Why do different shadowban checkers give different results?

One tool may only test from:handle search; another only tests autocomplete; a third looks at reply ranking and may never expand Show more replies. A cached result, a failed upstream request, or a logged-in versus logged-out session changes the answer too.

X Shadow Ban reports untestable instead of guessing when public data is insufficient, for example an account with no recent replies. Reply ranking is explicitly untestable unless a provider exposes it. A score of passed / measured x 100 is not comparable to a tool that counts untestable as a pass.