Organizations are under growing pressure to demonstrate measurable returns from their AI investments. The experimentation phase is ending, and business leaders increasingly want to understand how AI will improve productivity, accelerate decision-making, reduce costs, and drive growth.
Most enterprises understand the foundational requirements for success. They know AI depends on high-quality data. They know fragmented systems create challenges. They know modernization is necessary. Yet many continue to struggle to move from ambition to execution.
Gartner reports at least 50% of GenAI projects were abandoned after proof of concept by the end of 2025 due to poor data quality, risk controls, costs, or unclear value. McKinsey states only about 40% of companies report enterprise-level EBIT impact from AI.
The challenge is not a lack of interest in AI. The challenge is that many organizations are still operating in environments that were never designed to support the speed, scale, and data requirements that AI demands.
Enterprise complexity remains the biggest obstacle
Several factors can prevent organizations from realizing value from AI and broader transformation initiatives, particularly in larger enterprises.
Through the years, many organizations have grown through mergers and acquisitions, layering new systems onto existing IT environments and creating highly fragmented landscapes with “one of everything.” This increases operational complexity and makes modernization more difficult. Even upgrading systems to the latest version only goes so far in simplifying the environment and can sometimes add further complexity.
These fragmented systems also create challenges for AI initiatives, since data spread across multiple platforms with different structures makes it harder to generate reliable business insights.
INEOS Energy’s move from SAP ECC to SAP S/4HANA highlights how this complexity plays out in practice. Years of mergers and acquisitions had left the company’s existing environment burdened with redundant data, outdated configurations, and a fragmented organizational structure. Rather than simply lifting and shifting systems, the transformation required addressing that underlying complexity as part of the move to a modern ERP foundation.
The result is that organizations often spend significant time and effort trying to reconcile data before they can generate meaningful value from AI.
Most enterprise data is still not AI-ready
Expectations for enterprise technology initiatives have evolved significantly as organizations increasingly look to these environments to support AI, analytics, and broader business transformation efforts.
Most enterprise data is still not in a format suitable for AI, which creates an immediate disadvantage for organizations trying to operationalize these technologies. Organizations often face data quality challenges because system landscapes have grown over years, if not decades. Data has often accumulated over years without being consistently cleaned or standardized, resulting in empty fields, uncategorized records, invalid values, and other quality issues.
At the same time, enterprise data is typically spread across multiple systems, prompting many organizations to invest in master data management solutions to create “golden records,” though these efforts have often fallen short of delivering complete, reliable data. Bridging that gap has become one of the biggest challenges in modern enterprise environments. Operational enterprise data is often not AI-ready.
This challenge becomes even more complex when data exists in unstructured, static formats that are difficult to track and manage without significant manual work, requiring organizations to repeatedly transform and standardize information to create a usable enterprise-wide data set.
Organizations frequently underestimate the amount of effort required to prepare data for AI. Instead, the focus shifts quickly to models, tools, and use cases before the underlying data foundation is ready.
Complexity continues to compound
Customization presents another major challenge.
Many enterprise environments remain unnecessarily over-customized, even as standard software capabilities have evolved and made some of those modifications obsolete. While some customizations continue to support important business requirements, many remain in place long after their original purpose has disappeared.
Every unnecessary customization increases complexity, which makes upgrades, integrations, and modernization efforts more difficult.
Coop’s migration from SAP ECC to SAP S/4HANA demonstrates how customization can complicate modernization efforts. Its two ECC systems had been running for many years and had been extensively adapted, with numerous in-house developments and more than 400 interfaces. Before the migration could proceed, every interface had to be evaluated, and custom code and SAP simplification-item checks became a critical part of the preparation.
This illustrates why customization cannot be treated as a purely technical detail. Every additional interface, in-house development, and legacy adaptation increases the number of dependencies that must be understood, validated, and either carried forward or retired before a modern ERP environment can deliver its intended value.
Organizations also often retain large volumes of outdated data in active production systems, increasing complexity further. While some of this information may need to be retained for audit or compliance purposes, it does not necessarily need to remain part of day-to-day operations.
When organizations modernize without simplifying systems, retiring unnecessary customizations, and rationalizing legacy data, initiatives may succeed technically while still falling short of the business value expected.
AI needs a clearly defined business problem
Many companies are approaching AI initiatives with much broader expectations than they did historically.
At the same time, many organizations have not clearly defined the specific business problem they are trying to solve. Broad directives to “use more AI” or “improve sales with AI” lack the specificity needed for success, making outcomes harder to achieve and measure.
Successful AI initiatives start with a clearly defined business objective. Organizations need to understand what problem they are solving, how success will be measured, and whether AI is the right solution for that challenge.
Without that foundation, it becomes difficult to connect AI investments to meaningful business outcomes.
Organizations are modernizing the business, not just technology
Complexity, integration, and data challenges can significantly affect outcomes because organizations are not simply modernizing systems or data. They are modernizing the business itself.
That means ensuring critical operations continue seamlessly before, during, and after transformation takes place.
The guiding principle should be simple: keep the best and transform the rest. Data migrations should follow a defined process with clear validation points, traceability, and proactive data preparation ahead of major transformations. Organizations often see better results when they focus on smaller, targeted improvements rather than trying to change everything at once.
When these steps are followed, outcomes become far more predictable.
The race against time
One of the most critical challenges organizations face is the race against time.
It is not uncommon for modernization projects to be scheduled over five- to ten-year timelines. At that point, it becomes nearly impossible to realize the intended value from these initiatives simply due to the timeline involved.
The original business case was built on a set of assumptions that were true at the start of the program, and hopefully for the near future. But after several years, business conditions may have changed significantly due to macroeconomic shifts, technology disruptions, new business trends, and even geopolitical realignments.
Another example from Nebraska Public Power District shows the value of compressed execution. Companies traditionally take about two years to upgrade SAP ECC systems, especially when the upgrade is combined with other initiatives. NPPD completed its SAP S/4HANA migration in just nine months, while also consolidating CRM and core operational systems and reducing integration complexity.
As a result, much of the value organizations expected to capture may have already evaporated.
This is why a highly automated, software-based, and AI-assisted approach is critical to reducing timelines, aligning execution more closely with the original business case, and ultimately delivering the promised outcomes.
Closing the execution gap
The organizations generating the most value from AI are not necessarily those investing the most money.
They are often the organizations that have focused on reducing complexity, preparing their data, modernizing their environments, and aligning technology initiatives with clearly defined business objectives.
AI remains a significant opportunity for enterprises. But turning that opportunity into measurable ROI requires more than deploying new tools. It requires building the foundation that allows those tools to succeed.
Ultimately, the AI execution gap is not just an AI challenge. It is a data, modernization, and business transformation challenge. Organizations that address those foundational issues will be far better positioned to realize the value they expect from AI.
