ML-powered compliance intelligence across 2.1 million taxpayers.
NEXUS is Ghana's end-to-end tax administration and revenue intelligence platform. It unifies taxpayer data from nine source systems, automates compliance checks at scale, and runs a continuously learning ML anomaly detection engine that identifies high-risk filing patterns — enabling the GRA to close compliance gaps and dramatically increase voluntary tax compliance.
Revenue leakage from undetected under-reporting and registration gaps was costing the state hundreds of millions annually. The GRA's nine legacy source systems held overlapping, often contradictory taxpayer records. Compliance officers were manually reviewing flagged cases with no systematic risk scoring.
We built a master data management layer that performs entity resolution across all nine source systems, creating a single authoritative taxpayer profile. Over this, we deployed a gradient-boosted anomaly detection model trained on historical audit outcomes, producing a continuous risk score for every active taxpayer that guides compliance officer workloads.
25% uplift in voluntary tax compliance within the first fiscal year. The ML engine flagged GHS 340M in high-risk filings for targeted audit — recoveries that would not have been prioritised under the prior manual system. Full taxpayer profile consolidation achieved across all 2.1 million registered entities.
"NEXUS did not just give us a better database. It gave us a fundamentally new capability — the ability to know which taxpayers to look at. That intelligence is worth more than any number of additional compliance officers."