In recent years, Local businesses, including state-owned enterprises, have repeatedly struggled to implement critical IT systems on time and on budget.
From retail giants facing warehouse bottlenecks to telecommunications and municipal services plagued by billing and operational outages, the warning signs are clear: system breakdowns, missed deadlines and rising costs. The technology itself – erp softwareBilling platforms or supply-chain systems – often proven and widely adopted globally. What fails is the monitoring, planning and implementation of the programme.
Behind such headlines is a pattern that organizations often ignore: Large IT programs rarely fail suddenly. Rather, they are swept away. Warning signs appear early but are often ignored or misunderstood. Budgets expand slowly, deadlines slowly shrink, and by the time leadership recognizes the scale of the problem, recovery costs far exceed the original investment.
South Africa has seen this pattern before in large infrastructure programs. The construction of Eskom's Medupi and Kusile power stations suffered massive cost overruns and years of delays, reshaping the country's energy landscape and negatively impacting the economy. These mega-infrastructure projects exhibit surprisingly similar underlying dynamics: initial optimism, underestimated complexity, and problems that quietly grow over time.
In IT, risks can be even more difficult to detect. Across all sectors, South African organizations have experienced the consequences of large-scale digital programs that have struggled to deliver as planned. disruption in systems such as eNatisService disruptions affecting platforms administered by social grant agency Sasa and delays at airports due to immigration systems have demonstrated how technological instability can affect millions of citizens. In the private sector, Absa Group wrote off about R2.4 billion in software assets after projects failed to deliver expected value, and a major retail group is a facing trial Due to IT system implementation failures.
optimism bias
These results are not unusual in a global context. Research by Oxford University scholars Bent Flyvbjerg and Alexander Budzier describes what they call the “iron law of megaprojects”: over budget, over time, under profit – again and again.
One of the most powerful drivers of this pattern is optimism bias. Organizations routinely approve IT programs based on overconfident estimates. Budgets are often built around a P50 estimate, which implies only a 50% chance of meeting costs and set goals. Yet these projections are often presented as firm commitments to executives and boards.
Reading: The move to cloud is accelerating spending on IT services in SA
A more realistic planning approach requires a P80 confidence level, which accepts uncertainty and provides a buffer against the unknown complexities that inevitably arise in large-scale digital programs.
Software projects are particularly vulnerable because their progress is less visible than physical construction. A bridge that is only half built, this is clear to all. Conversely, a software platform may appear to be progressing well, while serious design flaws or integration failures accumulate beneath the surface. When deadlines approach, verification processes such as testing are often compromised, leading to operational failures.
Adding to the challenge is the extraordinary variation in software productivity. Research shows that some teams produce more than 10 times the output of other teams working under similar conditions. The scope is often poorly defined, and flaws often emerge months after critical design decisions have been made. As a result, projects can appear healthy, even as value quietly drains from the program.
By the time senior leaders recognize the drift, recovery may require large cost increases or even complete program redesign.
A practical method for improving planning accuracy is reference class forecasting, developed through the work of Budzuer and Flyvbjerg and used extensively by the author. Rather than relying solely on internal estimates, this approach compares proposed initiatives to the results of similar projects completed elsewhere. By basing forecasts on historical evidence, organizations can substantially reduce optimism bias and prepare more reliable budgets and programs.
Yet better forecasting alone cannot solve the deeper problem. Large IT programs succeed or fail largely because of the capabilities and governance structures surrounding them. Effective oversight requires leaders who understand technical complexity, recognize systemic interdependencies, and are able to detect early signs that a program is deviating from its path. Without these capabilities, even well-designed projects can unravel as unexpected interactions accumulate between systems, teams, and vendors.
Meeting this challenge requires more than better project management tools. Large digital programs demand leadership capabilities that combine systems thinking, rigorous problem investigation, interdisciplinary learning, and strong communication between technical and executive teams. When leaders develop these competencies, they are able to identify hidden interdependencies, challenge unrealistic assumptions, and detect early warning signs before a program goes beyond recovery.
complex ecosystem
This is especially important as South Africa accelerates its digital transformation. Government service platforms, financial systems, telecommunications networks, logistics infrastructure and retail supply chains depend on increasingly complex software ecosystems. When these systems fail, the consequences extend far beyond any one organization. They affect service delivery, economic productivity and public trust.
For decision makers, the lesson is simple. Three questions should be asked before approving the next big IT program:
- What do comparable projects tell us about realistic costs and timelines?
- Is the budget based on evidence, or on optimistic assumptions?
- Can those responsible for governance identify problems quickly?
Often, early warning signs can be found. A whistle-blower flagged the retail failure long before business operations were affected. Careful observation such as looking for projects that spend little time on feasibility and design. They move rapidly toward creation but often do so because of incomplete understanding. On the other end of the spectrum, projects that spend too much time describing requirements and design in detail are usually already struggling to reach a stable conclusion. In both cases there is a problem in delivery.
These questions may seem simple, but they address the root cause of why many digital initiatives struggle.
The reality is that most IT programs don't take off overnight. They gradually move away from their original objectives until recovery becomes extremely costly. The organizations that succeed are those that recognize drift early, while there is still time to correct course.
For South Africa, where digital capabilities will increasingly determine economic competitiveness, ignoring these warning signs is not costly. This is a strategic risk.
- The author, Bram Meyerson, is an executive member of Convocation Da Vinci Institute and CEO of QuantiMetrics