SC-401 deep review — September 27, 2026

The complete guide was read and 65 published October objectives in nine groups were mapped to its sections. Microsoft's study guide still says October 28 while the credential page says October 14. This conflict remains unresolved; July 28 stays the accepted baseline and historical objective snapshots are preserved.

Learning improvements

Four worked examples explain classifier precision/recall, Power BI label versus content access, JIT initial protection versus failure fallback, and retention end dates. The guide also corrects Data Explorer's item/content visibility and explains EDM normalization, audit retention at ingestion, September network DLP integration limits and the August priority-cleanup permanent-deletion preview.

The ninth lab asks learners to predict and verify these boundaries, with explicit entries for capabilities they could not execute. There are now nine labs and 42 answered checks. Four synthetic arithmetic assertions passed locally. No tenant or infrastructure lab was executed; priority cleanup is a tabletop exercise only.

Blog decision

George Smyrlis's From Oversharing to Enforcement, published in the Microsoft Purview Blog on April 23, 2026, was accepted for its discovery, remediation, prevention and monitoring workflow. Its public embedded article was read. The guide adds attribution, a reading estimate and an original synthetic-site exercise. Broad automation/auditing claims and preview coverage are qualified against current product documentation. The tenant workflow was not executed.

Evidence and limits

ADLC_Docs/operations/2026-09-27-sc-401-deep-review.json records prior catalog entries, guide hashes, objective positions/hashes and section mappings, source observations, findings, blog intake and validation. The 65-objective count describes the published October revision; it does not replace the current July baseline.

One paid course page was access-blocked. Public practice and partner pages do not prove signed-in access, assessment completion, course content or video playback. Feature support, licenses, providers and effective identities still need validation in the learner's tenant. Research and repair used the same AI context; independent and human review remain pending. Freshness remains blocked by the official date conflict even though the content review is recorded as performed. Source/objective mapping checks passed; that result does not resolve lifecycle freshness. The guide and exam catalog both retain review-required status.

Publication uses unit-test, repository, strict site-build, generated-site and diff checks. The Microsoft review tracker retains the remaining queue and dated follow-ups.