SARS Is Watching More Closely Than Ever — And Your Data Matters
How data analytics, AI, and connected systems are reshaping tax compliance for South African businesses
There was a time when a SARS audit seemed like bad luck. Businesses submitted their returns, kept their records reasonably organised, and hoped they would avoid attracting attention. Audits certainly occurred based on internal risk assessments, but they often felt random — more like an occasional compliance inconvenience than part of a sophisticated, data-driven system.
That era is over.
Over the past few years, SARS has become one of the most technologically advanced revenue authorities in Africa. What was once mostly reactive and manual has turned into predictive, automated, and highly accurate. For businesses still using disconnected systems and manual reconciliations, this change creates a real compliance risk.
From Instinct to Intelligence — Powered by AI
SARS has been collecting data for years. What has changed is what they do with it.
Today, SARS draws on a vast network of third-party information — banks, employers, customs authorities, the Deeds Office, payroll providers, investment platforms, and crypto exchanges — automatically cross-referenced against what taxpayers declare. VAT returns are compared to PAYE submissions. Declared turnover is assessed against banking activity. Import and export records are tested against income declarations.
Artificial intelligence accelerates all of this. SARS already uses AI-driven systems across its operations: its Lwazi Assistant handles taxpayer queries without human intervention, and during filing season, millions of data points from IRP5s, medical schemes, and investment providers are processed automatically to generate assessments at scale. On the enforcement side, machine learning models identify patterns — sudden increases in wealth, undeclared foreign income, inconsistencies between lifestyle and declared earnings — that would previously have taken auditors weeks to uncover manually.
For businesses, the enforcement lens looks somewhat different but is no less powerful. AI-driven analysis can detect mismatches between a company’s VAT outputs and inputs across multiple periods, flag turnover trends that diverge unexpectedly from sector norms, or identify inconsistencies between PAYE declarations and the payroll activity visible in banking data. Businesses with multiple entities — group structures, intercompany loans, or related-party transactions — face additional scrutiny, as AI tools are well-suited to identifying patterns across connected taxpayers that would be invisible when looking at any single return in isolation.
The result is a risk-scoring environment where each registered taxpayer has a digital risk profile. Audits are not random; they are triggered.
What Triggers a Flag?
Businesses do not have to act dishonestly to attract scrutiny. Inconsistencies — not fraud — are usually what draw attention. Common risk indicators include:
- VAT input claims that appear unusually high relative to the size or nature of the business
- Turnover figures that do not align with prior years, industry benchmarks, or expected activity
- PAYE declarations inconsistent with payroll-related banking activity
- Differences between supplier declarations and customer VAT claims
- Significant cash transactions without clearly identifiable corresponding income
Many of these discrepancies are entirely innocent — timing differences, classification issues, or human error. But innocent explanations still need to be explained and defended, and that process takes time, documentation, and professional support. Prevention is considerably cheaper.
The Real Problem: Your Systems
Traditionally, tax compliance was treated as a technical exercise — appoint the right accountant, apply the correct rules, submit on time. Those fundamentals remain important. But compliance is increasingly a systems and data issue as well.
Businesses operating across multiple disconnected platforms are particularly vulnerable. A separate payroll system, standalone invoicing software, spreadsheet-based reconciliations, and manually maintained records may each function adequately on their own — but together they create fragmented data environments where the risk of error multiplies at every handover point.
Over time, VAT, PAYE, and income tax submissions can begin telling subtly different stories about the same business. And that is precisely what modern SARS systems are designed to detect.
What To Do About It
Three questions are worth asking honestly:
- Do your VAT, PAYE, and income tax submissions tell a consistent story? If placed side by side, would the numbers broadly align and make commercial sense?
- How much manual intervention sits between your operational systems and your tax submissions? Every spreadsheet adjustment and manual transfer is an additional opportunity for error.
- Is your relationship with your advisor proactive — identifying inconsistencies before SARS does — or purely reactive at filing deadlines?
The businesses that navigate this environment most effectively are not necessarily those with the most complex tax structures. They are the ones with clean, reliable data, integrated systems, and advisors who actively monitor for risk. In a compliance world driven by data analytics and automated risk detection, your systems and processes are now as important as your technical tax position.
For more information on reviewing your compliance position or identifying data gaps in your tax submissions, speak to the Nubis team.