Command Center
Real-time shrink intelligence across the store network, powered by POS-exception and computer-vision AI.
Shrink Rate Trend
% of sales · trailing actual vs. AI-forecasted trajectory
AI Insight Feed
Auto-generated from live network data
Cause of Loss
Share of total shrink value, network-wide
Cause mix calibrated to NRF / Appriss National Retail Security benchmarks
Shrink Rate by Category
vs. industry benchmark (dashed)
Shrink Risk Heatmap
Category × Region · risk-weighted score
Priority Cases
Highest severity, most recent first
POS Exception Intelligence
Behavioural analytics on voids, no-sales, refunds and discount overrides — cross-referenced against employee, shift and register to surface sweethearting and register manipulation.
Exception Types
Share of flagged transactions, trailing 30 days
Exception Volume by Hour
Network-wide, last 7 days
Employee Risk Radar
Exception rate vs. transaction volume · bubble = $ exposure
Top Flagged Employees
Ranked by composite risk score
Register Exception Log
Live transaction-level exceptions
| Exception ID ↕ | Type ↕ | Employee ↕ | Store ↕ | Amount ↕ | Time ↕ | Risk ↕ |
|---|
Payment & Card-Not-Present Fraud
Grounded in the real ULB/Worldline European card-transaction dataset (284,807 transactions, 492 confirmed frauds) — the industry-standard public benchmark for payment fraud analytics, rescaled to this network's transaction volume.
Fraud Rate by Transaction Amount Band
Real distribution from the ULB Credit Card Fraud Detection dataset
Source: Machine Learning Group – ULB, "Credit Card Fraud Detection" (Kaggle / Worldline, Sept 2013 European cardholders)
Avg. Transaction Amount — Fraud vs. Legitimate
Real values from the reference dataset
Self-Checkout vs. Staffed Lane Shrink
Industry-benchmarked differential
Fraud Score Distribution
AI model confidence across scored transactions today
Sample Transactions
Real anonymised rows from the reference dataset (PCA components V1–V4 shown)
| Time (s) | V1 | V2 | V3 | V4 | Amount | Class |
|---|
AI-Prioritised Investigation Cases
Every flagged incident — theft, fraud or error — ranked by expected recoverable value and evidence confidence. Approve to open a formal case file.
Case Queue
Sorted by AI priority score
| Case ID | Type | Store | Est. Loss | Confidence | Category | Opened | Priority | Action |
|---|
Case Resolution Funnel
Trailing 30 days, network-wide
Shrink Reduction ROI Calculator
Model the impact of a loss-prevention intervention on a store
Current Shrink Rate
–
–
Projected Shrink Rate
–
–
Annual Value Recovered
–
–
Payback Period
–
–
Store & Category Risk
Format-level and category-level shrink benchmarking with ABC-style prioritisation for audit and camera-coverage investment.
Risk × Value Matrix
Stores placed by shrink risk (X) and revenue value (Y) · click a cell to filter
Most-Targeted Categories
Ranked by shrink value at risk, network-wide
Store Format Comparison
Multi-metric normalised view, top formats
Shrink Risk — Category × Region
Darker = higher risk-weighted shrink exposure
Store Directory
All stores across the filtered network
| Store ↕ | Name ↕ | Format ↕ | Region ↕ | Revenue ↕ | Shrink % ↕ | Loss Value ↕ | Risk ↕ |
|---|
Alerts & Anomaly Detection
Statistical + ML anomaly detection surfaces theft signals, fraud spikes, expiry-linked write-offs and register deviations as they happen.