How Capitec Got to 26 Million Clients Without Losing Its Nerve

Graham Lee has been at Capitec for 23 years, long enough to have joined while it was still making the transition from micro lender to bank. A year into the group CEO seat, he sat down with Colin Iles on the AWS Founders Series to answer a question that gets harder to dodge as the numbers get bigger: why do 26 million South Africans bank here, and can a 25 year old company keep behaving like a young one?

Four principles, and the discipline to keep them

Twenty five years ago a small group walked the streets of South Africa asking what was excluding people from banking. The answer was not a shortage of banks. South Africa already had a strong banking industry and a sophisticated financial system. The problem was complexity, fees hidden behind asterisks that hit you when you least expected them, branches unreachable for someone working on a mine or a farm, and customers treated poorly when they walked in.

Capitec was founded on the antithesis of those four things. Simple and intuitive. Accessible to everybody, through longer hours, more places and every channel. Affordable. And the same quality of personal service for every single person. "Everybody gets the same colour card," Lee said, and the service should make a client feel it is a privilege for the bank to serve them, not the other way round.

Michiel, Riaan, André du Plessis and Gerrie had all of this in mind from the start, according to Lee, who picked it up as an early employee around the time the banking licence arrived. The principles were baked in, not retrofitted. What makes them more than a poster on a wall is that they work as a filter: affordability, in particular, is not a pricing decision but an organisational design decision.

Designing for scale before you have it

"We seek to provide the greatest value at the price point," Lee said, drawing the distinction from being cheapest. The trap he avoided was pricing high on the assumption you can cut later once volume arrives. Start there and you are already on the back foot. Capitec designed and priced for the scale it intended to reach.

Behind that sits a long term view, reduced to a sentence: "If we treat our clients well, our clients will treat our shareholders well." It is held in place structurally rather than by good intentions. There are no separate bonus or profit pools for different parts of the business, just one shared bottom line. For senior executives a significant proportion of earnings comes as long term incentives in the form of at the money options over four, five and six years, which pay zero if the share price does not grow.

Lee also insists that planning for today is not enough. Anyone wanting to be successful needs to work in at least four time frames at once: protect and grow what you are already good at, because if you lose sight of it you will lose it; accelerate the businesses that should deliver material revenue in one to three years; incubate what you started a couple of years ago for value in three to seven; and put the strategies, capacity and actual hires in place now for growth ten years out. All four need action today, not just strategy today.

The word that gives you focus

Asked how he handles the volume of decisions landing on a group CEO, Lee started somewhere unglamorous: "The only word that gives you focus and priority is the word no." Saying yes to more things is more likely to be distracting. Saying no well means it does not consume your time, and, done properly, does not consume anyone else's either.

Then there is a framework of lenses. The first is always the client. Is it clear why this is better for the client, and how it creates value for them? Then the fundamentals, then whether the organisation would be proud of it, then whether it sits in line with the cultural anchors of the business, which Lee argues is the most important test for strategy.

He is unusually relaxed about regulation, which he thinks can be productive rather than merely obstructive, provided people remember what the goals are. The failure mode is not the committee, it is treating the committee as a meeting: the attendance, the minutes, the pack. "The committee is the three of us, not the meeting dates, or the minutes." With Grant, the CFO, and Ismail, the CRO, a decision can move fast and still be properly sense checked.

Culture is a daily fight

Losing a culture like Capitec's, Lee said, is the easiest thing in the world. Keeping it is the hardest. It is not switched on once, it has to be fought for daily, and the strongest reinforcement is not communication but decisions, each one either building the culture or breaking it.

The mechanisms are the people mechanisms. Hire deliberately for resonance with the values, and be explicit up front about what is good and what is hard about working there so people join with open eyes. Then rate values and doing things the right way above outcomes at goal setting, appraisal, promotion and reward. Asked directly whether people have been exited for failing that test, the answer was yes.

The dashboard matters too. If the daily focus were the share price or the bottom line, that alone would break the culture. Instead Capitec watches what it calls product health, a combined measure of client experience, stability and protection against crime and fraud, with downtime treated as a first order problem rather than an IT footnote.

Staying close to clients is procedural rather than sentimental. On the day of the interview the CFO, Grant, and the entire senior finance team were in Rustenburg for the week, working through branches and walking the streets to speak to the small businesses around them. Basani, the executive for personal banking, was in Thohoyandou doing the same. It is built into goal planning and demanded of the group executive.

Branches, competitors, and the client lens

Colin put an old observation to him: two Capitec branches on opposite sides of the same street in downtown Joburg, at a time when everyone else was closing branches. Lee offered worse examples, including two branches in the same mall at Gateway. The logic is accessibility, and it follows the data and the footfall. Capitec's vision is to serve 95 percent of South Africa, and within that group are many clients still transacting in cash and many in remote areas not yet comfortable with digital. There will be more branches at the end of this year than there are now.

"Serving humans with humans is a competitive advantage," Lee said. Digital tools work best where they empower people to serve other people better, particularly on the hard decisions, and banking is full of hard decisions.

The same client lens sorts the competitive noise. Lee likes Amazon's famous indifference to competitors but will not copy it. With every retailer seemingly planning a bank, Pepkor bringing a ready made branch network in its store footprint, Shoprite in the mix and Revolut newly in the market with a subscription model, ignoring the field would be arrogant. The discipline is to ask what they do better for the client, now and in five years, learn from that, and stop there.

AI is already in production across all three tiers

Colin framed AI adoption as three tiers: staff access to tools, developers and internal functions, then customer facing value. Lee's answer was that Capitec is delivering real value across all three.

Generative AI is live inside compliance operations and anti fraud operations. An AI agent embedded in business bank credit processing has cut the turnaround time on a credit decision by close to 95 percent. On the front line, staff have real time contextualised insight on the client sitting in front of them, which the client sees at the same time because of the side by side consulting model. Call centre agents know what a client is phoning about before the conversation starts.

Lee described being astounded at a recent internal tech and data conference by how much had changed in six months, across penetration testing, cyber security, marketing, front ends and payment systems. The framing stayed disciplined: "The goal is not embedding Gen AI." The goal is the level of customer service you aspire to, and Gen AI is the tool.

On whether autonomous banks run by ten or twenty people will exist, Lee thinks they will, and that they will be competitors. Capitec would build one inside the house rather than exile it to a startup campus.

Friction is a product decision

Colin recounted being blocked from making a transfer while on a call, the app having flagged a possible vishing attack. Frustrating in the moment, and exactly the point. "You want all the bad guys to have 100 percent friction," Lee said, while good clients get as close to frictionless as possible. The way through is personalisation, so two clients making the same payment do not necessarily get the same experience.

Trust rests on three focal areas: stability, security, then client experience. Nothing breaks trust like not being available when a client needs to buy prepaid electricity at three in the morning. And when there is friction, or a no on a credit application, explaining why is what lets you deliver the message without breaking the relationship. Lee was candid that they do not always get this right.

Credit is time, not money

The best line of the hour was about credit. "Credit done well is not money, it's time and opportunity." A small business bringing forward its ability to produce and sell. A young person getting into a house years earlier than they otherwise could. A seller releasing twenty years of value and retiring on it. The harms people rightly worry about always trace back to the wrong amount, given to the wrong person, who could not afford it.

On thin file lending, Lee made a distinction worth stealing: do not confuse the input with the output. A credit record is an input. What you actually want is a forecast of each individual's future behaviour, their propensity and their ability to repay, and none of that is a point in time fact.

So the approach is to use every dataset with predictive strength: the credit bureau, transaction information, contact data from the cell phone networks, and more becoming available all the time. Store it permanently, govern it properly, get the metadata right, and publish features so the rest of the company can use them, since something predictive of mortality may also be predictive of credit risk.

Payments is where the disruption lands next

Lee named payments as the space most likely to be disrupted over the next year or two, and said it will arrive quickly and significantly. New technology, alternative rails and the South African environment are all pushing the same way. Payments will get cheaper, more accessible, and more embedded inside whatever the client is actually trying to accomplish. Capitec's stance is to lead rather than be swept along, engaging with regulators and industry bodies including PayInc, and developing solutions others could use as blueprints.

The caveat is crime. Pix in Brazil is widely read as a success story, and largely is, but Lee pointed at the less publicised part: how much money was stolen in the first couple of years. Controls have since gone in and may be working, but reactively, after billions of dollars were lost. South Africa should put those controls in proactively rather than waiting for clients to be robbed first. Cross border payments are the other obvious target, on speed, on knowing your fees in advance, and on the size of them.

The hard parts, and what comes next

Asked for the difficult moments, Lee reached for the credit cycle rather than the headlines. 2008, when African Bank had its first set of problems. The cycle coming out of COVID, when he was running the personal bank and had mounting credit losses that were his to own and fix. Viceroy, he said, was a blip, and more useful as a lesson in how to manage a fake crisis than as a crisis in itself. "All of those difficult times aren't really that difficult," he concluded, because the destination is clear, the behaviour is clear, and he likes and trusts the people he does it with.

As for the pipeline: more products already well down the delivery road that he would not be drawn on, then business banking to disrupt, an insurance business that today is a tiny slice of the market and only life cover, an MVNO he sees taking Capitec meaningfully forward, and enterprise payments and data and insights businesses that move it beyond banking. In the longest time frame, beyond South Africa's borders.

For a 25 year old bank being asked whether it can still behave like a young one, that is a reasonably convincing answer.