By Garrick Schermer, AI Data Strategy & Governance Lead
When MIT’s NANDA report recently highlighted that 95% of GenAI projects fail, many were quick to seize on the negative headlines. But the failures aren’t surprising—and more importantly, they aren’t inevitable.
The story of failure comes down to familiar themes:
- Learning gaps and inflexibility within enterprise IT and leadership.
- Misaligned priorities—chasing trendy use cases instead of focusing on practical automation opportunities.
But if these pitfalls are predictable, then so too are the paths to success. The failures aren’t a sign that AI doesn’t work—they’re a signal that too many organizations are approaching AI the wrong way.
The Pitfalls (and How to Avoid Them)
1. Learning Gaps and Inflexibility
Technology is racing ahead, but enterprise readiness often lags. Leaders know they “need AI” but aren’t sure how to adapt systems, processes, or mindsets to make it work.
The fix? Invest in data literacy and governance. There is no AI strategy without a data strategy. If your team doesn’t understand the data they have, how it’s governed, or why it matters, AI will remain a shiny object with little impact.
We help organizations start from where they are: uncovering what’s possible with the data they already have and reinforcing that good governance is the bedrock of AI success.
2. Misaligned Use Cases
It’s tempting to chase the hype—writing marketing copy, experimenting with chatbots, or running pilots that showcase “cool” but low-value applications.
But lasting impact doesn’t come from experiments that look good on paper—it comes from building a strong operational foundation first. Back-office automation may not be glamorous, but it’s where AI has proven value. Tasks like invoice processing, logistics optimization, and compliance reporting drive real cost savings and efficiencies—and those efficiencies free up field teams to focus on generating revenue.
Leaders have at times dismissed back-office automation as unimportant—but that mindset ignores the ripple effect: when the back office is efficient, the front lines thrive.
The Success Formula: How the 5% Win
The top 5% of GenAI projects don’t succeed by chance—they succeed because they follow a clear, disciplined playbook. Here’s what sets them apart: Start with a problem worth solving.
Ask: What decision do we need to make? What problem are we trying to solve? What story do the data patterns tell us?
Treat AI as a tool, not the solution itself.
The 5% don’t use AI just because it’s new or exciting—they use it where it directly addresses a problem. This keeps projects grounded, budgets on track, and adoption high because users see clear value.
Define clear goals and a plan to measure impact.
If you can’t articulate what success looks like, AI can’t deliver it.
Leverage best-in-class partners.
The most successful teams don’t try to build everything in-house. They tap proven enterprise tools from OpenAI’s enterprise offerings to predictive modeling platforms like DataRobot—and trusted implementation partners to reduce risk and accelerate time-to-value.
Partnership accelerates success—and frees you to focus on your core competitive advantage.
From Failure to Opportunity
The 95% failure rate should not discourage us. It should energize us to avoid the mistakes we already know don’t work.
With the right foundation—data literacy, governance, and pragmatic use cases—the possibilities of AI are enormous. Start with the unglamorous but impactful opportunities and let the wins build momentum toward more transformative projects.
AI isn’t a solution in search of a problem. It’s a toolset waiting to help you solve the problems that matter most. Contact SDI to explore how a strong data foundation can make your projects succeed.
.jpg)
.jpg)
.jpg)
