Has relationship banking reached the end of the road? It is an interesting debate because digital banking has changed how financial institutions operate.
Applications now move faster. Documents are reviewed in minutes. Credit assessments happen at remarkable speed. Yet speed on its own does not guarantee a better lending decision.
For many South African businesses, every set of financial statements tells a story. Numbers reveal performance, but they do not always explain why a business experienced a difficult month or why cash flow looked different during a particular period. When technology reads figures without understanding the circumstances behind them, good businesses can be judged unfairly.
The future of business lending is unlikely to be a choice between technology or people. Success will depend on combining both.
Note: This article is adapted from an original feature in Daily Maverick. If you’d like to explore the original insights, we welcome you to visit the article.
Faster Processing Cannot Tell the Whole Story
Automated credit scoring can process an application in minutes. What it often can’t do is understand why a business looks the way it does on paper.
A model built to flag risk will treat a late customer payment or an uneven month of revenue as a warning sign, even when the underlying business is sound. It doesn’t know that a large corporate client pays 60 days late as standard practice, or that a retailer’s slow month is simply the off-season.
This matters because the gap between what the data shows and what is really happening can be enormous. A business can appear unstable on a spreadsheet while running a tight, profitable operation in reality. An algorithm reading only the numbers has no way of telling those two situations apart.
South African Businesses Operate in a Context Algorithms Don’t Model
Local conditions make this problem more pronounced than in many other markets. South African SMEs deal with irregular customer payments, seasonal cash flow swings, load shedding disruptions, and a large informal economy that doesn’t always show up cleanly in financial statements.
A business owner navigating these realities isn’t necessarily mismanaging their finances; they’re operating in an economy that behaves unpredictably by nature.
An experienced lender who has seen dozens of similar businesses can recognise the difference between a temporary wobble and a structural problem. That kind of pattern recognition, built from years of exposure to how South African businesses actually function, is difficult to encode into a scoring model.
Automation Should Handle the Routine, Not the Judgement Calls
None of this is an argument against technology. Automation earns its place in modern lending. It can process documents, verify data, flag anomalies, and cut turnaround times dramatically. Used well, it frees up human bankers from administrative work so they can focus on the decisions that actually require thought.
The mistake is asking automation to do a job it was never designed for: making the final call on complex, nuanced credit decisions. When a business’s story doesn’t fit neatly into a standard model, someone with experience needs to look past the numbers and ask better questions. Technology is a tool for efficiency, not a substitute for judgement.
Trust Still Wins Business, Even in a Digital Market
One of the more overlooked points in the debate is how much trust still influences who a business chooses to bank with. Owners want a lender who understands their industry, their growth plans, and the specific pressures they face, not a generic product built for the average applicant. A banker who takes the time to understand a business can offer a solution shaped around that business, rather than a one-size-fits-all facility that technically qualifies but doesn’t quite fit.
This is where relationships still outperform pure automation. A tailored financing solution, backed by someone who understands the client’s goals, tends to serve both the lender and the business better over time than a standardised product approved purely on a score.
An Instant Yes Isn’t Always the Right Yes
There’s an understandable appeal to instant approval. Nobody enjoys waiting weeks for a lending decision. But speed becomes a liability when it replaces proper assessment.
A fast approval that ignores the full context of a business’s situation can lead to poor outcomes on both sides: a lender taking on more risk than it realises, or a viable business being declined because a model misread its situation.
Good lending decisions require weighing the full picture. That takes a bit more time than an automated yes or no, but the outcome tends to be sounder for everyone involved.
The Winning Model Combines Both
The most compelling argument in the piece is that this isn’t a contest between digital banking and personal relationships. It’s about knowing which parts of the process should be automated and which still need a human involved. Data analysis, document processing, and routine administration are exactly where AI adds value.
Context, judgement, and strategic advice are where people remain irreplaceable.
Banks and alternative lenders that combine these two strengths deliberately, rather than defaulting entirely to one or the other, will be best placed to serve SMEs properly. The technology handles the volume and the speed. The banker handles the nuance and the trust.
Final Thoughts
Relationship banking wasn’t killed by technology, it was tested by it. And what that test has revealed is that the businesses best served going forward will be the ones working with lenders who use technology to work faster without giving up the judgement that comes from actually understanding a business. For a business owner choosing a funding partner, the question isn’t whether a lender uses AI. Most now do. The better question is whether that lender still has someone on the other end who can look past the numbers and see the business for what it is.