As organisations around the world accelerate their investments in artificial intelligence (AI), digital transformation, and data analytics a growing number of industry experts are warning that many companies are still approaching these initiatives in fundamentally flawed ways.
Despite the enormous potential of new technologies, the majority of transformation programmes struggle to deliver lasting value.
According to Dr Pierre Le Roux, MD of digital consulting firm Moyo, the high failure rate of transformation initiatives is not a new phenomenon and long predates the current wave of excitement around AI.
“We are living in a time of extraordinary technological possibility,” says Dr le Roux. “Yet if you look at the track record of strategy execution, digital transformation programmes, and large IT initiatives the success rate has historically been far lower than most organisations would like to admit.”
He says artificial intelligence is now experiencing a similar pattern.
“Industry research suggests that the failure rate of AI projects is currently as high as 90%,” he says. “While that number will improve over time as the technology matures, it highlights a deeper structural problem in how organisations approach change.”
According to Dr le Roux, the root cause often lies not in the technology itself, but in the underlying condition of the organisation attempting to implement it.
“Many organisations are trying to implement increasingly sophisticated technologies on top of fragmented systems, inconsistent data, and unclear processes,” he explains. “When those foundations are weak, scaling any new capability becomes extremely difficult.”
Consultants working closely with enterprise clients frequently encounter similar challenges, he says. These include duplicated systems across departments, inconsistent data governance, overlapping processes, and unclear accountability structures.
“These issues create an environment where it is possible to demonstrate isolated technology successes, but very difficult to scale those successes across the enterprise.”
As a result, many organisations celebrate early AI use cases that deliver promising results, but struggle to expand those solutions beyond small pilots.
“The real failure often happens after the first successful use case,” says Dr le Roux. “The pilot works well, but when organisations try to scale it across the business the complexity of the enterprise starts to slow everything down.”
He believes companies should reflect carefully on how they evaluate the success of transformation initiatives. “If an organisation had a 70% failure rate in its production environment, or if only 30% of financial transactions in its ledger could be reconciled, no executive team would accept that,” he says. “Yet when it comes to transformation programmes, similar levels of failure are often accepted as part of the process.”
Dr le Roux argues that this mindset needs to change if organisations want to fully benefit from technologies such as artificial intelligence
“AI is an incredibly powerful capability, but it is not a magical solution that automatically fixes organisational complexity,” he says. “It depends heavily on the quality of the data environment, the integration of enterprise systems, and the clarity of business processes.” This is why organisations that successfully scale AI tend to approach transformation differently.
Rather than focusing primarily on individual technology use cases, they prioritise the design of strong organisational foundations. “That means building coherent enterprise architecture, improving data governance, removing duplication across systems, and establishing clear accountability structures,” says Dr le Roux. “It is essentially an engineering approach to enterprise transformation.”
Firms such as Moyo increasingly focus on helping organisations address these structural challenges before attempting to scale advanced technologies. “Once the foundational work is done, technologies like AI and advanced analytics can start to deliver the enterprise-wide improvements that organisations are looking for,” Dr le Roux says.
Without those foundations, however, many companies risk repeating the same pattern that has characterised previous digital transformation initiatives – promising pilot projects followed by slow, fragmented adoption.
“The next phase of AI adoption will be less about experimentation and more about discipline,” Dr le Roux adds. “Organisations that succeed will be the ones that combine technological innovation with strong engineering thinking and enterprise design. In the end, it is not the number of tools you deploy that matters, but the strength of the foundation you build.”