Amazon Keyword Research API: Competition Signals from One Search Call

For Amazon keyword research through an API, request the keyword's first results page from /api/amazon-search and read the competition off it: how many results are sponsored, the price band, how many reviews the top 10 organic products have, and which words their titles share. One call costs one credit and returns up to 50 results. It does not tell you search volume - it tells you what a new product would compete against on that page.

What can one search call tell me about a keyword?

Each result carries position, asin, productTitle, the price as a string in listingPrice.price.total, reviewRating (null when the page shows none), reviewAmount (0 when no count is shown or found), on_sale (true when a higher struck-through or list price is shown) and the sponsored flag. Products without a displayed price are skipped. That is enough for six signals.

How do I measure keyword competition in Python?

keyword_report() makes one call and returns the signals as a dict; the block at the end compares keywords, keeps those whose organic price band fits your price and sorts them by review barrier. Title terms are split on letters and digits, so they work in German or French - not in Japanese, which does not separate words with spaces. The stop-word list is English and short on purpose. get() and new_session() come from the Python quickstart.

keyword_report.py - six signals from one search call

"""Competition signals for one Amazon keyword from a single search call (1 credit)."""
import re
from collections import Counter
from statistics import median, quantiles

from sellermagnet import get, new_session

STOP = {"and", "for", "the", "with", "pack", "set", "new", "of", "in", "to", "a", "by"}


def keyword_report(query: str, marketplace_id: str, count: int = 50) -> dict:
    results = get(new_session(), "amazon-search", q=query, marketplaceId=marketplace_id, count=count)["searchResults"]
    organic = [r for r in results if not r.get("sponsored")]
    top10 = organic[:10]
    prices = sorted(float(r["listingPrice"]["price"]["total"]) for r in organic)
    words = Counter(w for r in organic
                    for w in set(re.findall(r"[^\W_]+", (r.get("productTitle") or "").lower()))
                    if len(w) > 2 and w not in STOP)
    p25, p50, p75 = quantiles(prices, n=4, method="inclusive") if len(prices) > 1 else (prices * 3 or [None] * 3)
    ratings = [r["reviewRating"] for r in top10 if r.get("reviewRating")]
    return {
        "keyword": query,
        "results": len(results),
        "sponsored_share": round(100 * (len(results) - len(organic)) / len(results)) if results else None,
        "price_p25_median_p75": tuple(round(p, 2) if p is not None else None for p in (p25, p50, p75)),
        "top10_median_reviews": median(r.get("reviewAmount") or 0 for r in top10) if top10 else None,
        "top10_under_100_reviews": sum((r.get("reviewAmount") or 0) < 100 for r in top10) if top10 else None,
        "top10_avg_rating": round(sum(ratings) / len(ratings), 2) if ratings else None,
        "on_sale_share": round(100 * sum(bool(r.get("on_sale")) for r in results) / len(results)) if results else None,
        "title_terms": [w for w, _n in words.most_common(8)],
    }


if __name__ == "__main__":
    my_price = 34.99                                   # keep keywords whose organic price band fits
    reports = [keyword_report(kw, "ATVPDKIKX0DER") for kw in ["usb c charger", "usb c charger 100w", "gan charger"]]
    fits = [r for r in reports if r["price_p25_median_p75"][0] is not None
            and r["price_p25_median_p75"][0] <= my_price <= r["price_p25_median_p75"][2]]
    for r in sorted(fits, key=lambda r: r["top10_median_reviews"]):
        print(r["keyword"], r["top10_median_reviews"], r["price_p25_median_p75"], r["title_terms"][:5])

Example output for one keyword (illustrative values)

{'keyword': 'usb c charger', 'results': 50, 'sponsored_share': 20,
 'price_p25_median_p75': (26.78, 40.55, 47.79), 'top10_median_reviews': 267.5,
 'top10_under_100_reviews': 2, 'top10_avg_rating': 4.15, 'on_sale_share': 32,
 'title_terms': ['usb', 'charger', '65w', 'gan', 'fast', 'wall', 'charging', 'iphone']}

How do I compare keywords and pick the realistic ones?

Run the report for a list of related keywords and sort by the review barrier. A broad keyword where the top 10 have hundreds of reviews each is a long-term goal; a narrower one where half the top 10 have fewer than 100 reviews is one a new product can reach. Check the price band before committing: a keyword is only useful if your price lands inside it. Once you are listed, the keyword rank tracker follows your position, and the market size guide estimates what the category sells.

Can I check a keyword from the command line?

Yes. For a quick look before writing code, jq can count sponsored results and list the review counts of the first organic ones.

Sponsored count and top-10 organic review counts with jq

curl -sG "https://sellermagnet-api.com/api/amazon-search" \
  -H "X-Api-Key: $SELLERMAGNET_API_KEY" --max-time 120 \
  --data-urlencode "q=usb c charger" \
  --data-urlencode "marketplaceId=ATVPDKIKX0DER" \
  --data-urlencode "count=50" \
  | jq -r 'if .success then (.data.searchResults
           | "sponsored: \(map(select(.sponsored)) | length) of \(length)",
             "top-10 organic reviews: \([.[] | select(.sponsored | not)][:10] | map(.reviewAmount) | join(" "))")
           else error(.message) end'

Bottom line: one search call per keyword gives the sponsored share, the price band, the review barrier and the vocabulary of the page - enough to shortlist keywords a new product can realistically reach. Every field is on the search endpoint page, and a free account includes 150 credits - 150 keywords.