Every Secho scan produces a score from 0–100 and a letter grade. Here's exactly how each scan type calculates its result.
All scan types use a 0–100 point scale with letter grades. The model differs slightly per scan type based on what is being measured.
Each vendor is scored across 7 categories. The overall score is a weighted average of those categories, then prohibited vendor penalties are applied on top.
Findings-based model. Score starts at 100 and each finding deducts points based on severity. Total penalty is proportionally capped to the number of checks run.
Same findings-based model as cloud audits. Org-level findings (e.g. 2FA not enforced org-wide) carry more weight than per-repo findings since they affect the entire organization.
Identical findings-based model to cloud audits. AI-specific checks (Vertex AI exposure, training data access, service account hygiene) contribute findings with the same severity weights.
File-based scoring. Score reflects the proportion of files that are clean vs. flagged, weighted by finding severity across all scanned documents.
Scores map to letter grades consistently across all scan types. The portal shows both a raw score and an adjusted score.
How edge cases are handled across all scan types.
Score is 100/100, grade A+. This applies to all scan types — a completely clean document audit, a GCP project with zero findings, or a domain with no prohibited vendors all score A+.
No. The score is always clamped to a minimum of 0. Even if the penalty calculation exceeds 100 points, the score floors at 0 (grade F).
Accepting a finding in the portal removes it from the penalty calculation. The adjusted score and grade update immediately. For TPRM/cloud scans with prohibited vendor caps, accepting the vendor finding also lifts the cap, potentially raising the score significantly.
The max_penalty denominator scales with the scope of the scan — a GCP project with 200 checks run has a higher denominator than one with 50, so the same number of CRITICAL findings has a proportionally smaller impact on a larger scan. This prevents small scans from being unfairly penalized for having fewer possible checks.
No. Benchmark mappings (CIS, FedRAMP, NIST 800-53, NIST AI RMF, SOC 2) are informational only. They appear in the portal as a separate tab to help with compliance reporting but have no effect on the score or grade.
Yes — deep mode can find additional findings (or confirm false positives), so the score may differ. Deep mode AI analysis findings are counted the same way as pattern-matched findings. INFO-severity AI summary notes are never counted toward the score.