Scorecard Percentile / Benchmarking
“How does my whole agent stack compare — dimension by dimension?”
The fleet benchmark ranks ONE scalar readiness score. AgentSmack's whole-stack percentile lab answers the question a CISO asks the instant they see their unified posture number: how does my stack compare to everyone else's, per dimension? Feed a target production scorecard and a cohort of peer scorecards and it ranks the overall posture AND every present dimension against the cohort via the SAME percentile engine the fleet benchmark uses — so the report says “your tenant_isolation is bottom-quartile but your credential_safety is top-decile”, and a buyer's posture number becomes competitively contextualized and prioritized. Honest-empty: a cohort smaller than five peers, or a dimension nobody in the cohort covers, yields an honest no-comparison — never a fabricated percentile. It is a benchmarking META-LENS over already-computed scorecards (no new surface, no new dimension, no DB). Fill a sample cohort below to watch the standing swing with no live infra.
Insufficient cohort — 0 peer scorecard(s) supplied, fewer than the 5 needed to rank honestly. No percentile is fabricated from too small a cohort.
Honest-empty — the cohort of 0 is too small to rank against (a statistically-meaningful comparison needs at least 5 peers). No percentile is fabricated from too small a cohort. Add more peer scorecards to benchmark dimension by dimension.