How decisions land
Every submission is routed through measurable signals. The dashboard shows which decisions are high-confidence, which need a second look, and how much traffic repeats.
99.68%
High-confidence calls
6,558 of 6,579 checks were classified with high confidence, giving teams a stable baseline of automatic decisions they can rely on.
0.32%
Low-confidence watch
Only 21 checks landed in the low-confidence bucket — a narrow surface for review queues and manual overrides instead of open-ended triage.
1.88%
Duplicate submissions
124 duplicate events were detected so campaigns, support queues, and content moderation can skip obvious repeated abuse immediately.
92.51%
Unique content seen
6,086 distinct content hashes across 6,579 checks — enough variety to prove the product handles real, diverse form traffic, not synthetic drills.