What does the evidence actually show?

Why large IT programmes fail: An overview of three widely cited studies.

Core thesis

In large IT programmes, deviation from the plan is the norm. Where respondents are asked about causes, they mostly name steering, coordination and decision paths and only rarely technical delivery. Three methodologically very different studies arrive at the same finding of high plan deviation over twelve years. Their figures cannot, however, be combined into a single rate.

Source basis

The analysis is based on three published studies that differ considerably in survey year and methodology. The most recent is the 2024 BCG study "Most Large-Scale Tech Programs Fail", a survey of around 1,000 companies in 59 countries. Considerably older, but still influential today, is the study by Bloch, Blumberg and Laartz (McKinsey & Company with the University of Oxford, 2012) of 5,400 IT projects. According to it, large IT programmes exceed their budget by 45 percent on average and deliver 56 percent less value than planned. The Standish CHAOS Report 2020 ("Beyond Infinity") classifies 31 percent of the projects examined as successful, 50 percent as "challenged" and 19 percent as failed. Two of the three sources are older than five years, one of them older than ten. The 2024 BCG survey shows that the basic pattern has not dissolved since.

Analysis

Together, the three studies show two things. First, in large IT programmes deviation from the plan is the norm. This applies to the Standish classification, according to which only just under a third of projects are successful, just as it does to the McKinsey and Oxford figures with their systematic deviations in budget and value. The expectation that a large programme stays within time, budget and scope without active counter-steering is not supported empirically. Second, steering topics come to the fore where causes are asked about. Only one of the three sources does this, however. In the BCG survey, more than 60 percent of the programmes name the absence of an end-to-end master plan with a critical path and clear dependencies as a central cause. 60 percent point to the absence of an active PMO that monitors value delivery and identifies emerging risks.

The differences between the sources are just as revealing as their similarities, because the three sources do not measure the same thing. Standish classifies project outcomes according to its own, repeatedly criticised definition of success based on time, budget and satisfaction. McKinsey and Oxford quantify deviations against planned values. BCG surveys which causes the companies themselves state. A failed project at Standish is therefore not the same as a "failed program" at BCG, and the figures cannot be combined into a single failure rate. What is robust is the shared direction of the findings and less the individual percentage. Three methodologically different approaches arrive at the same result of high plan deviation over twelve years.

Context

Beyond the analysis, two recurring patterns can be derived. The first is a missing master plan in which the dependencies are described explicitly. The second is a missing active PMO with the mandate to counter-steer. Both patterns lead to the same question, namely whether a programme has the steering building blocks that its actual complexity demands.

Limits of this analysis

This overview has three limits. First, it is based exclusively on published third-party research. loumiTECH has not conducted any empirical survey of its own. Second, the sources were not selected systematically, as an academic literature review would require. Publication bias is possible, because striking failure figures are published and cited more often than unremarkable successes. Third, two of the three sources come from consulting firms. BCG and McKinsey have a commercial interest in the diagnosis that programmes fail on steering, because they sell steering services. The Standish Group also markets its CHAOS data commercially. The findings should therefore be read as consistent indications and not as a neutral measurement.