ROI is about more than profitability when it comes to AI adoption – here’s what enterprises are looking for

AI doesn't have to deliver a traditional return on investment (ROI) in order to be worthwhile, according to a KPMG survey.

While ROI is often defined by the financial benefits delivered by the technology, enterprises are measuring returns across a broader range of areas, the consultancy noted.

Notably, the survey found more than three-quarters (76%) of respondents were more confident in their ability to measure ROI in terms of productivity gains. Other key metrics considered by firms include performance and quality of work (71%) and the speed and accuracy of decision making (67%).

Profitability, meanwhile, was cited by 64% of respondents as a key factor in measuring ROI.

When it comes to more strategic or indirect benefits, confidence drops sharply. Just 14% said they were confident in measuring ROI from improved analytics used for business decision‑making.

All told, around two-thirds (65%) of respondents said their organisation will continue to invest in AI regardless of their ability to measure tangible returns.

"This shift in mindset by business leaders from viewing AI as something that must deliver an immediate return to one that sees AI as a long-term investment, recognizing it as a strategic enabler for enterprise‑wide transformation, is an important milestone," said Leanne Allen, head of AI at KPMG UK.

"To get real value, businesses need to be intentional about where and how they deploy it, ensuring it supports their business goals and enables new ways of working. The organizations that take this more thoughtful, enterprise‑led approach will be the ones who fully realise the benefits of the technology.”

Shifting the goalposts or changing mindsets?

This shift in mindset highlighted by Allen comes amid a backdrop of growing concerns about ROI in recent months.

A survey from PwC in January found many executives have grown frustrated at the lack of returns, although this is often measured in financial terms.

Just one-in-eight CEOs told the consultancy that AI has delivered both cost and revenue benefits, for example.

Separate research from Forrester this month also found AI isn’t having the desired impact that many had hoped. The consultancy warned that this was due to a “lack of understanding” of the technology, however, rather than the technology’s flaws.

Forrester noted that enterprises recording success with AI are taking a more “deliberate” approach to adoption, focusing primarily on delivering value for customers instead of immediate financial, productivity, and efficiency gains.

Still a long way to go for IT leaders

KPMG’s survey found IT leaders still face an array of challenges when it comes to unlocking returns on AI investment.

Almost half (46%) said skills gaps were a key hurdle, while risk considerations such as data privacy and cybersecurity were also a leading barrier to demonstrating AI‑related ROI.

The second-biggest factor was difficulty quantifying indirect or long‑term benefits, cited by 40%.

Indeed, the main challenges to AI strategy in the next 12 months were risk management such as data privacy and cybersecurity (41%), followed by the quality of organizational data and employee adoption (32%).

Despite these concerns, 58% of respondents said their organizations were planning to invest more than $50 million in AI over the next year, with half of these investing more than $100 million.

70% of UK business leaders also said that AI would remain a top priority for investment, even if a recession occurs.

Different stages of adoption

Organizations are at different stages when it comes to AI agent implementation, KPMG found, with many focusing on integration in a wide range of domains.

Amongst the 94% either using or planning on using AI agents, 19% are scaling them across multiple functions, 13% are developing or implementing multi-agent or agentic AI systems, and 8% are orchestrating multiple AI agents across workflows.

In terms of security, 48% are looking at AI agents as augmented support for employees and providing training to their workforce. Security considerations are key here, respondents noted.

More than one-third (39%) are taking a human-in-the-loop approach while 37% aren't allowing AI agents to access sensitive data without human oversight.

"Whatever form of AI an organization is using or planning to use, it is critical to have the right guardrails in place to minimize risks — whether that’s accuracy issues, unreliable model outputs or the behavioural risks that come from how people use the technology," said Allen.

"It’s not only about ensuring the technology can be trusted, but also the people operating it. Building in appropriate, proportionate controls from the start — through design, development, deployment and continuous monitoring in production — will lead to better, safer outcomes.”

Skills in the spotlight

Just over six-in-ten are either already upskilling their current workforce or are planning to, 52% are recruiting prompt engineers or AI architects and 48% are redesigning job roles.

Only 15% said they weren't confident that their current talent pipeline was good enough to meet the needs of an AI-enabled workforce.

“As AI changes how work gets done and as human and AI agents increasingly collaborate, job roles themselves are being reshaped. That means the fundamental skills required include adaptability, critical thinking, judgment, and the ability to work effectively with AI systems, said Allen.

"Many of these skills already exist within the workforce and can be unlocked with the right training, role redesign and learning opportunities.”

FOLLOW US ON SOCIAL MEDIA

Follow ITPro on Google News and add us as a preferred source to keep tabs on all our latest news, analysis, views, and reviews.

You can also follow ITPro on LinkedIn, X, Facebook, and BlueSky.