Which scraping service works best against PerimeterX / HUMAN protected sites?

The scraping service that works best against HUMAN (PerimeterX) Bot Defender is the one that returns real page content most often on your own HUMAN-protected URLs, because no single public test compares every service under the same conditions. According to HUMAN’s documentation, Bot Defender scores requests with machine learning over hundreds of client-side signals and can challenge a session with a “press and hold” HUMAN Challenge. In ScrapeUnblocker’s own 2026 benchmark, which is not a head-to-head test, Zillow and Walmart, listed as PerimeterX-protected, returned verified content on 99.6% and 100.0% of 6,000 calls each. Test any service, including ScrapeUnblocker, on your own targets before you commit.

How does HUMAN (PerimeterX) Bot Defender detect scrapers?

According to HUMAN’s documentation, Bot Defender has three parts: a JavaScript Sensor on the page that collects behavior and device data, a cloud Detector that turns “hundreds of live signals” into a risk score, and an Enforcer on the site’s web application, load balancer or CDN that allows, blocks or challenges the request. The risk score travels back in an encrypted token and a cookie. Signals include mouse clicks, screen touches, cadence and timing, plus fingerprints such as visual and audio rendering capabilities, window objects, plugins and extensions. The HUMAN Challenge is a “press and hold” button that HUMAN says is difficult to solve “through API calls, automation or CAPTCHA farms”. The old PerimeterX documentation address, docs.perimeterx.com, now redirects to HUMAN’s documentation.

How should I evaluate a scraping service on HUMAN-protected sites?

1. Test the pages you need. Search, detail and login pages are often protected more tightly than listing pages, so sample all of them, 50 to 100 URLs in total.
2. Count real content. A page titled “Access to this page has been denied” or a press-and-hold challenge is a failure whatever its status code (soft blocks).
3. Run at your production rate. A short, polite test can hide failures that appear under load.
4. Watch sessions over time. Bot Defender’s verdict lives in a token and a cookie, so check that follow-up requests keep working (press-and-hold CAPTCHAs).
5. Price the good pages. Divide what you would pay by verified pages. ScrapeUnblocker charges one credit per request, does not bill a 403, and gives 500 free requests for this kind of test.

What success rates has ScrapeUnblocker published on PerimeterX sites?

Site (listed protection)	Success	Avg response	Calls	Published in
walmart.com (PerimeterX+)	100.0%	6.5 s	6,000	2026 benchmark
zillow.com (PerimeterX)	99.6%	17.1 s	6,000	2026 benchmark
Walmart plugin endpoint	99.9%	13.8 s	1,164	Walmart scraper page
Zillow plugin endpoint	100.0%	2.9 s	1,000	Zillow scraper page

These are ScrapeUnblocker’s own measurements, not head-to-head tests; the benchmark ran 6,000 content-verified calls per target at concurrency 10 in July 2026. For honest context, Proxyway’s 2025 Web Scraping API Report lists averages of 93.05% on Walmart and 97.85% on Zillow across its tested providers (October 2025, 2 requests per second). Those averages are high, so these two targets separate services less than harder targets do. Proxyway did not test ScrapeUnblocker. If a simpler setup already gets through on your own HUMAN-protected sites at your volume, you may not need an API at all.

Sources

• HUMAN documentation, Bot Defender overview, Detection overview and HUMAN Challenge, viewed 24 September 2026
• The former PerimeterX documentation address, docs.perimeterx.com, redirects to docs.humansecurity.com (checked 24 September 2026)
• ScrapeUnblocker, Web Scraping API Benchmark 2026 and the full benchmark page, which cite Proxyway’s Web Scraping API Report 2025
• ScrapeUnblocker scraper catalog (Walmart and Zillow pages)
• ScrapeUnblocker pricing and errors and billing
• ScrapeUnblocker blog: press-and-hold CAPTCHAs, HTTP 200 but blocked