Amazon Product Search Scraper

Every product on an Amazon search page — price, rating, rank and demand.

Search Amazon the way you normally would. Paste the URL. Get every product behind it as structured data, 30 fields each — from any marketplace.

Why this one

List price and sale price, side by side. Both initial_price and final_price come back, so the discount is a subtraction rather than a guess. That is the field most price-tracking datasets quietly leave out.

Real demand signals, not just star ratings. bought_past_month carries Amazon's own "N bought in past month" figure, alongside the rating and the total number of ratings. Popularity and sentiment, on the same row.

Search rank, preserved. page_number and rank_on_page record exactly where each product sat in the results, and sponsored marks the paid placements. That is what makes this usable for share-of-shelf and ad monitoring, not just a price list.

Any Amazon marketplace. amazon.com, .co.uk, .de, .co.jp, .com.au, .in and the rest — the URL decides. Run several marketplaces in one go and compare.

What you get

The product — name, ASIN, brand, product URL, image URL, and the variations offered.

The money — final price, initial list price, and the currency.

The demand — rating out of 5, number of ratings, units bought in the past month, units sold, and any badge such as "Overall Pick" or "#1 Best Seller".

The placement — page number, rank on page, whether the result is sponsored, whether it is a banner product, and the total results the search reported.

The offer — Prime eligibility, Subscribe & Save, coupons, Amazon Fresh, delivery options, business type, and any sponsored video.

Sample output

One real row, abridged:

{
  "asin": "B0CF3VGQFL",
  "name": "Redragon Mechanical Gaming Keyboard Wired, 11 Programmable Backlit...",
  "final_price": 29.99,
  "initial_price": 42.21,
  "currency": "USD",
  "rating": 4.3,
  "num_ratings": 67,
  "bought_past_month": 3000,
  "badge": "Overall Pick",
  "sponsored": "false",
  "is_prime": false,
  "page_number": 1,
  "rank_on_page": 1,
  "keyword": "mechanical keyboard",
  "total_results": 30000,
  "input": { "url": "https://www.amazon.com/s?k=mechanical+keyboard" }
}

That row came from a live run — 26 of its 30 fields were populated. Export to JSON, CSV, Excel or XML, or pull it straight from the API.

Fields Amazon does not show on a search result come back empty rather than guessed at. Prices, ratings, rank and ASIN are near-universal; brand, badge, business_type and sponsored_video appear only on some listings, because Amazon only shows them on some listings.

How it works

1. Search on Amazon and copy the URL from your address bar.
2. Paste it into Amazon search URLs. Add more URLs to run several searches, or several marketplaces, at once.
3. Set a limit if you want one, and run it.

The search term and the marketplace are read out of the URL, so there is nothing to type twice.

What carries over, and what does not. The search term and the marketplace are used; Amazon's own result filters — department, brand, price band, rating, Prime — are not. A URL of /s?k=laptop&rh=p_89:Apple is collected as a plain search for laptop on that marketplace. Filter the results on your side, on the fields below, rather than in the Amazon URL.

Input

Field	Required	What it does
Amazon search URLs	yes	One or more Amazon search results URLs, from any marketplace. The search term and the marketplace are read from each URL.
Max products per URL	no	Stop after this many products from each search. Prefilled at 10, so a first run is a quick sample. Raise it up to 1,000, or clear the field to collect 1,000 per URL.
Max products in total	no	Stop the whole run after this many products, across all searches. Up to 10,000; empty means 10,000.

Limits are applied during collection wherever Amazon allows it, so a limited run genuinely does less work. Every run is bounded, so a broad search can never run away with itself.

Run it once to see the shape of the data, then open it up. The prefilled 10 products per URL keeps a first run quick and cheap.

Built to be trusted with real work

Stop a run at any time and keep what it found. Aborting stops the collection and delivers every product completed up to that moment. Nothing is abandoned and nothing half-fetched is passed off as a result.

You are never charged for a product you did not receive. Searches that failed are reported separately and never reach your dataset. Amazon returns a whole results page at a time, so a small limit can overshoot — anything past your limit is dropped before delivery and never billed.

Results come back in rank order. When a limit trims a search, what you keep is the top of the results, not an arbitrary slice.

Clean, stable output. Field names come straight from the source schema and are not renamed or reshaped between runs.

No invented data. A field Amazon does not show comes back empty. Nothing is inferred, averaged or filled in to make a row look complete.

What people use it for

• Price and promotion tracking — watch list price against sale price across a category, and catch discounts as they land.
• Share of shelf — measure how much of page one a brand occupies, organic against sponsored.
• Competitor and ad monitoring — see who is paying for placement on which search terms, in which marketplace.
• Product and demand research — rank a category by units bought last month, rating and review volume before committing to a line.
• Catalogue building — pull ASINs, brands, images and prices for a whole search in one pass.

Free accounts

On a free Apify plan a run collects up to 10 products, whatever the limits above say. Nothing beyond that is collected, so a capped run finishes quickly. Upgrade to a paid plan and the limits above apply in full.

Good to know

• A URL is checked for shape and for a search term in it. A product page or a URL with no search term is refused with a message saying so, rather than becoming a collection that returns nothing.
• Two URLs that ask the same marketplace for the same term are collected once, even if their other filters differ, because those filters are not used.
• Whether a search returns any products is not knowable until collection runs.
• Amazon paginates in whole pages of roughly 40–60 results, which is why a request for 5 products may collect a page and then trim it.
• If a search cannot be collected, it is reported in the log and counted in the run summary rather than silently dropped.