Amazon Scraper API
Noah Bennett avatarNoah Bennett· ChocoData Amazon data expert · 12 min read
Amazon Scraping Blog

Best Proxies for Amazon Scraping in 2026 (Tested)

Updated at

The Answer

The best proxies for Amazon scraping in 2026 are country-matched residential or mobile proxies from a tier-one provider. Datacenter proxies fail on Amazon at 60 to 70% rates because Amazon’s WAF classifies them by ASN before the HTTP layer ever runs. Residential proxies from Bright Data, Oxylabs, Decodo, or Webshare succeed at 90 to 99% on product detail pages. Mobile proxies (CGNAT IPs from real phone carriers) succeed at 95%+ even on the harder seller-profile and back-end pages, at roughly 2x the price of residential. For most teams, country-matched residential at $4 to $12 per GB is the sweet spot. For teams that prefer not to manage proxies at all, a managed scraper API like Amazon Scraper API bundles the residential pool plus rotation plus retries for around $0.90 per 1,000 successful requests, with 1,000 free on signup.

Why Does Amazon Need Special Proxies?

Amazon needs special proxies because their WAF (Web Application Firewall) classifies incoming traffic by ASN (Autonomous System Number) in the first packet, before any HTTP-level signals are evaluated. The ASN identifies the network operator, and Amazon maintains internal lists of which ASNs are commercial cloud providers (DigitalOcean, AWS itself, Linode, Hetzner, OVH) versus residential ISPs (Comcast, Verizon, Deutsche Telekom).

A request from a commercial cloud ASN gets a different scoring profile than a request from a residential ISP. The commercial cloud profile triggers more aggressive bot detection: stricter rate limits, faster CAPTCHA escalation, and a lower threshold before a “Robot Check” page replaces the product HTML.

This is why a Python script running on a $5 DigitalOcean droplet can succeed for a few minutes and then suddenly start receiving robot checks for every request. The script did nothing wrong. The ASN got flagged.

The fix is to route requests through proxies whose IPs sit on residential or mobile ASNs that Amazon’s WAF classifies as low-risk by default.

What Are the Different Types of Proxies?

The four proxy types that matter for Amazon scraping, ranked by Amazon success rate:

  • Mobile proxies (CGNAT IPs from real cellular carriers) - Highest success rate (95%+) on every Amazon page type including seller profiles. Most expensive at $15 to $50 per GB.
  • Residential proxies (IPs from real home internet connections, sourced via opt-in SDKs in legitimate consumer apps) - High success rate (90 to 99%) on product detail pages, slightly lower on seller profiles. $4 to $12 per GB.
  • ISP proxies (datacenter IPs that the ISP has classified as residential at the ASN level) - Medium success rate (75 to 88%) on product detail pages. $1 to $4 per GB.
  • Datacenter proxies (commercial cloud IPs, no residential association) - Low success rate (30 to 70%) on Amazon. $0.50 to $2 per GB.

The price differential reflects the underlying cost of acquiring the IPs. A datacenter IP is a $5/month VPS. A residential IP is a fraction of bandwidth on a real home connection that someone agreed to share. A mobile IP is a fraction of a cellular plan.

The success rate differential reflects how Amazon’s WAF scores the ASN. Residential and mobile look like real users; datacenter looks like a bot.

Static vs Rotating Residential Proxies

Two configuration modes within residential and mobile categories:

  • Rotating residential assigns a fresh IP for each request (or per session, configurable). Best for high-volume scraping where being seen as the same user across many requests is a problem (Amazon flags repeated requests from one IP for the same product).
  • Static residential (sticky sessions) holds the same IP for a configurable duration (10 minutes to 24 hours). Best for paginated workflows where you need to maintain a session cookie across multiple requests, or for back-end pages that require a stable session.

For most Amazon product scraping, rotating wins. For paginated reviews (when accessed pre-login-wall) or any flow that uses Amazon’s session cookies, sticky wins.

Top 7 Proxy Providers for Amazon Scraping

The seven providers that consistently rank in independent Amazon-scraping benchmarks, with realistic 2026 pricing.

1. Bright Data

Bright Data operates the largest residential proxy network globally with 150+ million IPs across 195 countries. Their network depth means even when Amazon flags entire sub-pools, fresh ones rotate in fast.

Pricing on residential is roughly $8.40 per GB at the entry tier, dropping to $4 per GB at high commitment. Mobile proxies sit at $24 per GB. They charge separately for unblocking infrastructure (SERP API, Web Unlocker) which adds another layer for scraping protected pages.

Best for: Enterprise teams scraping at multi-TB scale who can negotiate Custom plans. Compliance-conscious industries that need formal data-access agreements.

Trade-offs: Highest price tier of the top providers. Pricing is opaque without a sales call. The unbundled product structure (proxies, Web Unlocker, SERP API sold separately) makes total cost prediction difficult.

2. Oxylabs

Oxylabs is a Lithuanian-headquartered provider with a 100+ million IP residential pool and one of the largest mobile pools in Europe. Their proprietary OxyMouse module emulates real browser interaction patterns that bypass several Amazon anti-bot signals beyond the IP layer.

Residential pricing starts at $8 per GB and drops to $4 per GB at commitment. Mobile sits at around $22 per GB.

Best for: Teams scraping across many e-commerce targets beyond Amazon. EU-based teams who want a vendor with EU-resident infrastructure for GDPR purposes.

Trade-offs: Premium price point. The two-product split (proxies separate from Web Scraper API) means most production workflows pay for both.

3. Decodo (formerly Smartproxy)

Decodo (rebranded from Smartproxy in 2025) runs a 65+ million IP residential pool and a strong mobile tier. Pricing is more aggressive than Bright Data or Oxylabs at the entry level.

Residential is $7 per GB at the entry tier, dropping to $3 per GB at commitment. Mobile is around $14 per GB. They run aggressive 60-day cookie windows on their affiliate program, which has driven a lot of independent benchmark coverage in their direction.

Best for: Mid-market teams scraping 100GB to 5TB per month. Teams that want both proxies and a Web Scraping API from one vendor.

Trade-offs: Smaller pool than Bright Data, so coverage gaps can show on niche country/city targets. Rebrand-related documentation churn in 2025.

4. Webshare

Webshare is a budget-friendly proxy provider with a 30+ million IP residential pool. Pricing is the most aggressive among the top tier: roughly $2.99 per GB on residential, dropping to under $1 per GB at high commit.

The trade-off is that the pool is smaller and the IP rotation logic is simpler than Bright Data or Oxylabs. For Amazon product detail page scraping at modest volumes (under 500GB/month) the success rates are competitive. For aggressive seller-profile scraping or for Amazon during Prime Day-level traffic spikes, the smaller pool shows.

Best for: Cost-sensitive teams scraping under 500GB per month. Solo developers and small consultancies. Anyone whose pricing math fails above $4 per GB.

Trade-offs: Lower success rates on harder Amazon endpoints. Slower customer support response times. Less third-party benchmark coverage.

5. Evomi

Evomi is a Swiss-based newer entrant with a focus on ethically-sourced residential IPs (the SDKs that fund the network are explicit about the data-sharing trade with users). Pricing sits between Webshare and Decodo at roughly $3 per GB on residential.

Their pool is smaller than the top three (around 12 million IPs), which limits scale, but the IP quality is high enough that success rates on Amazon match providers twice the price.

Best for: Teams that prioritize ethical sourcing and EU-resident infrastructure. Pre-Series-A startups balancing cost and performance.

Trade-offs: Smaller pool. No proprietary anti-bot stack on top of the proxy layer (unlike Bright Data’s Web Unlocker or Oxylabs’ OxyMouse).

6. NetNut

NetNut is an Israeli provider with a 52+ million IP residential pool sourced through ISP partnerships rather than consumer SDKs. The ISP-partnership model gives them stable IPs that survive longer between rotations than the SDK-sourced pools.

Pricing is around $9.50 per GB on residential at entry tier. Their static residential offering (“ISP proxies”) is competitive at $1.50 per GB.

Best for: Teams that need stable IPs (sticky sessions for paginated workflows). Workflows that can use ISP-tier proxies instead of full residential.

Trade-offs: Mid-market price. Smaller mobile pool than Bright Data, Oxylabs, or Decodo.

7. IPRoyal

IPRoyal is a Lithuanian provider with a competitive entry-tier residential price (around $7 per GB on Pay-As-You-Go). Their differentiator is per-port residential proxy plans that suit smaller teams who do not want to deal with bandwidth-based billing.

Best for: Solo scrapers and small teams who prefer fixed monthly proxy bills.

Trade-offs: Smaller pool. Less proprietary anti-bot infrastructure than the top three.

How Do You Choose Between Residential, ISP, and Mobile Proxies?

The decision tree for Amazon scraping specifically:

For product detail page scraping at any volume: rotating residential. Success rate matches mobile at half the price.

For paginated workflows that need session continuity: static residential or ISP proxies with sticky sessions. The fixed-IP behavior preserves session cookies that Amazon uses for cross-request state.

For seller profiles, back-end search filters, and any page Amazon protects more aggressively: mobile proxies. The CGNAT IPs from real cellular carriers carry the lowest detection signal.

For under 5,000 ASINs per month: Webshare or IPRoyal at the entry tier. The price differential matters more than the pool depth at this volume.

Above 100,000 ASINs per month: Bright Data, Oxylabs, or Decodo at a Custom plan. The pool depth matters more than the entry price at this volume because anti-bot escalations during peak periods hit smaller pools harder.

What Is the Cost of Proxies for Amazon Scraping?

Residential proxy costs scale linearly with bandwidth. A typical Amazon product detail page is 200KB to 500KB depending on category. Search result pages are larger (1MB+). For planning, assume 250KB per scrape average across a mixed workload.

That math gives:

  • 10,000 ASINs at 250KB = 2.5GB. At $7 per GB residential = $17.50 per 10k requests.
  • 100,000 ASINs at 250KB = 25GB. At $5 per GB (volume tier) = $125 per 100k requests.
  • 1,000,000 ASINs at 250KB = 250GB. At $3 per GB (high-commit tier) = $750 per 1M requests.

Compare to managed scraper APIs at $0.50 to $0.90 per 1,000 requests:

  • 10,000 ASINs at $0.90/1k = $9.
  • 100,000 ASINs at $0.90/1k = $90.
  • 1,000,000 ASINs at $0.50/1k (Custom) = $500.

The managed API path is cheaper above roughly 50,000 requests per month, equivalent at smaller volumes, and cheaper still when you account for engineering time spent maintaining the DIY scraper. Amazon Scraper API bundles the proxy, retry, and orchestration layers; pricing starts at $0.90 per 1,000 successful requests with 1,000 free on signup.

How Do You Integrate Residential Proxies in Python?

The pattern most Python Amazon scrapers use combines curl_cffi (for browser TLS fingerprint impersonation) with an HTTP-level proxy URL.

from curl_cffi import requests

PROXY_URL = "http://username:[email protected]:8080"

response = requests.get(
    "https://www.amazon.com/dp/B08N5KWBKK",
    impersonate="chrome",
    proxies={"http": PROXY_URL, "https": PROXY_URL},
    timeout=30,
)

Three things in this snippet matter:

  • impersonate="chrome" makes curl_cffi replay a real Chrome TLS handshake. Without it, Python’s default requests library sends a TLS fingerprint that Amazon flags in the first packet regardless of the proxy.
  • The proxies dict tells curl to route through the residential proxy. The provider issues credentials in username:password format.
  • The 30-second timeout matches Amazon’s typical p95 response under load.

For country-matched routing (German IP for amazon.de), most providers support country selection via the username field: username-cc-de:[email protected]:8080. The exact syntax varies per provider; Bright Data uses customer-zone1-country-de:password, Webshare uses username-de:password.

How Do You Rotate Residential Proxy IPs?

Two rotation strategies:

  • Per-request rotation (rotating residential): the provider returns a fresh IP for each new request. Default behavior on most rotating endpoints. Right for high-volume product scraping.
  • Per-session rotation (sticky residential): the provider holds the same IP for N minutes (configurable, typically 10 to 30 minutes). Right for paginated workflows where session cookies need to persist.

In Python, both are configured at the proxy URL level. Bright Data uses session-NNN in the username for sticky behavior. Webshare uses a session ID parameter. The provider documentation is the source of truth.

For Amazon specifically, per-request rotation handles 80% of workloads. Switch to sticky only when a workflow demands it.

What Are Best Practices for Amazon Proxy Scraping?

Five operational practices that materially affect success rates:

  • Match country to TLD. Scrape amazon.de through a German IP, amazon.co.jp through a Japanese IP. Cross-country requests trigger anti-bot in milliseconds even on premium pools.
  • Cap requests per IP. No more than 3 to 5 requests per minute from any single residential IP to amazon.com. Even on rotating endpoints, ensure the rotation logic enforces this cap.
  • Rotate User-Agents in lockstep with IPs. A single User-Agent across thousands of IP rotations is its own fingerprint.
  • Respect rate limits with exponential backoff. When Amazon serves a 429 or a robot check, wait 5 to 10 seconds before retrying with a fresh IP. Hammering immediately makes the block stick longer.
  • Cache aggressively. A product page changes maybe once per day for most categories. Caching responses for 30 to 60 minutes cuts your bandwidth, your cost, and your detection risk by the same factor.

FAQ

What’s the cheapest proxy provider for Amazon scraping?

Webshare at $2.99 per GB is the cheapest tier-one provider. IPRoyal sits close at around $3 per GB for residential. Below that, you are buying smaller pools and lower success rates, which makes the per-request math worse despite the lower per-GB price.

Do free proxies work for Amazon scraping?

No. Free proxy lists are dominated by IPs that are already on Amazon’s permanent blocklist (often because they were used for previous abuse). Success rates on free proxies against Amazon are typically below 5%. The bandwidth saved is more than offset by the time wasted on retries.

Can I scrape Amazon without proxies?

Yes, but only at very small volumes (under 100 requests per day from one home IP) and only for product detail pages. Anything above that triggers Amazon’s per-IP rate limit, and your home IP gets flagged for hours or days. For any production workflow, residential proxies or a managed scraper API are the only realistic options.

How do I avoid getting blocked when scraping Amazon?

Combine four things: country-matched residential proxies, browser-impersonation TLS (via curl_cffi), conservative request rates (3 to 5 per minute per IP), and aggressive response caching. Those four together yield 90%+ success rates against Amazon in most categories. See our bypass Amazon CAPTCHA guide for the deeper anti-bot patterns.

Should I use mobile proxies or residential for Amazon?

Residential is the default for cost reasons. Mobile is the upgrade for harder targets: seller profiles, back-end search filters, and any page Amazon serves more aggressive anti-bot on. Mobile costs roughly 2x residential at most providers, which is why most teams reserve them for the 5 to 10% of their Amazon workload that needs them.

What’s the difference between scraping with proxies vs using an Amazon Scraper API?

A scraper API bundles the proxy layer, the TLS-fingerprint layer, the retry-orchestration layer, and the response-parsing layer into one HTTP endpoint. You pay per successful request rather than per GB of bandwidth, and you do not maintain the infrastructure when Amazon rotates its anti-bot stack. For workloads above ~50,000 requests per month, a managed API like Amazon Scraper API is cheaper than residential proxies plus engineering time. For smaller, full-control workloads, raw residential proxies are still the right call.

Sources