Moyo MD: Most AI projects fail as companies overlook weak foundations

Image supplied

According to Pierre Le Roux, managing director at Moyo, high failure rates in transformation programmes are not new and predate the current surge in AI adoption.

“We are living in a time of extraordinary technological possibility,” he says. “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.”

Le Roux notes that AI projects are now following a similar pattern, with some estimates suggesting failure rates of up to 90%.

“While that number will improve over time as the technology matures, it highlights a deeper structural problem in how organisations approach change,” he says.

He argues that the issue is less about the technology itself and more about the state of the organisations implementing it. Many businesses are attempting to deploy advanced tools on top of fragmented systems, inconsistent data and poorly defined processes.

“These issues create an environment where it is possible to demonstrate isolated technology successes, but very difficult to scale those successes across the enterprise,” he says.

As a result, companies often succeed in early-stage pilot projects but struggle to expand those solutions across the business.

“The real failure often happens after the first successful use case. The pilot works well, but when organisations try to scale it, the complexity of the enterprise starts to slow everything down,” he says.

Le Roux adds that many organisations tolerate failure rates in transformation initiatives that would not be acceptable in other parts of the business.

“If an organisation had a 70% failure rate in its production environment, no executive team would accept that. Yet when it comes to transformation programmes, similar levels of failure are often accepted,” he says.

He believes companies that successfully scale AI take a different approach, focusing on building strong organisational foundations before deploying advanced technologies. This includes improving data governance, integrating systems and establishing clear accountability structures.

“AI is an incredibly powerful capability, but it is not a magical solution that automatically fixes organisational complexity,” he says.

Without these fundamentals in place, organisations risk repeating a familiar cycle of promising pilot projects followed by slow and fragmented adoption.

“The next phase of AI adoption will be less about experimentation and more about discipline,” Le Roux says. “In the end, it is not the number of tools you deploy that matters, but the strength of the foundation you build.”