Software Confidence

Software confidence in the Censys Platform measures how reliably Censys identifies software and versions on Internet-exposed assets. Confidence scores range from 0 to 1 and help users assess the likelihood of false positives in detected software fingerprints.

Software confidence values are displayed in the following fields:

• host.services.software.confidence
• web.software.confidence

Confidence levels

Confidence values can be one of the following four levels.

• Confidence Level 1.0 (Very high)
  • Based on exact matches of known, structured fields (e.g., RFC-defined values, protocol specifications).
  • Examples: Exact favicon hashes, HTTP body hashes, or banner content hashes.
• Confidence Level 0.75 (High)
  • Uses regex-based matching on multiple unique fields or attributes.
  • Examples: Non-HTTP regex-based recog fingerprints.
• Confidence Level 0.5 (Moderate)
  • Based on heuristic approaches, such as non-RFC fields and inferred metadata.
  • Examples: HTTP endpoint regex matches or body regex matches.
• Confidence Level 0.25 (Low)
  • Includes broad or uncertain matching criteria, potentially capturing false positives.
  • Examples: Single-field keyword-based recognition.

Notes

• Confidence levels do not currently impact CVE (vulnerability) scoring.
• Software confidence fields are available exclusively to Enterprise-tier users.