The shift toward digital payments in South Africa is well underway, although it is not always easy to quantify.

A range of data sources provides insight into this transition. In this blog, we examine selected data to better understand how payment behaviours are evolving.

The SARB in its Quarterly Bulletin publishes data on the number and value of Electronic Funds Transfers (EFTs) and credit card transactions. Notably absent is data on debit card transactions. Nevertheless, the data on credit card and EFT transactions presented below paints a very curious picture. We have continued growth in credit card transactions and simultaneously a decline in EFTs post 2023 (shaded in pink).

Why are EFTs declining? The short answer is they aren’t. The SARB’s EFT data includes salary payments but excludes Payshap, an unfortunate omission given that Payshap is the most exciting thing to happen in the payments space in South Africa since the previous exciting thing to happen in payments, which no one can remember.

PayInc, the clearing house formerly known as BankservAfrica, publishes a rival time series which excludes salary payments but includes Payshap. A comparison of those numbers from August 2023 to August 2025 is provided below. Some (but not all) of what we have lost in EFTs we have made up in Payshap.

Focusing on Payshap, to get values and volumes we must, unfortunately, resort to eyeballing a chart in a recent SARB report[1]. It looks like there were around 35 million Payshap transactions in March 2025 (give or take a few hundred thousand), a massive jump from the roughly three million (give or take) a year earlier. The company indicated[2] that transaction volumes in December 2025 exceeded 60 million. Because of competitive concerns, they cannot disclose more. While that growth is impressive, it is by no means clear how we would assess the extent to which Payshap has displaced cash or replaced other EFT mechanisms that are inferior with respect to both price and latency, attributes that are particularly important in payment systems.

A final point about reported EFT transactions; neither the SARB nor PayInc give us any inkling of what is happening to so-called on-us EFT transactions. These transactions occur between customers of the same bank. Data on volumes and values are not published. If banking market shares are relatively stable, we could argue that on-us trends are likely to mirror off-us trends. But market shares are not stable. Capitec’s growth continues at pace. According to the latest annual report, the bank now has 24 million clients, more than double the next largest competitor (FNB). Capitec also has a growing base of merchants and a compelling payments proposition supported by Capitec Pay, an on-us system. It is entirely plausible to suspect that on-us is growing faster than off-us. Arguably, it is these evolving competitive dynamics that have finally sharpened other banks’ incentives to improve the performance of off-us payment solutions.

Turning our attention to credit card transactions, here too growth is noticeable with transaction volumes growing by 10% per annum and (nominal) values by 14% per annum between August 2023 and August 2025.

Once again, we have some questions. Is the trend in the number and value of credit card transactions driven by growth in the credit card base or does it reflect growth in merchant adoption (and card usage)? Should we be concerned that this incarnation of digitalisation is simply credit card debt mounting at unsustainable levels? To answer these questions, we would want to turn to the National Credit Regulator (NCR) and the data it publishes. But we can’t. The NCR aggregates credit cards with store cards and other revolving credit facilities, making it impossible to isolate trends in credit card utilisation or performance. So much for that.

Likewise, the SARB does not report on credit card debt specifically but includes it together with other unsecured loans.

Happily, when the regulator closes a door, the private sector often opens a window. On the question of growth in the credit card accounts, we can explore data submitted from lenders to credit bureaus via the South African Credit and Risk Reporting Association (SACRRA). According to the latest SACRRA 71point4 Data Submission Monitor, the number of credit- and garage card accounts grew at a modest 2.4% per annum between June 2023 and June 2025. Transunion reports[3] that the number of credit card customers (as opposed to accounts) grew by 3.7% over roughly the same period. As noted, the per annum growth in volume and value of transactions over the same period (as per the SARB) was 10% and 14% respectively. Based on this data we conclude that while we do have growth in customers, much of the growth in credit card transactions is from more intensive usage.

There are several trends that are driving increased credit card use. Growth in online shopping is top of the list. Online retail spend has grown by more than 30% per annum since Covid, and now accounts for roughly 10% of retail spend in the country. According to World Wide Worx’s recent study[4] 50% of online shoppers use card as their preferred payment mechanism[5], although we don’t know whether this is credit cards or debit cards. Pity.

Bank transfers, the next most frequently cited preferred payment option, come in at 26%.

A trend to watch is Buy Now, Pay Later (BNPL), currently preferred by only 5% of customers but probably used by a lot more of them. Each BNPL purchase can generate three, six, or even twelve separate transactions. BNPL therefore inflates transaction volumes far more rapidly than transaction values. This makes BNPL deceptively powerful: it can reshape payment flows, operational load and risk profiles long before it shows up meaningfully in spend data. In short, it scales fast. And quietly. The problem is that we can’t watch this trend at all. There is virtually no reliable data on BNPL. None. No consistent reporting, no transparency, no visibility.

Another critical trend is the dramatic increase in the number of merchants offering card facilities as new (non-bank) entrants have reduced the price of POS devices and enabled access. Here too we are not blessed with abundance in terms of regularly published and accurate data. But there are some fascinating studies or at least one of them; a recent SARB working paper on revenue-based financing[6]. The paper explores data provided by a mysterious ‘fintech’ that offers “card machines, point-of-sale systems, and revenue-based financing options” to no less than 250 000 small businesses in South Africa. Let’s pause there: 250 000. That is a lot of businesses[7].

The paper’s primary interest is on testing various credit market theoretical constructs and frankly it does go on a bit. But one table at the back contains some fascinating descriptive data on these businesses and their sales which is reproduced below for ease of reference:

The problem with card-based digitalisation is that it is expensive. If we use a flat 2.7% merchant fee and assume a somewhat symmetric distribution of values / volumes, the average bakery will pay over R680 per month in merchant fees while a hardware store would fork out over R1 200. The total merchant fees generated off this sample of 35 418 small merchants is around R22.6 million per month, split between the Fintech, the acquirer and the issuer. Cards are expensive, but may still be better than cash, apparently.

Which brings us, lastly and finally, to cash itself. A key metric is cash held by households and businesses, proxied by the SARB’s Monetary sector liabilities: Banknotes and coin in circulation (KBP1312M). We explore this stock variable, normalised relative to retail sales, admittedly a flow variable, to get a picture of cash intensity in retail sales.

As per the data in the pink area of the chart, the stock of cash held by households and businesses is not keeping up with retail sales for the first time in a very long time. One interpretation is that households and businesses need less cash to transact as digital payments adoption rises. An alternative, or perhaps coinciding, explanation is that cash is circulating with greater velocity, with a given stock of cash supporting more transactions. Either way, it looks like the role of cash is shifting.

Bringing it all together, the admittedly limited data suggests that something meaningful is shifting in South Africa’s digital payments landscape. Yet, frustratingly, it remains unclear what exactly is changing and where those changes are taking place. This is somewhat paradoxical, given that one of the core promises of digitalisation is greater visibility.

Indeed, the SARB implicitly recognised this limitation when it commissioned a survey of consumers to support its 2023 Payments Study Report. However, recall-based surveys introduce their own constraints: respondents may misremember or misreport their payment behaviour. Asking consumers what payments they are making and how they make them is only a modest improvement on inference and assumption. Arguably, for a regulatory authority tasked with overseeing the payments system a more appropriate (and lower cost) alternative would be to exercise that authority directly and require banks and other payment system participants to provide robust, anonymised transactions data for the next study.