Start with the rules and the official options
Amazon is one of the most frequently scraped sites in the world, and one of the most defended. Before you build anything, two points:
- Amazon's conditions of use restrict automated data collection. Collecting data at scale may breach those terms, and Amazon actively blocks automated traffic.
- Official alternatives exist. Sellers and vendors have API access to data about their own business, affiliates have programme APIs with product information, and many commercial data providers sell Amazon data under their own agreements.
For many teams, an official API or a data provider is cheaper and lower-risk than collecting pages yourself. If you've considered those and still need to collect publicly visible pages, this guide covers how Geonode proxies fit, and how to do it carefully. See also Amazon scraping proxies.
Which Geonode product to use
| Need | Geonode product | Notes |
|---|---|---|
| Product pages at moderate volume | Rotating residential | Datacenter IPs are challenged quickly on Amazon |
| A small number of products, no maintenance | Scraper API | Pay per page; the API handles fetching |
| Different marketplaces (US, UK, DE, JP) | Residential with country targeting | Match exits to the marketplace |
| Logged-in seller workflows on your own account | Not proxies | Use the official seller APIs |
Datacenter proxies rarely work for Amazon at any meaningful rate. Rotating residential is the realistic DIY option.
Targeting the right marketplace
Each Amazon marketplace has its own domain, catalogue, prices and delivery logic. Match the Geonode exit country to the marketplace: US exits for the US store, German exits for the German store, and so on. Some product details, delivery estimates and offers also depend on the delivery location set in the session. If you need consistent location-dependent data, set the delivery location once per sticky session and fetch a batch of products in it.
Session settings
- Rotate per request for independent product pages when delivery location doesn't matter.
- Use short sticky sessions (a few minutes) when you set a delivery location first, or when paging through search results.
- Keep rates low per IP and per session. Amazon is sensitive to bursts.
Proxy tip: You can try every step in this guide with a small balance: Geonode bills residential bandwidth per GB with no expiry, so testing costs very little. See our complete Geonode guide for the full product overview.
Page weight and bandwidth
Amazon product pages are heavy when loaded in a browser, with large images and many scripts. On per-GB residential traffic, that adds up quickly.
- Prefer an HTTP client for server-rendered product data.
- If you use a browser, block images, media and fonts, and avoid loading review pages you don't need.
- Fetch only the fields you need and only as often as they change.
Because Geonode's residential bandwidth doesn't expire, small or occasional Amazon jobs don't waste a monthly allowance. See cheap proxies.
Detecting blocks
Amazon commonly responds to suspected automation with CAPTCHA pages or simplified error pages that still return HTTP 200. Detect them by checking for an element present only on real product pages, such as the product title container, and treat anything else as a failure. Retry with a new IP after a pause, cap attempts, and slow down globally if the block rate rises. See CAPTCHAs and proxies.
When the Scraper API makes more sense
If you need data for a few hundred products and don't want to maintain a scraper against Amazon's changing defences, sending URLs to Geonode's Scraper API and receiving HTML or Markdown is often simpler. Compare its per-request cost with the residential bandwidth and engineering time a DIY scraper would need. See the Geonode Scraper API.
A careful approach
- Limit collection to public product information you genuinely need, such as prices and availability.
- Avoid personal data, including reviewer profiles.
- Don't log in to accounts for scraping.
- Keep volumes and rates modest.
- Document what you collect and why.
- Reconsider official APIs and data providers as your needs grow.
See ethical web scraping and are proxies legal?.
Common problems
- Constant CAPTCHAs: rate too high, or the client's fingerprint looks automated. Slow down and use a realistic client.
- Prices differ from what you see manually: delivery location or marketplace mismatch. Match both.
- Pages missing fields: Amazon serves different layouts in tests and by category. Make parsers tolerant and log unparsed pages for review.
- Bandwidth higher than expected: images and scripts loading in a browser. Block them or switch to an HTTP client.
FAQ
Can I scrape Amazon with datacenter proxies?
Rarely at any useful scale. Amazon identifies hosting ranges quickly. Residential proxies or an API are the realistic options.
Is scraping Amazon legal?
Collecting public data isn't automatically illegal, but it can breach Amazon's terms, and the legal position depends on what you collect, how and where. Get advice for commercial projects. See are proxies legal?.
How do I get prices for different Amazon marketplaces?
Use residential exits in the marketplace's country and the marketplace's own domain, and set a delivery location in a sticky session where it affects results.
Is the Scraper API cheaper than proxies for Amazon?
For small volumes, often yes once you include engineering time. For large volumes, a well-tuned DIY setup is usually cheaper per page.
Editorial note: Proxy providers change their pools, features and prices often, so treat any figures here as typical ranges at the time of writing and confirm current terms with the provider before you buy.