An AI pilot cuts a recurring task from three hours to one. Employees like it. The demonstration goes well. Leadership calls the pilot successful.

Then the CFO asks what changed financially.

Often, the answer is unclear.

That gap is showing up across organizations. PwC’s 2026 Global CEO Survey, covering 4,454 CEOs, found that only 30% reported increased revenue from AI during the previous 12 months, while 26% reported lower costs. More than half saw neither benefit.

The problem is partly how organizations measure AI ROI. Time saved is an operational result. It is not automatically a financial return.

Measure what happens after the time saving

Suppose an accounts-payable team uses AI to reduce invoice-processing time by 30%. If staffing costs remain unchanged, transaction volumes stay flat and the additional capacity goes unused, the P&L may barely move.

The pilot still worked. The organization simply failed to convert productivity into value.

That distinction should shape the business case before implementation.

Instead of measuring only hours saved, finance should establish a baseline and connect the AI use case to an outcome the business already measures: lower processing cost, increased capacity without additional hiring, faster collections, fewer errors, improved conversion, reduced external spending or additional revenue.

This is why BLS treats AI integration and process improvement as an operating-model issue rather than a software exercise. The workflow around the technology determines whether productivity becomes financial value.

A finance-grade approach to AI ROI

A useful AI business case starts with four numbers:

Baseline: What does the process cost today, including labour, software, external services and rework?

Investment: What will AI actually cost once licences, implementation, integration, training, governance and ongoing oversight are included?

Operational change: What measurable improvement occurred — processing time, throughput, error rate, cycle time or capacity?

Financial conversion: How does that improvement affect revenue, cost, working capital or avoided future spending?

That final step is where many pilots stop too early.

BCG’s recent research similarly identifies a disconnect between AI initiatives and P&L accountability, with only 14% of surveyed CEOs saying their organizations clearly define P&L impact for all AI initiatives. BCG argues that scaling value requires workflow redesign and clear business accountability, not simply more pilots.

Finance should be involved before the pilot starts

If the CFO is asked to calculate AI ROI after implementation, measurement has already become harder.

Finance should help establish the baseline, expected value and measurement period before the pilot begins. Operations needs to determine what will change in the workflow. Technology needs to establish what implementation requires.

That cross-functional discipline is central to how BLS Consulting approaches organizational AI adoption.

Before approving another AI pilot, leadership should be able to complete one sentence:

If this works, we expect it to improve ________, and we will measure that by ________.

If the blanks cannot be filled with something the business already values, the pilot may demonstrate that the technology works. It has not yet demonstrated a business case.