A large retailer once spent $1.4 billion trying to modernise its IT systems. The project was eventually abandoned. Falling further behind competitors, the company tried again, this time with a $600 million supply-chain system.
That effort failed too, and the company ultimately filed for bankruptcy. This isn’t a cautionary tale from decades ago. Instead, it’s one of the real examples cited in McKinsey’s research on why large technology projects fail. It’s exactly the kind of outcome enterprise leaders are trying to guarantee never happens to them.
The Numbers Behind Why Nobody Trusts a Delivery Date Anymore
McKinsey and the University of Oxford analysed more than 5,400 large IT projects, defined as those with initial budgets exceeding $15 million. The findings were stark: on average, these projects ran 45 per cent over budget and 7 per cent over schedule. Additionally, they delivered 56 per cent less value than originally promised. Half of all large IT projects massively blow their budgets.
The most alarming finding wasn’t the average. It was the tail. McKinsey classified a subset of projects as “black swans,” those with cost overruns between 200 and 400 per cent. These overruns are severe enough to threaten the survival of the company that funded them. Furthermore, these aren’t rare outliers confined to government bureaucracy or legacy industries. They happen consistently enough, across enough industries, that “on time and on budget” has become something enterprise leaders explicitly demand rather than quietly assume.
Software projects specifically carry the highest risk of both cost and schedule overruns among all the project types McKinsey studied. This matters given how much of a modern enterprise’s competitive position now depends on the software it builds or buys. A retail chain, a bank, an insurer—none of them can treat a major software initiative as a low-stakes bet anymore. The data says it very much isn’t one.
That’s the real context behind the demand for guaranteed delivery. It isn’t a preference for extra reassurance. Rather, it’s a direct response to a well-documented, decades-long pattern of large technology projects going badly wrong in expensive, sometimes company-ending ways.
What Actually Changes the Odds
The projects that avoid this fate usually get a few basic things right: skilled people doing the actual work, business and engineering teams pulling in the same direction, and real checks along the way that catch a problem while it’s still small and cheap to fix.
As software initiatives become more demanding, many organisations are turning to AI-powered engineering to strengthen delivery capacity without sacrificing quality. The right engineering partner can help teams move faster, solve technical challenges earlier, and maintain momentum from planning through deployment.
Why More Code Isn’t the Same as More Reliability
It’s tempting to assume that faster code production automatically means faster, more reliable delivery. However, the data doesn’t support that leap. A team that ships code quickly but skips rigorous testing, integration validation, and requirements alignment isn’t more reliable. Instead, it’s just moving toward the same failure modes faster.
The enterprises actually improving their odds are the ones treating speed and reliability as separate problems that both need solving, not assuming one automatically produces the other. Fast code generation paired with weak governance still produces a black swan. It just gets there on a shorter timeline.
What Enterprise Leaders Actually Mean by “Guaranteed”
No serious engineering partner can promise a project will never encounter a problem. What enterprise leaders are actually asking for, when they demand guaranteed delivery, is a fundamentally different risk profile than the one McKinsey’s data describes as typical: a project structured from the start to catch and correct problems early, rather than one that discovers a 200 per cent overrun only after the damage is already done.
That’s a reasonable thing to demand, given the numbers. A 45 per cent average overrun isn’t a minor rounding error in a budget. For a $15 million project, it’s nearly $7 million in unplanned spending. In the black swan cases, it’s an existential risk to the business. Enterprise leaders asking for delivery certainty aren’t being difficult. Rather, they’re responding rationally to a well-documented history of what happens when nobody asks for it.
The Bottom Line
The retailer that went bankrupt after two failed IT modernisation attempts wasn’t unlucky. It was a predictable outcome of the same pattern McKinsey found across thousands of large IT projects: misaligned teams, weak governance, and a technology-first mindset that never adequately addressed the human and organisational factors driving the overrun. Demanding guaranteed delivery isn’t enterprise leaders being cautious for its own sake. Instead, it’s a direct, rational response to decades of data showing exactly what happens to the organisations that don’t, and it’s why the vendors who take that demand seriously, building real evaluation rigour and organisational alignment into how they work, tend to be the ones enterprise leaders keep coming back to.