7% of Amazon Pages Returned Errors With HTTP 200. Here's the Proof.

We set up daily screenshots of twelve Amazon product pages to track price changes. Three days in, half the captures weren't product listings. They were error pages. "Sorry, something went wrong on our end." Every uptime check said the pages were fine. HTTP 200 across the board. We didn't plan an availability experiment, but that's what we got.

What four weeks of screenshots revealed

We ran the test for 28 days, capturing each URL every six hours. Out of 1,338 usable screenshots, we found three distinct error types, all returning HTTP 200. Generic "something went wrong" templates made up 5.1% of captures (68 total), with visual diffs of 60–75% against normal product pages. CAPTCHA challenges showed up in 1.3% (17 total), where Amazon's bot detection served a verification page instead of the listing. Then there were the sneaky ones: partial renders at 0.9% (12 total). The page looked almost normal, but the price section was blank and the "Add to Cart" button was missing. Those only registered 15–20% visual change, easy to overlook if you aren't paying attention.

Add it up: 92.7% of captures were clean product pages. The other 7.3% were broken in some way. Our uptime monitor reported 100% availability for the entire period because it only checked HTTP status codes. The real user-visible availability was closer to 93%. But the errors weren't random. They clustered around late-night maintenance windows and high-traffic periods. That pattern only showed up because we had a timestamped visual archive to scroll back through. We actually went back and checked a Tuesday-night spike against Amazon's AWS status page, and sure enough, they'd had a partial regional outage that never made the news.

The failure modes we found aren't Amazon-specific. Any site that returns a friendly error page with a 200 status will fool traditional monitoring. Soft errors, silent API failures, partial renders: they're everywhere on the modern web. If you want to set up this kind of detection for your own pages, our availability monitoring guide walks through which URLs to watch, what thresholds to set, and how to route alerts so nothing slips through.