Identity Normalization Health

This is the help text the console shows in the help panel on this page.

This report shows the quality of AutoCISO’s identity matching process. Normalization is the act of linking raw account names (as they appear in each system) to verified employee identities. Low confidence matches may result in misattributed access or missed anomalies.

Confidence Tiers

• Auto-verified (≥80%): High-confidence matches.
• Manual Required: Every review task below 80% confidence — these need a human decision in the Review Queue.
• Low Confidence (<60%): Likely mismatches, a subset of Manual Required — review these first.

Confidence Distribution

The bars show how many review tasks fall into each 20-point confidence band (0–20% up to 80–100%), so you can judge overall data quality at a glance.

Low-Confidence Table

Lists pending review tasks below 60% confidence, lowest confidence first — the matches that still need a decision. Turn on Show history to include tasks that were already approved, rejected or remapped. The stat cards and the distribution chart always cover all review tasks, and PDF always exports the pending list. Columns: Raw Name and Raw Email (as seen in the source system), Asset, Confidence, Status and Risk. Export with CSV or PDF.

Why This Matters

Poor normalization health means your anomaly detection and access reviews are working with incomplete data. Improving match quality directly improves the accuracy of every other report in AutoCISO.