advertisement
The CIO’s Two Clocks
When Absa published its first-half technology results in August, its AI section read like a set of stopwatch times. Opening a small business account had fallen from two days to under 30 minutes. Financial spreading for credit decisions, which previously took two to five days, now took four hours. Call holding times in its contact centre had dropped by 44 per cent, while effectiveness in its debt review process had risen from 28 per cent to 90 per cent.
Johnson Idesoh, Group Chief Information and Technology Officer at Absa, summed up the test he applies: technology “only matters when it changes the experience of customers and the performance of the business.”
The same bank shows the other side of the ledger. Absa wrote off $133 million in software assets in its 2025 financial year following a change in strategy and faster-than-expected obsolescence. It recorded a further $11 million impairment at the interim stage. Technology now accounts for about 28 per cent of the group’s operating costs.
advertisement
Within one company, two clocks are running. One measures AI in minutes saved this quarter. The other tracks investments that can lose their value before they have paid back.
That tension sits behind much of this week’s discussion at Africa Tech Festival in Cape Town. On stage, the debate centres on data centres, power and sovereign compute, with investments measured against 2030 and beyond. In the boardroom, the question is more immediate.
A Harris Poll survey of 685 enterprise CIOs found that 97 per cent faced greater board pressure to demonstrate measurable AI returns than they did in 2025. PwC’s latest global CEO survey found that 56 per cent of chief executives had seen neither higher revenue nor lower costs from AI over the previous 12 months.
advertisement
Africa now has its own evidence of how that gap is being closed, at least in banking. The State of AI in African Banking 2026, published in July by Backbase and African Banker magazine, surveyed 277 senior executives across 37 countries.
Two-thirds of banks, or 67.1 per cent, measure returns on their AI investments in some form. Among those that do, 85.1 per cent said the results met or exceeded their original projections. “AI pays off. The return is real, but it requires the discipline to measure it,” the report concluded.
The weak spot is the one-third of banks that do not measure returns at all. Yet 82 per cent of these institutions still plan to increase AI spending over the next 12 months.
advertisement
Measurement is also uneven within the same institutions. Finance teams track AI returns 82 per cent of the time, compared with just 50 per cent among C-suite executives. Pan-African banks, among the continent’s largest and most complex lenders, recorded the lowest measurement rate of any group, at 54.2 per cent.
The use cases executives rated as most impactful were also those with the clearest scorecards: fraud detection and transaction monitoring, followed by credit scoring. Nedbank’s use of a conversational AI platform reportedly cut live-agent chat volumes by more than half within a year.
The CIOs bridging the two timelines tend to do three things.
First, they prioritise work whose results can be measured in weeks rather than years, establishing baselines from the outset. Absa did this by tracking account-opening times and call handling performance.
Second, they concentrate spending rather than spread it across too many initiatives. Theo Mabaso, Sanlam’s Group Chief AI Officer, has argued for directing AI investment towards areas with the greatest potential impact rather than simply increasing budgets. In March, Sanlam appointed him to the role and placed him on its executive committee, alongside his technology responsibilities. The group said the move would embed AI measures and support revised capital allocation.
Paul Hanratty, Sanlam Chief Executive, said AI “has moved beyond experimentation”.
Third, they manage AI as a portfolio, cutting projects that fail to deliver. Gartner research has found that companies which consistently track AI returns, manage initiatives as a portfolio and discontinue underperforming projects reported positive returns on 81 per cent of their initiatives.
The longer clock, however, does not stop. African banks spend an average of 55.7 cents of every IT dollar maintaining legacy systems, according to the Backbase report. Legacy integration was also the barrier most frequently cited by respondents.
Those systems remain the foundation on which AI projects depend, and replacing them is a process measured in years. For CIOs, the practical lesson emerging from the continent’s banks is to use measurable short-term gains to build credibility and secure the funding needed for the slower infrastructure rebuild underneath.
For boards, the lesson in the numbers is equally direct. Banks that demanded evidence were more likely to find it. Those that did not measure returns were, in most cases, still spending.