Finance AI returns depend on more than deployment

Finance AI returns depend on operating capability

Finance leaders seeking better decisions and more accurate forecasts need to assess whether their organizations can turn AI use into financial performance. Deploying the technology gives people access; preparing staff and workflows helps them use it effectively. For chief information officers (CIOs) and technology leaders, the investment decision therefore includes how the organization will govern AI, measure its contribution and develop the people using it.

Data quality and workforce capability need different responses

Preparing for that work means addressing organizational barriers, because skills gaps often pose greater obstacles than the technology itself. Nikki McAllen, global head of finance advisory of KPMG Australia, said data and talent had been identified among financial services’ main AI barriers two years earlier, and both remained problems. KPMG provides advisory services, so it has a commercial interest in businesses seeking help with these challenges. McAllen’s assessment puts management’s response at the center of readiness.

Leaders need a different response for each barrier. “Functions that will likely pull ahead are the ones that recognize these as distinct problems requiring distinct responses,” McAllen said. That means assessing the quality of data used in financial work separately from the abilities people need to apply AI effectively. Grouping both under a broad readiness label can obscure which problem an investment addresses.

This distinction also changes what leaders should expect purchases and training programs to achieve. McAllen said: “Data quality is not a technology fix. Workforce capability is not a training plan.” Her point makes both continuing management responsibilities. A deployment plan needs to address each, because unresolved problems in either can prevent finance teams from realizing AI’s advantages.

Performance findings connect AI choices with business outcomes

With those barriers in view, KPMG’s findings suggest leaders should examine where they expect AI to create value. KPMG found stronger performance in work requiring substantial judgment than in applications focused on reducing expenses. The comparison gives finance leaders a reason to assess whether AI improves decisions and forecasts alongside whether it reduces the resources needed to complete an activity.

The expected improvement determines what to measure. An expense-reduction application calls for assessing resource use, while a judgment-intensive application calls for assessing the decisions or forecasts it supports. Both concern financial performance, but they ask different questions of the technology. KPMG’s comparison therefore supports giving judgment-intensive work a clear place in the investment case.

The type of AI also matters to that case. Organizations deploying agentic AI, systems that can pursue an objective through multiple actions, reported stronger forecasting accuracy and return on investment, or ROI, in KPMG’s findings. For finance and technology leaders, this makes system selection part of deciding how to achieve the intended result. The choice belongs alongside an assessment of the organization’s ability to use it effectively.

Organizational oversight also featured prominently in KPMG’s research. Its spring research found that data security, privacy and risk were leading influences on AI strategy for nearly all finance leaders for that year. These concerns shape how organizations handle information and manage AI use, so governance belongs within investment and implementation decisions. It lets leaders assign responsibilities and oversee how the technology enters financial work.

KPMG’s third-quarter report connected more developed governance with significantly stronger reported outcomes than peers, in some cases “three to six times better.” The range describes reported outcomes, so its meaning depends on the outcome being compared. It is not a universal multiplier for ROI. Even with that limit, the reported gap gives finance leaders a financial reason to treat governance as part of AI execution.

Put business value, governance and measurement inside execution

To turn these performance considerations into action, KPMG’s third-quarter report identifies four top priorities: define AI’s business value, make governance an execution priority, incorporate measurement into implementation and shape the workforce beyond technical instruction. These responsibilities work together rather than forming a prescribed sequence. They connect the result finance wants with the oversight, assessment and people needed to pursue it.

Defining business value gives the other priorities a purpose. Leaders need to specify the financial contribution expected from AI-enabled work, whether that means better decisions, more accurate forecasts or improved financial performance. They can then connect the activity being introduced to the result it should support, making the intended business gain an explicit part of implementation.

Governance helps establish the conditions for achieving that gain. McAllen connects trust directly to performance: “Trust, embedded in how performance gets built, is what separates the organizations capturing value from the rest,” she said. Her argument places responsibility for trustworthy use within the work of producing financial results, so governance continues as AI becomes part of finance operations.

Measurement lets leaders assess those results during implementation. A deployment milestone records progress in introducing the technology, while an outcome measure assesses its contribution to the business goal. Building that assessment into implementation connects execution with the value defined at the start. Leaders can then judge whether AI use is delivering the intended benefit.

Workforce preparation determines whether people can apply the technology within those arrangements. KPMG describes successful finance teams as establishing governance, measurement and workforce abilities as enabling conditions for AI use. For technology leaders, that means preparing staff and workflows alongside the systems being introduced. The workforce priority therefore requires a broader view of capability than instruction in operating a tool.

Build workforce capability beyond technical training

PwC’s April report makes that broader responsibility more concrete. It found that technical proficiency is critical, while CIOs must also foster human-centered capabilities to advance AI within their organizations. PwC, which also provides advisory services, has a commercial interest in businesses investing in support for AI and workforce change. Its report identified coaching, agility, judgment and empathy as capabilities relevant to that work.

These four capabilities concern how people apply technical knowledge as their responsibilities change. Coaching means helping others develop, while agility means responding to change. Judgment involves assessing what a situation requires, and empathy involves understanding other people’s perspectives. For CIOs planning workforce development, these examples define a practical task: build technical proficiency alongside the human abilities people need to use it effectively.

Key executive takeaways

• Separate data and workforce readiness: Finance and technology leaders need distinct plans for data quality and staff capability. Assess each barrier separately so investments address the problem limiting AI performance.
• Target measurable financial gains: KPMG found stronger reported performance in judgment-intensive work than in expense-reduction applications. Finance leaders evaluating AI investments can use decision quality and forecasting accuracy alongside cost savings to define expected value.
• Build governance and measurement into execution: Finance leaders need to define AI’s business value, assign oversight responsibilities and measure outcomes during implementation. KPMG linked more developed governance with stronger reported outcomes, reinforcing its role in capturing value.
• Develop human capabilities alongside technical skills: CIOs need workforce plans that build coaching, agility, judgment and empathy alongside technical proficiency. These capabilities help staff apply AI effectively as workflows and responsibilities change.