes back empty all cost nothing. You pay for delivered rows.
Amazon does publish an official API — the Product Advertising API — and for a lot of teams it is the wrong tool for a reason that has nothing to do with quality: eligibility. PA-API is an affiliate product. You need an Associates account in good standing, and Amazon expects that account to keep making qualifying sales; access is throttled against those sales and can be withdrawn. A price-monitoring team, a brand protecting its listings and a researcher building a dataset are none of those things.
Beyond eligibility, the two answer different questions. PA-API answers "tell me about this item I already know", keyed by ASIN. This collector answers "what does a shopper searching this keyword actually see" — including the sponsored placements PA-API has no reason to expose, in the order Amazon really ranked them, from an IP in the country you care about. If you already hold the ASINs and just want catalogue attributes, use the Amazon product API collector instead, which is the ASIN-keyed half of the same pair.
A self-hosted Amazon scraper is a weekend to build and a permanent job to keep alive. The public GitHub projects that rank for this query mostly demonstrate the parse, which is the easy 20%. The other 80% is the part that decides whether you have data on Monday morning:
The trade is straightforward: you give up control of the parser and get a published output schema, someone else's on-call, and a bill that only moves when rows arrive. If you would rather own the fetch layer and just want the network under it, our residential proxies are the same exits this collector runs on.
Run your keyword set on a schedule and diff price_value per ASIN over time. Because delivery promises and prices are pinned to the exit country, a EU/US price comparison is two runs with a different country, not two separate stacks.
Ask where your ASINs land for the keywords that matter, and use the sponsored flag to separate what you earned from what you paid for. Global rank makes the series comparable across days even when Amazon changes page size.
New ASINs appearing against your keywords are new competitors. A daily run over a category's keyword set gives you the delta without anyone reading a search page.
Search is the cheap discovery step: collect ASINs at $0.001 each, then push only the ones you care about through the Amazon product API collector for the full detail page. Discovery wide, enrichment narrow.
One run delivers up to 200 products; the default is 40. Pagination happens inside the run, so you never handle offsets — for wider coverage, run the same keyword against several countries, or several keywords in parallel, each billed on what it actually delivers. Throughput follows your plan's rate limit rather than the collector: 60 requests/minute on pay-as-you-go, up to 1,200 on the top tier.
On the legal side: collecting publicly visible listing data is generally lawful in most jurisdictions, and courts have repeatedly declined to treat public web data as computer intrusion — but Amazon's terms of service restrict automated access, the analysis differs by country, and none of this is legal advice. Personal data carries its own rules regardless of where you found it. Get advice for your actual use case.
There is a free allowance rather than a free product: every account gets $2 of usage each month with no card, which at $0.001 per product is about 2,000 delivered products. That is enough to run a real keyword set for a week and measure the data before you spend anything. Genuinely free scrapers exist on GitHub — the difference is who maintains the parser and the exit pool when Amazon changes something.
$0.001 per delivered product — $1 per 1,000 — with no subscription and no credit pack to buy up front. Volume tiers take up to 30% off that unit price, and GET /v1/scraper/collectors returns the exact price your API key pays.
It is one HTTP POST, so requests is enough — no SDK to install. Post your keyword to the collector's run endpoint with your key in the Authorization header and read results from the JSON response. The Python, Node and cURL versions of the same call are on this page.
No. The collector runs on our residential network and handles rotation, retries and the bot-check interstitial internally. Your API key is the only credential you manage. If you would rather run your own fetcher, you can buy the same residential exits directly instead.
Not this collector — it delivers search result cards, where Amazon publishes the rating and the review count but not the review text. Use the review count and star rating from these rows for ranking work, and treat individual review text as a separate job.
Pass an ISO country code and the collector uses that country's Amazon site and a proxy exit in the same country — so the prices, currency and delivery promises are the ones a local shopper sees rather than a US page served through a European IP.
Collecting publicly visible listing data is generally lawful in most jurisdictions, but Amazon's terms restrict automated access and the analysis differs by country and by what you do with the data afterwards. Treat anything that identifies a person under the applicable privacy rules, and take proper legal advice for your case — this answer is not it.
Skip the marketing. QuantumProxies.io publishes a machine-readable site map, Markdown for every page, a free MCP server, and APIs billed from the same balance as your proxies.
Connect in one command: npx -y quantumproxies-mcp