Track Amazon Bestseller-Oriented Search Results (2026)

Thirdwatch's Amazon Product Scraper can request Amazon's bestseller-oriented sort for category keywords, then return structured placement and demand-proxy fields across 19 marketplaces. It is designed for repeat search-shelf research—not official BSR extraction.

▶ Skip the setup: Run the bestseller-and-review movement Task on Apify →

What “bestseller tracking” means here

Set sortBy to bestSellers and provide category-specific keywords such as robot vacuum or mechanical keyboard. Amazon returns a bestseller-oriented search order. The Actor captures each result's ASIN, page, position, monthly-bought label, price, rating, review count, sponsored status, Prime flag when displayed, and badge.

This is not the same as Amazon's /Best-Sellers/zgbs hierarchy or the official Best Sellers Rank shown on product-detail pages. If your analysis requires the literal category leaderboard or numeric BSR, choose a source built for those pages.

Configure an incremental category monitor

{
  "queries": [
    "wireless headphones",
    "robot vacuum",
    "air fryer",
    "mechanical keyboard"
  ],
  "country": "us",
  "category": "all",
  "sortBy": "bestSellers",
  "maxResults": 200,
  "monitorMode": "all-changes",
  "monitorStoreName": "amazon-bestseller-keywords-us"
}

The first run emits a baseline. Later runs emit new ASINs and rows with available price, rating, or review-count changes. position is included on every returned row, but position-only changes do not currently trigger an all-changes event. If position movement itself is the primary signal, use monitorMode: "off" and retain full snapshots.

Pull a full weekly snapshot

import datetime
import json
import os
import pathlib
import requests

response = requests.post(
    "https://api.apify.com/v2/acts/thirdwatch~amazon-product-scraper/run-sync-get-dataset-items",
    params={"token": os.environ["APIFY_TOKEN"]},
    json={
        "queries": ["robot vacuum", "air fryer", "mechanical keyboard"],
        "country": "us",
        "sortBy": "bestSellers",
        "maxResults": 200,
        "monitorMode": "off",
    },
    timeout=3600,
)
response.raise_for_status()

week = datetime.date.today().strftime("%G-W%V")
pathlib.Path(f"amazon-search-{week}.json").write_text(
    json.dumps(response.json(), indent=2)
)

maxResults is the total run cap, not a per-query target. The Actor currently visits at most seven pages per query. Use focused runs when you need balanced category samples.

Calculate review and placement movement

import pandas as pd

previous = pd.read_json("amazon-search-2026-W28.json")
current = pd.read_json("amazon-search-2026-W29.json")

key = ["asin", "domain", "searchString"]
joined = current.merge(previous, on=key, suffixes=("", "_prev"))

joined["reviews_delta"] = (
    joined["reviews_count_value"] - joined["reviews_count_value_prev"]
)
joined["position_delta"] = joined["position"] - joined["position_prev"]

signals = joined[
    (joined["is_sponsored"].fillna(False) == False)
    & (joined["reviews_delta"] > 0)
    & (joined["rating_value"] >= 4.0)
].sort_values(["reviews_delta", "position_delta"], ascending=[False, True])

print(signals[[
    "title", "monthly_bought", "reviews_delta",
    "position_prev", "position", "price_value"
]].head(25))

Review growth is a lagging proxy for purchase activity, and the monthly-bought field is a bucketed lower bound. Combine them with placement, rating, and sponsored status instead of labeling any one threshold a guaranteed “rising bestseller.”

Sample output

{
  "title": "Cordless Robot Vacuum and Mop",
  "asin": "B0ABC12345",
  "price_value": 179.99,
  "rating_value": 4.4,
  "reviews_count_value": 8421,
  "monthly_bought": "1K+ bought in past month",
  "monthly_bought_value": 1000,
  "is_prime": true,
  "is_sponsored": false,
  "badge": "Amazon's Choice",
  "page": 1,
  "position": 6,
  "domain": "amazon.com",
  "searchString": "robot vacuum",
  "scraped_at": "2026-07-20T08:15:00+00:00"
}

Avoid false conclusions

1. position is captured search placement, not official BSR.
2. bestSellers is a search sort, not the literal Amazon Best Sellers page.
3. A newly visible product may reflect ranking churn, personalization, availability, or sponsored insertion—not necessarily a new launch.
4. monthly_bought_value is the lower edge of a public bucket such as 1K+, not exact sales.
5. Prime may be absent because Amazon omitted the label from that search card.
6. Compare the same query, country, sort, filters, and sample size every time.

Related Amazon workflows

• Scrape Amazon products for price monitoring
• Build an Amazon product-research tool
• Monitor Amazon competitor prices and ratings
• The complete guide to scraping e-commerce data

Run the Amazon Product Scraper on Apify or start with the preconfigured movement-tracking Task.