Produced as you go, not assembled the month before

Accreditation against NCCA standards or ISO/IEC 17024 turns, in practice, on whether a body can produce records of decisions it made years ago: how the passing standard was set, whether the items work, who decided a contested result and on what basis, whether renewal requirements are actually verified. Bodies fail on documentation far more often than on practice.

A platform that treats those records as a reporting afterthought leaves a body reconstructing them under deadline, from email. CredHub writes them as the by-product of running the program, because the same records are what makes a decision defensible on the day it is made.

• How the cut score was set. The standard-setting study: method, date, facilitator, panel, the minimally-competent-candidate definition, the rounds, the recommendation and the rationale — with the superseded studies still there. Psychometrics.
• That the examination measures something. Difficulty, point-biserial and upper–lower discrimination, and distractor analysis per item, per sitting. Exam development.
• That eligibility is applied consistently. Modelled requirements, each application storing the rules as they stood that day, structured answers, evidence per requirement, and a reviewer decision with a reason. Exam administration.
• That candidates have due process. A concern raised in words, a window to answer, a decision with a reason, and an appeal heard by somebody who took no part. Item challenges with an outcome that propagates to every affected score. Exam integrity.
• That recertification is real. A versioned renewal policy, CE claims against approved providers, and a percentage of renewals audited against evidence. Credential management.
• Who did what. Every governed act — a result released, a credential revoked, a fee waived, a cut score adopted, a candidate seated past a full window — is written to an event record with the actor named.