Succeeding with AI initiatives

Derek Ashmore at Asperitas explains why so many enterprise AI initiatives fail, and how CIOs can flip the script

 

AI has captured the imagination of business and technology leaders alike. The promise is compelling: greater efficiency, new sources of insight, better customer experiences, and entirely new business models. Yet, for all the investment and hype, most enterprise AI initiatives still fail to deliver meaningful return on investment (ROI).

 

The reasons are not mysterious. AI’s failures follow familiar patterns of misaligned goals, flawed execution, and cultural resistance. But the good news is that CIOs and IT leaders can break the cycle. By tying AI directly to business outcomes, building realistic roadmaps, and reshaping organisational processes, enterprises can transform AI from a costly experiment into a sustainable driver of value.

 

 

Why AI initiatives miss the mark

AI projects often stumble before they even get off the ground because expectations are not aligned with business outcomes. Business leaders may chase AI because it is trendy, while technology groups pursue the latest model architecture. Without deliberate communication and alignment, projects become solutions in search of problems.

 

Data quality is another persistent obstacle. AI depends on clean, complete, and well-documented data. Yet many organisations discover too late that their data is inconsistent, biased, or missing critical context. As a result, models produce unreliable results that erode trust.

 

Timelines also doom many projects. Executives often expect AI to deliver transformative results in months, when in reality, unanticipated challenges inevitably extend development and deployment. When results don’t arrive on schedule, enthusiasm wanes and support dries up.

 

Skills shortages add another layer of difficulty. AI projects require both technical AI expertise and deep domain knowledge. Many teams lack the right blend, leading to mistakes, rework, and stalled progress. Compounding the issue, organisations frequently attempt “big bang” releases that overwhelm teams with complexity instead of starting with smaller, incremental wins.

 

 

The leadership practices that drive AI ROI

Successful AI adoption starts with leadership. CIOs and technology executives must treat AI as a business investment rather than a science experiment. That means tying AI features directly to business ROI outcomes such as cost savings, revenue growth, or risk reduction.

 

Measurement and optimisation are equally important. Success cannot be defined only in technical terms like accuracy or latency. True ROI requires metrics that track user adoption, operational efficiency, and financial performance. CIOs should set up dashboards that provide this visibility from the start.

 

Leaders also need to set realistic expectations. Overpromising undermines trust; transparency builds it. Communicating that AI adoption will be iterative and cumulative helps the business understand both the risks and the rewards. Delivering small, incremental changes keeps risk manageable while steadily building organisational mastery.

 

Finally, leaders must approach AI with a portfolio mindset. Not every investment will pay off, and that is acceptable. AI should be treated like a venture portfolio, where multiple bets are placed, some fail fast, and others scale dramatically.

 

 

Building the right talent and culture

AI success hinges as much on people as it does on technology. CIOs must invest in both recruiting new talent and upskilling existing employees. Continuous learning programs keep practitioners current with fast-moving AI platforms and tools, while also giving them confidence to apply new methods to real problems.

 

Equally critical is rethinking business processes. AI enables capabilities that old processes were never designed for. CIOs should encourage teams to refactor workflows to capitalise on AI, not simply bolt AI onto existing methods. Incentivising adoption and process improvements ensures that AI isn’t seen as a burden but as an enabler.

 

Resistance and roadblocks are inevitable, so organisations need mechanisms to address them quickly. One effective approach is to establish AI “SWAT teams” that specialise in facilitating adoption, troubleshooting issues, and clearing obstacles. This keeps projects moving forward and prevents momentum from stalling.

 

 

Adapting organisational processes for AI

AI doesn’t just augment existing processes; it requires organisations to rethink them. Decision-making, for example, can no longer rely exclusively on human judgment. The most effective enterprises combine human expertise with AI recommendations, creating a collaborative decision framework.

 

The traditional IT project approach of ad hoc experimentation must give way to a portfolio-based model. AI initiatives should be tracked like investments, with performance metrics guiding whether projects are scaled, adjusted, or retired.

 

Governance also needs to evolve. Hierarchical approval chains are too slow for AI-driven insights, which often demand real-time responses. Instead, organisations should establish lightweight guardrails that allow decisions to be made quickly while maintaining appropriate oversight.

 

 

Monitoring and measuring AI progress

CIOs cannot simply launch AI projects and hope for the best; they must monitor and measure rigorously. This begins by defining success upfront, with benchmarks that encompass both technical and business objectives. Baseline measurements should be taken before projects launch so impact can be measured accurately.

 

Embedding measurement into governance is key. Dashboards with predictive indicators should be integrated into leadership’s regular performance review process. Tracking “time-to-value” milestones provides visibility into how quickly AI initiatives are generating ROI.

 

Measurement should not just report on progress but also drive continuous improvement. Regular review cycles allow leaders to refine AI strategies, reallocate resources, and identify opportunities for scaling. The best feedback loops are dynamic, feeding directly into strategy and execution.

 

 

Avoiding the pitfalls

The same mistakes repeat across enterprises, and CIOs can avoid them by learning from others. Scaling too quickly before proving value in a pilot is one of the most common errors. Security and privacy must also be built in from the start, not treated as afterthoughts.

 

Chasing hype is another trap. AI should never be adopted for its own sake; it must solve a business problem. Likewise, organisations that measure only technical KPIs without linking them to business outcomes miss the point. Accuracy or precision metrics matter only insofar as they contribute to savings, growth, or risk reduction.

 

Perfectionism is another killer of ROI. In many cases, off-the-shelf solutions aligned to business goals can deliver faster value than custom-built systems. Striving for technical elegance at the expense of business value delays results and increases risk.

 

 

From failure to value creation

AI is not destined to fail in the enterprise. The organisations that succeed are those that combine ambition with discipline. They tie AI directly to business outcomes, align expectations with reality, build talent pipelines, and adapt organisational processes to the speed and demands of intelligent systems.

 

For CIOs, the challenge is not just technological but cultural and strategic. Success requires reshaping how organisations think about investment, governance, and decision-making. It demands building measurement into the DNA of projects and creating feedback loops that drive continuous improvement.

 

The enterprises that embrace these practices will move beyond stalled pilots and wasted budgets. They will position AI not as an experimental playground but as a core driver of efficiency, innovation, and competitive advantage. In short, they will flip the script from AI failure to AI value creation. 

 

---

 

Derek Ashmore is AI Enablement Principal at Asperitas

 

Main image courtesy of iStockPhoto.com and BlackJack3D