DEV Community

g4dr
g4dr

Posted on

Build an Amazon Affiliate Feed with Live Prices in 2026

#amazon #affiliate #webscraping #api

Static prices kill affiliate conversions. A "best standing desks" table that says $299 while Amazon shows $349? Your reader bounces. Live data converts β€” here's how to build a self-updating product feed with Apify actors.

πŸ€– Try it live β€” chat with a Poe bot that generates the exact pipeline for your niche: Amazon Product Miner Β· Lead List Generator Β· Job Postings Miner Β· SERP Rank Tracker Β· Social Lead Finder

The stack

  • Apify β€” apify/amazon-crawler handles Amazon's anti-bot wall. Free credits to start.
  • Nightly schedule β€” fresh prices every morning, zero maintenance.

Step 1 β€” Token

Create an Apify account β†’ Settings β†’ Integrations β†’ API token β†’ export APIFY_TOKEN=...

Step 2 β€” Scrape ASINs in batch 

from apify_client import ApifyClient

client = ApifyClient("APIFY_TOKEN")
asins = ["B0ABC12345", "B0DEF67890", "B0GHI11223"]

run = client.actor("apify/amazon-crawler").call(run_input={
    "productUrls": [{"url": f"https://www.amazon.com/dp/{a}"} for a in asins],
    "proxyCountry": "AUTO_SELECT_PROXY_COUNTRY",
})

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["asin"], item["title"], item.get("price"), item.get("inStock"))
Enter fullscreen mode Exit fullscreen mode

Fields: asin, title, price, listPrice, stars, reviewsCount, inStock, thumbnailImage, url.

Step 3 β€” Scheduled snapshots β†’ live table 

import json, os

DB = "prices.json"
current = {}
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    current[item["asin"]] = {
        "price": item.get("price"),
        "stock": item.get("inStock"),
        "stars": item.get("stars"),
    }

old = json.load(open(DB)) if os.path.exists(DB) else {}
for asin, p in current.items():
    if asin in old and p["price"] and old[asin]["price"]:
        drop = (old[asin]["price"] - p["price"]) / old[asin]["price"]
        if drop > 0.10:
            print(f"PRICE DROP {asin}: {old[asin]['price']} -> {p['price']}")

json.dump(current, open(DB, "w"))
Enter fullscreen mode Exit fullscreen mode

Put the actor on a daily schedule in Apify Console β†’ this script runs via cron/Actions β†’ your comparison pages always show today's price.

Step 4 β€” Reviews for content angles

junglee/amazon-reviews-scraper with filterByRating: "critical" pulls the worst reviews first β€” exactly what you need for "pros & cons" sections that readers trust:

curl -X POST "https://api.apify.com/v2/acts/junglee~amazon-reviews-scraper/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"productUrls": [{"url": "https://www.amazon.com/dp/B0EXAMPLE"}], "maxReviews": 100, "filterByRating": "critical"}'
Enter fullscreen mode Exit fullscreen mode

Pro tips

  • proxyCountry must match the marketplace (US β†’ .com, DE β†’ .de) or data comes back wrong/blocked.
  • Batch ASINs in one run; re-run only failures.
  • Pair prices with your Associates tag in the outbound URL β€” ?tag=yourtag-20.

More tools on awesome-web-scraping. Agent users: npx skills add g4dr/apify-skills β€” the apify-amazon-products skill does this end-to-end (repo).

What niche are you building a feed for?

Top comments (1)

Collapse
 
sinarezaei profile image
Sina Rezaei •

The part I’d be careful with is treating the latest scrape as the current truth. Imagine a comparison page tracking 5,000 ASINs across multiple Amazon marketplaces. A crawler can finish successfully while returning stale prices, missing offers, or incomplete stock data for a subset of products. If that snapshot replaces the previous dataset wholesale, a temporary scraping failure can look like thousands of products suddenly disappearing or changing price.

I’d keep the last-known-good snapshot, validate each batch before publishing it, and track price_observed_at, marketplace, and scrape status per ASIN. Price-drop alerts should also ignore missing or invalid values instead of treating them as real changes. For production, I’d evaluate Amazon’s Creators API before building around scraping, especially since PA-API 5.0 was deprecated in May 2026. The API’s rate limits and eligibility requirements still need to be part of the design, though. A feed that knows when its data is unreliable is much more useful than one that simply refreshes every night.