Decoding AI’s worth

M Muneer

Artificial Intelligence (AI) is driving enterprises to embrace it for innovation, efficiency, and competitive advantage. From retail to healthcare, manufacturing to finance, AI is reshaping how businesses operate and engage with their stakeholders. 

Yet, as organisations heavily invest in AI systems and tools, a vital question looms: how can the return on investment be justified? Measuring ROI is not merely an accounting exercise; it is a strategic necessity to ensure substantial investments deliver tangible business benefits.

AI promises transformative potential. However, without a clear evaluation framework, its promises can remain unfulfilled. Leaders must look beyond AI’s novelty and focus on its ability to drive measurable outcomes aligned with business goals.

One of the primary reasons to measure ROI is to ensure alignment with business objectives. AI is a powerful enabler, but its success hinges on its application to specific, measurable goals such as increasing revenue, reducing costs, enhancing customer experiences, or streamlining operations.

For example, in retail, AI recommendation engines aim to increase the average cart sizes or improve customer retention rates. These metrics work as benchmarks to determine an AI initiative’s value. Take Amazon’s AI-driven personalisation system, which analyses customer behaviour to suggest products, contributing to over 35% of its revenue. Similarly, in healthcare, AI-powered diagnostic tools must prove their worth by improving patient outcomes and reducing diagnostic errors.

AI integration entails substantial costs, from development and deployment to infrastructure upgrades, employee training, and maintenance. Decision-makers and stakeholders must justify these expenses with clear evidence of ROI.

For example, in manufacturing, AI-powered predictive maintenance has immense potential. By analysing equipment data to predict failures, companies reduce downtime, lower maintenance costs, and extend machinery lifespan. McKinsey reports that AI-driven predictive maintenance can cut maintenance costs by up to 25 per cent and reduce downtime by up to 15 per cent. 

Measuring ROI helps identify which AI initiatives to scale and which to discontinue, enabling better resource allocation. For instance, a customer service department might deploy an AI chatbot to handle routine inquiries. Metrics such as resolution time, customer satisfaction scores, and cost savings can determine whether to expand its use or refine its capabilities. For instance, Sephora’s virtual assistant, which provides personalised recommendations, exemplifies this approach—its measurable improvements in customer engagement and satisfaction justified scaling the project. 

AI also brings inherent risks, from under performance to ethical concerns like data privacy and bias. Enterprises need accountability frameworks to ensure AI systems meet predefined benchmarks and operate within acceptable risk parameters. In banking, for example, AI-driven fraud detection systems must balance accuracy in identifying fraudulent transactions with minimising false positives that inconvenience customers. Metrics like detection accuracy and customer impact are crucial for justifying AI use.

AI excels at automating repetitive tasks and augmenting human capabilities, boosting productivity. However, these gains must be measured to establish value. For example, marketing teams using AI tools can significantly reduce time spent on manual data analysis, allowing them to focus on better marketing strategies. PwC estimates that AI adoption could contribute up to $15.7 trillion to the global economy by 2030, with productivity gains accounting for a lion’s share of this impact. 

ROI measurement is not a one-time exercise but a continuous process that provides a baseline for improvement. Regular evaluation allows businesses to refine AI initiatives and optimise results. Walmart’s use of AI for inventory management serves as a compelling example. By using machine learning to predict demand and plan stock levels, the company reduced waste, improved product availability, and achieved cost savings while enhancing customer satisfaction.

A robust framework for measuring AI ROI includes:

Defining clear objectives tied to business goals.

Identifying key metrics such as cost savings, revenue growth, or customer satisfaction.

Accounting for all costs, including development, deployment, and maintenance.

Evaluating tangible and intangible benefits, such as increased sales and improved brand reputation.

Regularly monitoring and refining AI projects to ensure alignment with organisational objectives.

AI is hailed as a game changer, but its true power lies in its measurable impact. As the saying goes, “What gets measured gets managed.” Without a robust framework for evaluating ROI, even the most sophisticated AI initiatives risk being reduced to buzzwords.

(The writer is a start-up investor and co-founder of the non-profit Medici Institute for Innovation)

Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.