Yea the header trick and BeautifulSoup thing worked fine early in the year, but Amazon's leveled up hard - they've got datadome v3 now, which literally analyzes your HTTP/2 frames and can sniff out static headers and basic requests library stuff. So you're gonna need something like curl_cffi to mess with TLS fingerprinting or Playwright to act like a real human, and spoiler alert, this article doesn't touch any of that. And that whole just throw residential proxies and headers at it and you'll dodge 70-80% of captcha thing is stale.you're looking at more like 35-55% success these days, which means skipping over managed solutions like apify is kind of a mistake - especially if you're running 200k+ pages monthly, because apify'll actually save you money compared to building and maintaining your own janky infrastructure. Nobody here mentions behavioral biometrics - stuff like tracking mouse jitter, click timing, and scroll speed - which Amazon's now using to flag sketchy bot behavior no matter how good your headers look