Audit Analytics Workspace

Phase 3.6 Benford screening (optional)

Off by default. Requires at least 100 eligible non-zero values. Screening results are recorded as audit context and do not automatically change exception risk scores.

Phase 3.6 Benford screening (optional)

Off by default. Requires at least 100 eligible non-zero values. Screening results are recorded as audit context and do not automatically change exception risk scores.

Phase 3.6 Benford screening (optional)

Off by default. Requires at least 100 eligible non-zero values. Screening results are recorded as audit context and do not automatically change exception risk scores.

Phase 3.6 Benford screening (optional)

Off by default. Requires at least 100 eligible non-zero values. Screening results are recorded as audit context and do not automatically change exception risk scores.

Phase 3.6 Benford screening (optional)

Off by default. Requires at least 100 eligible non-zero values. Screening results are recorded as audit context and do not automatically change exception risk scores.

Model profile

Risk rating thresholds

Scores remain capped at 100. Thresholds must satisfy Medium < High < Critical.

Audit-test weights

Weights are additive for tests triggered on the same source row and are capped at 100 before the configured rating thresholds are applied.

One-click Phase 3.1 validation

Tests threshold logic, 77 pack weights, score caps, model serialization, workbook persistence, snapshot immutability and custom-weight behavior without changing the active model.

One-click Phase 3.2 validation

Runs deterministic synthetic fixtures to validate robust Z, IQR, peer-group detection, ensemble scoring, missing-value handling, score caps, risk-rating integration, reproducibility and workbook QA persistence. No audit dataset is required.

One-click Phase 3.3 validation

Uses deterministic synthetic fixtures to validate seeded random, systematic, stratified, high-value and MUS/PPS selection, exact sample sizing, duplicate prevention, source-row traceability, monetary eligibility and QA persistence. No audit dataset is required.

One-click Phase 3.4 validation

Uses deterministic synthetic time-series fixtures to validate monthly/quarterly/weekly bucketing, aggregation, period spikes/drops, moving averages, persistent trends, seasonality, concentration shifts, score caps, risk-rating integration, source-row traceability and workbook QA persistence. No audit dataset is required.

Relationship mapping

Primary key is required. Secondary key is optional but must be mapped on both sources if used. Amount/date/attribute comparisons are applied only to unambiguous 1:1 matched relationships.

One-click Phase 3.5 validation

Uses deterministic synthetic paired populations to validate composite-key matching, missing counterparts in both directions, amount/date tolerances, attribute comparison, duplicate/ambiguous relationships, score caps, risk ratings, source-side traceability, reproducibility and workbook QA persistence. No audit dataset is required.

Approved Phase 3.6 scope

Included: Accounts Payable invoice amount, General Ledger absolute debit/credit line amount, Expense/T&E amount, Procurement PO amount, Revenue/AR invoice amount.

Excluded by design: Payroll and Inventory packs, because salary bands, policy constraints, standard costs, quantities and other structured values can make Benford assumptions inappropriate.

Pack screening is optional and off by default. When enabled, at least 100 eligible non-zero monetary values must remain after the minimum-amount filter.

One-click Phase 3.6 validation

Uses deterministic synthetic distributions to validate first-digit extraction, expected logarithmic proportions, MAD calculation/classification, minimum-value filtering, approved pack scope, overrepresented-digit context, score neutrality, snapshot reproducibility and workbook QA persistence. No audit dataset is required.

Phase 3.7 scoring methodology

Priority Score is capped at 100 and uses a transparent model: up to 55 points from base Risk Score, up to 21 from multiple tests/signals on the same run/source row, advanced analytics context (Cross-Domain / Anomaly / Trend), limited Benford run context, unresolved age, workflow status and unassigned ownership. Closed items are suppressed to zero.

One-click Phase 3.7 validation

Uses deterministic synthetic exceptions to validate every scoring component, band boundaries, multi-test grouping, closed-item suppression, run isolation, sorting, explanation quality, risk-score neutrality, reproducibility and workbook QA persistence. No audit dataset is required.

How orchestration works

Each procedure defines a primary domain pack, recommended supporting engines, expected evidence and scenario rules. The engine checks which steps have evidence in the current workbook, groups related exceptions by source lineage, identifies corroborating test combinations and produces one procedure-level scenario per evidence cluster. Underlying exceptions remain unchanged.

One-click Phase 3.8 validation

Uses deterministic synthetic audit evidence to validate all seven procedure definitions, evidence-lineage grouping, cross-run consolidation, procedure coverage, named scenario rules, scenario scoring, duplicate suppression, closed-item handling, risk neutrality, reproducibility and workbook QA persistence. No audit dataset is required.

Runtime environment

One-click Phase 4.1 validation

Checks Excel host/API availability, secure origin, workbook read/write isolation, local storage, frozen Phase 3 registrations, external runtime dependencies, production configuration safeguards and QA persistence. No audit dataset is required.

Current release

Update status

One-click Phase 4.2 validation

Checks release metadata, semantic-version comparison, secure same-origin update feed, manifest/taskpane/schema alignment, Excel API floor, frozen analytics registrations, update safeguards, local release-state storage, Desktop UI wrapping and QA persistence. No audit dataset is required.

Current installer/runtime state

One-click Phase 4.3 validation

Run this after installing the Phase 4.3 Desktop Installer package and loading the localhost add-in in Excel Desktop. The test validates localhost HTTPS, runtime health, installer state, version alignment, scheduled startup, Excel trusted-catalog registration metadata, workbook isolation, frozen analytics engines and QA persistence.

Bulk disposition

Applies to all currently filtered exceptions. Audit history is retained.

Start a new audit engagement

This clears CTC run history, exceptions and review history in this workbook, but keeps your saved settings and profiles.

One-click Phase 4.5A validation

Validates localhost hosting, local CTC assets, Office.js dependency classification, workbook-local analytics/evidence, licensing and update network boundaries, false-claim prevention, native migration metadata and QA persistence. This test does not pretend the existing Office.js host is fully air-gapped.