Almost 95% of senior business leaders in the U.S. report that they are currently investing in AI, and the number of companies investing $10 million or more in it is expected to rise from 16% last year to 30% this year. But are they following the foundational functions for AI to thrive?

As businesses continue to integrate artificial intelligence into their operations, the focus is shifting from exploring its capabilities to understanding its impact. Indeed, AI promises unparalleled efficiency, enhanced productivity, and revolutionary customer experiences, but the critical question now is: “How do we measure the return on investment (ROI) of AI initiatives?”

The need to measure AI’s ROI is not merely an academic exercise; it is a vital business imperative. Remember the adage, what’s measured gets done; what’s expected doesn’t? Organisations worldwide are committing substantial resources into AI technologies, and the pressure to justify these expenditures is mounting—perhaps not as yet in India because of its slow adaption here. The journey to quantifying AI’s impact is not without complexities. Its benefits are many and probably intangible in the short term, making traditional ROI metrics inadequate—not very dissimilar to measuring the ROI of cultural change, for instance.

The stakes are much higher. According to Gartner, up to 30% of GenAI projects may falter, failing to progress beyond the proof-of-concept stage—a sobering statistic that reflects the challenges of deploying AI effectively. On the flip side, evidence of AI’s potential is compelling. The EY AI Pulse USA Survey reveals that nearly 75% of companies leveraging AI report measurable ROI. Key areas of impact? Operational efficiency (77%), employee productivity (74%), and customer satisfaction (72%). This conflict necessitates the importance of refined approaches to measuring ROI.

To start with, it’s crucial to recognise that AI initiatives do not conform to traditional models. Unlike straightforward financial investments, where ROI is directly tied to revenue or cost savings, the value AI delivers is indirect, multidimensional, and long-term. For example, productivity gains are not one-size-fits-all. AI’s impact on productivity varies across functions and industries. For software engineers, it might mean a higher number of features delivered or reduced debugging time. For human resources, it could involve streamlining the hiring process, reducing time-to-fill metrics, or enhancing candidate experience. Establishing a baseline productivity level before deployment and monitoring changes post-implementation provides a clear picture of AI’s contribution. AI accelerates decision-making by providing faster access to insights, a factor that directly translates into time and cost savings.

Operational efficiency is another dimension. The ability of AI to automate routine, time-intensive tasks is one of its most established advantages. Tasks like data entry, invoice processing, and quality control burden human resources with wasted time and inefficiency. AI eliminates these bottlenecks. For instance, Robotic Process Automation in financial operations has been shown to reduce manual labour costs by as much as 25%, according to McKinsey. In manufacturing, predictive maintenance powered by AI is transforming operational strategies. Companies like GE employ AI models to predict equipment failures, enabling timely interventions and significantly reducing downtime. This avoids costly disruptions and extends the lifespan of machinery, creating substantial long-term value.

AI’s influence extends beyond internal operations to customer-facing interactions. Metrics such as Net Promoter Scores (NPS) or customer satisfaction surveys capture the impact of AI on user experiences. When deployed effectively, it enhances personalisation, responsiveness, and overall service quality. Amazon’s AI-powered recommendation engine is a case in point. The personalisation strategy has been credited with driving 35% of Amazon’s revenue. Similarly, AI chatbots are revolutionising customer service by handling routine inquiries, reducing wait times, and offering instant resolutions. According to Juniper Research, these chatbots are expected to save businesses over $11 billion annually, demonstrating their role in enhancing customer satisfaction and reducing operational costs.

Revenue generation, perhaps the most direct and tangible measure of ROI, is another area where AI can shine. Dynamic pricing models illustrate AI’s capacity to drive sales and optimise revenue. Travel portals and airlines like Emirates use AI to dynamically adjust prices based on real-time demand, local events, and historical trends. This maximises earnings while ensuring competitive pricing for customers. Similarly, streaming platforms like Netflix rely on AI-powered recommendation systems to keep users engaged. An astounding 80% of the content streamed on Netflix is based on such recommendations—data-driven personalisation drives user retention and loyalty.

Measuring ROI for AI initiatives should begin with pilot projects. These small-scale implementations serve as controlled environments to test AI’s impact and gather actionable insights. For instance, deploying chatbots in one area or testing predictive maintenance in a single factory will enable evaluation of effectiveness, address potential challenges, and refine strategies before scaling up. This phased approach mitigates risks and ensures that resources are channelled into initiatives with proven value.

However, achieving meaningful ROI requires patience and a strategic perspective. The timeline for seeing returns varies based on the complexity of the project. Simple automation solutions, such as chatbots or data processing tools, might yield measurable benefits within three to six months. More advanced applications, like predictive analytics or dynamic pricing systems, take six months to one year to demonstrate significant value. Large-scale transformations, such as AI-driven supply chain optimisation, may require a longer horizon, exceeding 30 months or so. During this period, continuous monitoring and iterations are essential to align AI initiatives with overall strategy.

Despite the allure of quick wins, CEOs must embrace a long-term view of AI’s potential. It is not merely a tool for incremental improvement; but a strategic asset that can redefine businesses and industries. See a parallel between AI and foundational technologies like email or electricity. Just as these became indispensable to businesses and human race, AI will become integral to the fabric of businesses. Organisations that focus on creating long-term value, rather than seeking immediate financial returns, are better positioned to gain a sustained competitive advantage.

Ultimately, measuring ROI of AI requires a nuanced approach that transcends traditional financial metrics. It demands a comprehensive evaluation of operational efficiencies, enhanced productivity, improved customer satisfaction, and long-term strategic impact. By starting with pilot projects, establishing clear benchmarks, and adopting a holistic perspective, businesses can unlock AI’s transformative power. In doing so, they not only ensure agility and competitiveness in a data-driven world but also pave the way for sustainable growth and innovation.

Enterprises that master the art of AI’s ROI won’t just survive the data-driven era, but will thrive in it. After all, AI doesn’t just pay for itself; it reinvents the wallet!

(The author is a Fortune 500 advisor, start-up investor and co-founder of the non-profit Medici Institute for Innovation. Views are personal.)