How to Find Amazon Category IDs for the Bestsellers API

An Amazon category ID, also called a browse node ID, is the identifier at the end of a Best Sellers URL: in amazon.com/.../zgbs/electronics/281407 it is 281407. The SellerMagnet bestsellers endpoint takes it as category_id, together with a marketplace ID. Top-level categories use short slugs such as electronics; everything below them uses numeric IDs, and those numbers are different in every marketplace.

What is an Amazon category ID?

A browse node is one category in Amazon's catalogue tree, and its ID is how Amazon addresses it. The tree is deep and different in every storefront: as of September 2026 the amazon.de tree has 32,456 categories across 51 top-level departments and up to 9 levels, while amazon.com has 28,610 categories under 27 departments and goes 12 levels deep.

Two kinds of ID appear in the tree. The top level uses readable slugs - electronics, beauty, computers - which repeat across marketplaces. Every level below uses a numeric browse node ID, which does not: comparing the amazon.de tree with the amazon.co.uk and amazon.com trees finds zero numeric IDs in common. An ID copied from one marketplace is meaningless in another.

How do I find an Amazon category ID?

1. Open Best SellersOn the Amazon storefront you need, open the Best Sellers page.
2. Go to the categoryClick down through the departments until you reach the list you want.
3. Copy the last URL segmentThe part after the final slash is the category ID, for example 281407.
4. Confirm it in the treeCheck the ID exists in the category tree of the same marketplace before calling the API.

Read it from the Best Sellers URL

Open the category on Amazon's Best Sellers page and take the last path segment of the address. For Electronics > Accessories & Supplies on amazon.com the URL ends in /zgbs/electronics/281407, so the category ID is 281407. For a top-level department it is the slug itself, such as electronics. Ordinary category pages carry the same browse node ID in a node= parameter, such as /b?node=281407. Seller Central's Product Classifier and Browse Tree Guides list the same IDs for listing products; for API work the public tree below is quicker.

Search the category tree

To search by name instead, use the Amazon category explorer, or load the JSON tree behind it: one file per marketplace, where every node carries CategoryId, CategoryName, CategoryLink and its Subcategories. Names are in the marketplace's language, so the amazon.de tree says Elektronik & Foto. Amazon can rename, move or retire deep categories, so re-load the tree rather than hard-coding deep IDs for months. The script below finds every category whose name matches and prints its full path.

find_category.py - search a marketplace's category tree by name

import sys

import requests

TREE_URL = "https://sellermagnet-api.com/static/data/{code}.json"  # US, DE, UK, JP ...


def walk(nodes, path=()):
    for node in nodes:
        here = path + (node["CategoryName"],)
        yield here, node
        yield from walk(node.get("Subcategories") or [], here)


def find(code: str, text: str) -> None:
    tree = requests.get(TREE_URL.format(code=code), timeout=60).json()
    for path, node in walk(tree):
        if text.lower() in path[-1].lower():
            print(f"{node['CategoryId']:>14}  {' > '.join(path)}")


if __name__ == "__main__":
    find(sys.argv[1], sys.argv[2])  # python find_category.py US "accessories & supplies"

How do I call the bestsellers endpoint?

Send category_id, marketplaceId and optionally count (default 30, maximum 50) to /api/amazon-bestsellers. Each result carries its rank, asin, productTitle, price, reviewRating and reviewAmount. The script checks the ID against the marketplace's tree first, because the endpoint rejects an unknown ID only after the credit is charged. Marketplace IDs for all 23 storefronts are in the marketplace ID list.

bestsellers.py - validate the ID locally, then fetch the top 50

import os
from functools import lru_cache

import requests

API = "https://sellermagnet-api.com/api/amazon-bestsellers"
TREE_URL = "https://sellermagnet-api.com/static/data/{code}.json"


def all_ids(nodes):
    for node in nodes:
        yield str(node["CategoryId"]).lower()
        yield from all_ids(node.get("Subcategories") or [])


@lru_cache(maxsize=None)  # one download per marketplace, not per call
def known_ids(tree_code: str) -> frozenset:
    tree = requests.get(TREE_URL.format(code=tree_code), timeout=60).json()
    return frozenset(all_ids(tree))


def bestsellers(category_id: str, marketplace_id: str, tree_code: str, count: int = 50) -> list:
    if category_id.lower() not in known_ids(tree_code):
        raise ValueError(f"{category_id} is not a category in the {tree_code} tree")  # no credit spent
    resp = requests.get(API, params={
        "category_id": category_id, "marketplaceId": marketplace_id,
        "count": min(count, 50),  # above 50 is a billed 400
        "api_key": os.environ["SELLERMAGNET_API_KEY"],
    }, timeout=60)
    try:
        body = resp.json()
    except ValueError:
        body = {}  # some error pages are HTML
    if resp.status_code != 200 or not body.get("success"):
        raise RuntimeError(f"{resp.status_code}: {body.get('message')}")
    return body["data"]["bestsellers"]


for item in bestsellers("281407", "ATVPDKIKX0DER", "US")[:5]:
    print(item["rank"], item["asin"], item["price"]["price"], item["productTitle"][:60])

One entry of data.bestsellers (trimmed; example from an amazon.de list)

{
  "rank": 1,
  "asin": "B0B7SFSN99",
  "productTitle": "Sterntaler Schirmmütze Nacken - Unisex Baby- und Kinder Mütze ...",
  "price": {
    "price": 16.68,
    "currency_code": "EUR",
    "currency_symbol": "€"
  },
  "reviewRating": 4.6,
  "reviewAmount": 5152
}

Why is my category ID rejected?

The endpoint looks the ID up in the tree of the marketplace you asked for, at any depth, ignoring case. If it is not there, the answer is HTTP 400 with the message Invalid CategoryId - and because that check runs after billing, the request still costs a credit. The error-handling guide lists which errors are charged. The usual causes:

• The ID belongs to another marketplace. Numeric browse node IDs never carry over; look it up again in the target tree.
• A category name instead of an ID. Send 281407, not Accessories & Supplies.
• A URL fragment instead of the last segment. From /zgbs/electronics/281407, only 281407 is the ID.
• Ireland. amazon.ie runs on the amazon.co.uk catalogue, so its categories come from the UK tree.

How big is each marketplace's category tree?

Tree sizes vary widely, which matters if you cache or crawl them: the smallest trees have around 12,500 categories and the largest, amazon.de, more than 32,000. Measured on the public trees as of September 2026:

Bottom line: take the category ID from the Best Sellers URL or the category tree of the same marketplace, check it locally, and ask for up to 50 ranked products per call. The bestsellers endpoint page shows the full response, and a free account includes 500 credits.