The Federal Inland Revenue Service’s (FIRS) TaxPro Max platform, introduced to modernise and enable e-tax filing and payments by companies, represents a major step in Nigeria’s digital tax collection, but it lacks the advanced predictive capabilities needed to detect sophisticated tax fraud and revenue manipulation, a data scientist, Mr. Emeka Atuma, has said.
Atuma, an independent consultant and trainer in Data Science, made this assertion on Friday at the 10th Blakey’s National Tax Conference held in Lagos, where he presented a paper titled “Data and Artificial Intelligence in Administering Tax in Nigeria.”
The Principal Partner and CEO of Management Consulting and Business Advisory Services stated that while TaxPro Max can improve compliance through electronic filing, it cannot accurately identify tax avoidance schemes and more complex problems.
According to Atuma, TaxPro Max largely functions as an electronic filing and payment platform, with little analytical depth to assess the credibility of returns submitted by taxpayers.
“TaxPro Max is a start, but it does not have the predictive analytics to determine whether a filing is fraudulent or whether a taxpayer is deliberately manipulating the system,” said the Data Scientist.
He explained that advanced Artificial Intelligence (AI) systems can analyse historical data, detect unusual patterns and flag suspicious filings—such as companies consistently keeping revenue just below tax thresholds to reduce liabilities—but such intelligence is currently missing in FIRS’ TaxPromax.
Atuma’s explanations were in response to a member of the audience who complained about issues with his e-filing where he was wrongly assessed and an inaccurate tax amount was imposed on his firm and was never reviewed in spite of his protests.
“All I was told is go and pay,” the participant in the conference claimed.
Atuma noted that AI combines data science, machine learning and deep learning models to analyse large datasets and identify compliance gaps. Once reliable data is available, AI can distinguish between compliant taxpayers and those at high risk of evasion, allowing tax authorities to focus audits where they matter most.
However, he stressed that AI does not work in isolation. “AI can only function where quality data exists,” he said, adding that poor data integration across government agencies could undermine tax collection efficiency.
The expert identified fragmented taxpayer data as one of FIRS’ biggest constraints. Information on the same individuals and businesses is spread across multiple agencies, including the Corporate Affairs Commission, National Identity Management Commission and the Central Bank of Nigeria.
He insisted until these sets of data centres are aggregated, cleaned up and standardised, they might not be good enough to work with in advanced AI analytics.
He also cited infrastructure deficits—unreliable power supply, limited bandwidth and outdated legacy systems—as barriers to deploying advanced analytics at scale.
Rather than scrapping TaxPro Max, Atuma recommended extending the platform by embedding machine learning algorithms behind it.
This would allow the system to analyse stored tax returns, flag anomalies and support risk-based audits without replacing human judgement.
Such models, he said, can cluster taxpayers with similar behaviour and highlight filings that deviate sharply from established patterns for further investigation.
Such models, he said, can cluster taxpayers with similar behaviour and highlight filings that deviate sharply from established patterns for further investigation.
However, he stressed that privacy laws prevent the government from directly accessing citizens’ bank accounts even as he noted that they have access to bank accounts of individuals and corporate organisations.
He is of the view that AI systems should instead rely on indirect verification, notifying taxpayers of discrepancies and requesting clarification through due process.
Atuma added that even the most advanced AI system would struggle to capture Nigeria’s large informal economy without broader data collection mechanisms. Many economically active individuals operate outside the banking system, limiting visibility for tax authorities.
To address this, he suggested expanding data capture points beyond banks to areas such as SIM registration, healthcare access and other everyday services that require identity verification.
“AI cannot work outside available data,” he said, adding that telecoms data could provide a critical entry point for identifying economically active individuals without BVN, NIN, TIN or bank accounts.