A source that tries several accounts differs from one user mistyping a password. Measure distinct identities per source rather than raw records. Missing users cannot establish additional named targets. A spray hypothesis still needs context such as known scanners, identity provider policy and the observation window. Do not label every shared address as one person.
Work from the raw records before you read the alert label. Keep the source, time boundary and outcome visible in your result. Use the reference panel for the generic search syntax, then test the search against the task. A plausible story can still be wrong: cite the events that support your conclusion and name anything the logs cannot establish. The changed-log check rejects searches tied only to memorized event IDs.
A source that tries several accounts differs from one user mistyping a password. Measure distinct identities per source rather than raw records. Missing users cannot establish additional named targets. A spray hypothesis still needs context such as known scanners, identity provider policy and the observation window. Do not label every shared address as one person.