Beyond AI pilots: What boards should be measuring

Over the past two years, most boardrooms I engage with have heard some version of the same update. The number of artificial intelligence (AI) pilots keeps growing, and so do the productivity gains being reported. These are real achievements. However, they do not fully answer the central question for boards. Is AI helping the business perform better?

The State of AI in the Enterprise, 2026, a recent report by Deloitte, shows how wide this gap is. Two-thirds of organisations surveyed report productivity and efficiency gains from AI, and four in ten report lower costs. Only one in five report revenue growth, even though nearly three in four expect AI to grow their revenue.

For boards, this raises a practical question: how should AI success be measured?


Productivity, return on investment, and business outcomes each have a role. Productivity is an early signal that something is working, and return on investment indicates whether the spend is paying back. Business outcomes take longer to show, but they are what tell a board whether AI is strengthening the enterprise’s performance. Boards need all three, at different stages.

Getting there requires three shifts: a business case that can be proven, a redesign of how work gets done, and management systems that hold leaders accountable for the results.

Start with a business case that can be proven

Every significant AI investment should start with an outcome leadership already cares about, such as better margins or stronger customer retention. If the case cannot be expressed in these terms, it is not yet a business case.

In a consumer products company using AI for demand planning, forecasts may become more accurate, and planners may spend less time on routine work. That is useful, and it is only the first step. The investment should be judged on what follows from it: less inventory and better margins.

Each initiative needs a baseline, clear targets, an honest view of cost, and one accountable business owner. It also needs to identify the operational changes that will deliver the benefit, because a benefit with no change attached to it is unlikely to be realised.

The evidence should mature with the initiative. Early experiments prove feasibility, production deployments show operational improvement, and at an enterprise scale the financial KPIs must move. It is reasonable to expect less from a pilot than from a programme running at scale, as long as the bar rises at each stage.

Redesign how the work gets done

AI gives organisations an opportunity to rethink how work is done. Much of the value comes from how well those changes are embedded in the business.

An insurer’s claims operation shows why. AI can help with document processing, fraud detection, and claims assessment, and each brings some improvement on its own. Brought together across the full claims journey, they can reduce settlement times and cost per claim. That requires changes to how claims handlers work and who makes which decisions.

Deloitte’s 2026 Global Human Capital Trends research found that organisations leading in the deliberate design of how people and AI work together were nearly 2.5 times more likely to report better financial results.

This is where many programmes fall short. Employees need to understand how their work will change, and leaders need to keep reinforcing the new ways of working until they become routine.

Build AI into how the business is managed

This is perhaps the most important shift. AI outcomes must become part of how the company sets targets and reviews performance.

An initiative expected to improve service profitability should show up in margin targets and delivery reviews. Gains in sales effectiveness should be visible in conversion rates and revenue forecasts, and any capacity freed up in delivery should change hiring plans. If none of these numbers moves, the board should ask why.

Boards should ask management to establish a value realisation office, sponsored at the executive level, that brings together the business with finance, technology, HR, and risk. Its role is to track outcomes and make sure operational gains turn into financial results. Ownership must still remain with business leaders, with the other functions supporting them and finance in particular validating the economics.

Board reviews should separate projected benefits from realised outcomes and support clear decisions on whether to scale, redesign, or stop an initiative.

Not every initiative will generate returns within the same period. Some will deliver near-term financial benefits, while others will open up new products or business models over time. Each needs its own measures and time horizon, with the same level of accountability.

Organisations that do this consistently, turning technology investments into stronger performance year after year, build a lasting apex advantage.

The real test of AI is whether the enterprise performs better than it did before.

The author is a Partner at Deloitte India. Views are personal.