Amazon Search Results API: Page-One Results as JSON, Ads Flagged

An Amazon search results API turns a keyword into the products Amazon shows for it. With SellerMagnet, one GET request to /api/amazon-search with q, a marketplace ID and an optional count returns up to 50 results from the first results page as JSON: position, ASIN, title, price, rating, review count and a sponsored flag. It costs one credit per call. There is no page 2, sort or search volume - this reference shows exactly what you get.

What does the Amazon search results API return?

Each call returns data.searchResults, a list of products in the order the first results page shows them. position starts at 1 and counts every returned product, sponsored ones included. Products without a displayed price are skipped and do not use up a position, so position is the product's place among priced results, not a pixel slot. The response for q=playstation 5 on amazon.com looks like this:

Trimmed /amazon-search response (example values from the API docs)

{
  "success": true,
  "data": {
    "searchResults": [
      {
        "position": 1,
        "asin": "B0CL5KNB9M",
        "productTitle": "PlayStation®5 Digital Edition (slim)",
        "listingPrice": {"price": {"currency_code": "USD", "currency_symbol": "$", "total": "449.00"}},
        "on_sale": false,
        "reviewRating": 4.7,
        "reviewAmount": 7209,
        "sponsored": false,
        "discount_info": "",
        "mainImage": "https://m.media-amazon.com/images/I/51fM0CKG+HL._AC_UY218_.jpg",
        "link": "https://www.amazon.com/dp/B0CL5KNB9M"
      },
      {
        "position": 2,
        "asin": "B0DGY63Z2H",
        "productTitle": "PlayStation 5 Pro Console",
        "listingPrice": {"price": {"currency_code": "USD", "currency_symbol": "$", "total": "699.00"}},
        "on_sale": false,
        "reviewRating": 4.4,
        "reviewAmount": 1373,
        "sponsored": false
      }
    ]
  }
}

Get Amazon search results as JSON with curl

Send q (the keyword), marketplaceId and optionally count (1 to 50, default 30) with your key in the X-Api-Key header. marketplaceId takes any of the 23 supported marketplaces, for example ATVPDKIKX0DER for amazon.com. Allow up to 120 seconds: the server waits that long for a slow page. This curl call prints position, sponsored flag, price and ASIN per result:

Page-one results for one keyword with curl and jq

curl -sG "https://sellermagnet-api.com/api/amazon-search" \
  -H "X-Api-Key: $SELLERMAGNET_API_KEY" --max-time 120 \
  --data-urlencode "q=playstation 5" \
  --data-urlencode "marketplaceId=ATVPDKIKX0DER" \
  --data-urlencode "count=20" \
  | jq -r 'if .success then .data.searchResults[]
           | [.position, (if .sponsored then "AD" else "-" end), .listingPrice.price.total, .asin]
           | @tsv
         else error(.message) end'

Separate sponsored and organic results in Python

To use the Amazon search API from Python, filter on sponsored and number the rest yourself: the organic rank is the position among the non-sponsored results. Parse the price with Decimal so "449.00" stays exact. get(), new_session() and ApiError come from the Python quickstart, which also has a retry helper for 429 and 5xx answers.

search_results.py - page one to CSV with an organic rank

"""Page-one Amazon search results as CSV, sponsored flagged, organic rank added (1 credit)."""
import csv
import sys
from decimal import Decimal

from sellermagnet import ApiError, get, new_session


def search(query: str, marketplace_id: str, count: int = 50) -> list:
    if not 1 <= count <= 50:
        raise ValueError("count must be 1-50; a count above 50 is a billed 400")
    results = get(new_session(), "amazon-search", q=query, marketplaceId=marketplace_id, count=count)["searchResults"]
    rows, organic_rank = [], 0
    for r in results:
        organic_rank += not r["sponsored"]
        price = r["listingPrice"]["price"]
        rows.append({
            "position": r["position"],
            "organic_rank": None if r["sponsored"] else organic_rank,
            "asin": r["asin"],
            "price": Decimal(price["total"]),  # a string such as "449.00"
            "currency": price["currency_code"],
            "rating": r["reviewRating"],       # None when the page shows no stars
            "reviews": r["reviewAmount"],      # 0 when no count is shown
            "title": r["productTitle"],
        })
    return rows


if __name__ == "__main__":
    try:
        rows = search("playstation 5", "ATVPDKIKX0DER", count=20)
    except ApiError as err:
        sys.exit(f"search failed: {err}")  # status and the API's message
    print(f"{len(rows)} results, {sum(r['organic_rank'] is None for r in rows)} sponsored")
    with open("search.csv", "w", newline="", encoding="utf-8") as f:
        writer = csv.DictWriter(f, fieldnames=["position", "organic_rank", "asin", "price", "currency",
                                               "rating", "reviews", "title"])
        writer.writeheader()
        writer.writerows(rows)

Sponsored detection reads Amazon's ad label markup; the text fallback only knows the English "Sponsored" and German "Gesponsert" labels. For keyword-level analysis built on this list - review barrier, price band, title terms - see the keyword research guide; to follow one ASIN's position over time, use the keyword rank tracker.

Can I get page 2 or sort the results?

No. The endpoint reads only the first results page in Amazon's default order, across all departments; it has no page, sort or category parameter. count caps how many of that page's priced products come back, up to 50. For deeper coverage, search narrower keywords ("playstation 5 controller" rather than "playstation"), or read a category's ranked list from the bestsellers endpoint, which takes a category ID.

• No search volume, no autocomplete and no reverse-ASIN lookup: the API returns what the page shows.
• No price filter: filter the returned list in your code.
• geo_location (a ZIP or postcode) localises the results; it is best-effort, and the response does not say which location was used.

Search errors and what they cost

Checks before billing are free; anything the endpoint itself rejects is charged. A missing or blank q or an unknown marketplace ID is a free 400, while a count above 50 is a billed 400. A keyword whose page has no priced products is a normal 200 with an empty list, and is billed like any answer.

Bottom line: one /api/amazon-search call gives you page one for a keyword as typed JSON - position, ASIN, title, a string price, rating, reviews and the sponsored flag - for one credit. Every parameter is on the search endpoint page, and a free account includes 150 credits after e-mail verification.