A bureau report is a record of what a consumer has done on registered credit. It is a backward-looking document: repayment history on facilities that lenders have reported, accumulated over years, scored by algorithms that weight age and consistency. A bank statement is different in kind, not just in format. It shows what is happening in the consumer’s financial life right now: money in, money out, who they owe, how they spend, and whether they are under stress.
These are not competing tools. They answer different questions. But SA credit providers who rely primarily on bureau data are systematically missing information that only the statement contains, and some of that information is not optional. Regulation 23A requires it.
What the bureau is and what it isn’t
A bureau report captures registered credit behaviour: accounts opened with NCR-registered lenders, payment history across those accounts, current balances, credit utilisation, number of enquiries, derogatory flags (judgements, defaults, debt review). TransUnion, Experian, XDS, and CompuScan collect this data from credit providers who report to them through monthly submission cycles.
The bureau answers one question well: has this consumer serviced their registered credit reliably? It answers this question across a long time horizon, which makes it useful for understanding repayment character. A consumer with 10 years of clean bureau history genuinely has demonstrated something meaningful about their financial discipline on registered credit products.
What the bureau does not contain: income data of any kind. Current spending behaviour. Non-registered obligations. Anything that happened in the last 4–6 weeks before the enquiry. Any signal from unlicensed lenders, informal debt, stokvels, or private financial arrangements. And for 1.4 million consumers who open new credit accounts each year with thin or no bureau file (TransUnion SA data), the bureau has no answer at all.
Ten things only the bank statement reveals
Each of these is real, observable in standard transaction data, and absent from bureau reports. Each affects credit risk in ways that bureau-only assessment cannot see.
1. Gambling transactions. Betway, Hollywoodbets, Sportingbet, and online gambling platforms show as debits in bank statement data. Absa has publicly stated that gambling spend is a significant predictor of delinquency. Not whether the consumer has gambled at all, but frequency and share of wallet. A consumer directing 12% of monthly income to gambling platforms has a materially different risk profile from one with occasional R50 bets. The bureau has no visibility into any of this.
2. Cash withdrawal patterns. High-frequency ATM withdrawals are a financial stress signal. When a large share of outflows is cash, the credit provider loses visibility into actual spending, and high cash usage often correlates with informal debt obligations (mashonisa loans, private lending arrangements) that are serviced in cash and never appear as structured transactions. Bureau data records nothing of this.
3. Informal loan repayments. EFTs to individuals or to unlicensed lenders are off-bureau obligations. A consumer repaying R2,500 per month to “J Nkosi” or to an unregistered microlender has a real monthly obligation that reduces disposable income, but no bureau report will ever show it. Regulation 23A requires assessment of “all monthly debt repayment obligations,” which explicitly includes these.
4. Stokvel and burial society contributions. Fixed monthly debits to stokvel coordinators or burial societies are common in SA across income bands. These are genuine financial commitments; many consumers rely on stokvels as primary savings vehicles and would not default on them. They represent real monthly outflows that reduce the disposable income available for debt service. Invisible to bureau. Visible in the statement.
5. Medical emergency payments. A large one-off debit to a hospital, clinic, or medical specialist in the three-month window is a financial shock event. It may or may not affect ability to service new credit, but it is observable context. A consumer who paid R18,000 to a private hospital last month may have depleted emergency reserves significantly. The bureau will not show this for months, if ever.
6. Subscription creep. Streaming services, app subscriptions, cloud storage, software licences, gym memberships: small monthly debits accumulate. A consumer who added R800 per month in recurring subscription costs over the past year has reduced their effective disposable income by R800 per month without any bureau-visible credit event. Bureau data from 12 months ago would not capture this shift.
7. Income source changes. A salary credit from a different employer mid-statement, or salary deposits that transition from a corporate payroll to a freelance payment pattern, indicates employment change. This is critically relevant for vehicle finance and any long-term product. The bureau score from before the employment change is still a 730, but the income picture has shifted materially.
8. Income volatility. Irregular salary deposit dates indicate precarious employment: contract work, commission-based income, or gig employment where payment timing varies. A consumer whose salary arrives between the 15th and the 28th of different months, in amounts that vary by 25%, is in a fundamentally different position from a consumer with a fixed monthly payroll credit on the same date. The behavioural score captures this through the income coefficient of variation. The bureau has no income data at all.
9. BNPL debit orders. Buy-now-pay-later operators representing R815 million in SA transactions in 2025 (projected R1.3 billion by 2030) are largely not NCR-registered credit providers. Their repayment debit orders appear in bank statements as recurring obligations but are absent from bureau files. A consumer with three active BNPL repayments totalling R1,200 per month has obligations the bureau cannot see.
10. Loan stacking. Multiple lender income credits in a single month and multiple new debit orders to unfamiliar payees are detectable loan stacking signals, before bureau monthly submission cycles have captured the new obligations. Ocrolus data suggests a borrower applying for a second loan within 15 days carries four times the fraud risk of a borrower with a single active facility. That signal lives in the statement, not the bureau.
The bureau data lag: what happened last month isn’t there yet
SA credit bureaus update through monthly submission cycles, typically 4 to 6 weeks from a payment event to score update. This is not a bureau failure; it is a structural feature of how the data flows from lenders to bureaus. The lender reports monthly. The bureau processes and updates. The score reflects what was reported, not what is happening.
For most credit decisions, this lag is immaterial. A consumer’s five-year repayment history does not change meaningfully because a payment from last week has not yet landed in the bureau data.
For specific risk scenarios, the lag is the entire problem. A consumer who lost their job six weeks ago still shows a 730 bureau score. A consumer who took out three payday loans from Boodle, Lime24, and FairMoney last month may not have those obligations on bureau yet if the providers have not completed their monthly submission cycle. A consumer who started gambling heavily in March will not show bureau distress until missed payments filter through in May or June. The statement shows all of this in real time.
The credit-invisible blind spot
TransUnion SA estimates 1.4 million consumers open new credit accounts annually with thin or no bureau file. These are not financially inactive consumers; many have consistent incomes, controlled spending, reliable debit order histories, and healthy cash buffers. They simply have no credit footprint yet. The bureau’s answer to “what is the risk of lending to this person?” is “no data.”
The bank statement provides 12 months of observable financial behaviour where the bureau sees nothing. A consumer with 24 months of consistent salary credits, zero gambling transactions, healthy month-end balances, and orderly debit orders is demonstrably lower risk than the bureau’s silence suggests. As explored in our piece on credit-invisible consumers, thin-file does not mean high-risk, but you need the statement to know that.
Why Reg 23A requires the statement, not just the bureau
Regulation 23A(4) specifies the acceptable income verification documents: the latest payslips, three months of bank statements, or income statements. The credit bureau is not on the list. This is not an oversight; the bureau has no income data. Reg 23A’s verification requirement is structurally impossible to satisfy with bureau data alone.
The regulation also requires assessment of “all monthly debt repayment obligations of the consumer,” not just those on bureau. A credit provider who identifies obligations only by pulling a bureau report has a process that systematically undercounts obligations whenever informal lending, off-bureau BNPL, stokvel commitments, or unregistered lender repayments are present. The statement-based obligation enumeration is the only method that can catch all of them.
Beyond income and obligations, the August 2025 draft amendments to Regulation 23A (comment deadline 13 September 2025, promulgation expected 2026) tighten the expense norm requirements further: mandatory minimum household expense norms by income band that credit providers must apply regardless of what consumers declare. Applying these norms correctly requires knowing the consumer’s actual income, which requires the statement.
Combining both for a complete picture
The practical framework for credit providers is not to choose between bureau and statement. It is to use each for what it actually answers.
Bureau data provides the long-run repayment track record. Five years of clean bureau history from a consumer is genuinely informative about their credit discipline and financial character over time. That historical signal has real predictive value for long-term products like vehicle finance and home loans.
Bank statement data provides current-state financial reality. It tells you whether the salary is still arriving. Whether new obligations have appeared in the last month. Whether spending patterns have changed. Whether there are stress signals invisible to the bureau. And it satisfies the Reg 23A verification requirements that bureau data structurally cannot meet.
For a detailed look at how the two signals compare as scoring inputs, the post on bureau score vs bank statement score breaks down the mechanics. AffyScore’s analysis systematically extracts and classifies all ten statement signals covered above, documents each finding’s contribution to the affordability calculation and behavioural score, and produces a Reg 23A-compliant record at origination. The bureau tells you who they were. The statement tells you who they are.
