Correction, 7 September 2026: the earlier numerical benchmark scores on this page have been withdrawn because the underlying test records were not available. This page now describes an evaluation you can reproduce. WebsiteDown.org is our own product.
Choose the question before the checker
A one-off origin check, a provider’s incident page, a user-report service and a scheduled browser monitor answer different questions. A useful comparison starts with what you need to establish: whether a host responds now, whether users report problems, or whether a particular transaction keeps working.
A controlled evaluation
Use a public test site you control. Prepare a working root page, a failing application path and a path that redirects to the working page. Record the expected response before testing tools. Do not create failures on production services for this exercise.
- Enter the working origin and save the returned time, status and location coverage.
- Enter the failing path. Does the tool actually request it, or silently replace it with the origin?
- Check the redirect. Does the result distinguish the requested address from the final address?
- Repeat once after the stated cache window. Compare observation timestamps.
- Export the result. Can a colleague recover the exact observation without starting a new check?
What WebsiteDown can demonstrate
Our recorded example.com check contains eight HTTP 200 results and their original timestamps. The raw file lets you examine the output format. It does not establish comparative accuracy, incident detection speed or an uptime percentage.
WebsiteDown checks the origin and follows redirects. It does not test the application path in the second step or run a signed-in browser journey. Its eight configured locations belong to one hosting provider. If those limitations matter to your incident, add a different measurement method.
A comparison sheet without invented scores
Tool and version / URL:
Test time in UTC:
Input URL and actual target:
Expected response and returned response:
Available locations and provider diversity:
Original observation timestamps:
Redirect information:
Exported evidence location:
Failure the tool did not detect:Compare tools using the same inputs and close observation times. Keep unavailable results distinct from target failures. Select the tool that supplies the evidence your workflow needs; a single average score can hide a crucial missing capability. See the incident checklist for interpreting the result.