77% of Business Leaders Say AI Spending Is Stuck in Incremental Mode

Seventy-seven percent of business decision-makers — C-suite executives, directors, and managers — believe their organizations devote too much of their AI investment to incremental improvements, according to a survey commissioned by Avanade, the Microsoft-focused IT advisory provider. Only 11% expect AI to fundamentally change how their organizations operate, create value, or compete within the next year.

Four years into the mainstream AI era, the numbers describe widespread stagnation. Nearly two-thirds of respondents say AI has not yet delivered returns that justify the investment and effort, and three-quarters believe their organizations have overstated their own AI maturity. In short: most enterprises are paying for AI and not getting transformation back.

What separates confidence from change?

Ben Beath, Avanade's chief AI officer, argues the gap is structural, not technical. Many organizations and their CIOs see AI's potential but haven't embraced it, he says.

"Confidence is high, but confidence isn't the same as change," Beath said. "Having Copilot licenses and a few dozen use cases can feel to a leader like maturity, but it probably isn't if it hasn't changed the way the work is actually getting done."

The root cause, in his view, is strategic. Organizations are attaching AI to existing infrastructure rather than rebuilding around it.

"That's the product of bolting AI onto either broken or legacy processes," he said. "You just get a faster legacy process."

Many IT leaders chase tools that summarize email or generate basic reports. Those uses raise employee productivity but leave the organization's operating model untouched.

Takeaway: licenses and pilots count activity, not change.

Why do incremental projects keep winning approval?

Beath rejects the idea that leaders are unaware of the problem.

"Leaders aren't confused about this," he said. "They can see they're stuck in efficiency and productivity mode when it comes to AI. The issue isn't awareness; it's that incremental projects are easier to approve, they're easier to measure, and they're a lot less risky for whoever sponsors them."

The clients building AI into the business ask a different question. "They're saying, 'What would this process in our business look like if it was reimagined with AI?' and AI is embedded in every step of the way," Beath explained.

He also frames the problem as a leadership gap. When AI strategy sits only with the CIO, the focus stays on efficiency; when the CEO or CFO and business leaders partner with the CIO, transformation follows. His advice: don't ask where to apply AI — ask what business you would build today if you started from scratch with it.

Not every AI program is underperforming. CIO.com's State of the CIO survey found 59% of IT leaders report that some or most of their AI initiatives have met or exceeded ROI goals.

"It's not because the models are bad or because the pilots are poorly selected; it's because the hard parts of AI are where data meets the app, or where technology meets the process," Beath said.

Is it incremental versus transformational — or both?

Some practitioners reject the binary. Ed Keisling, chief AI officer at Progress Software, runs both tracks in parallel.

"I believe they go hand in hand," he said. "The incremental improvements allow organizations to begin to explore the art of the possible with AI, and in some instances, can compound across an entire workflow to bring transformational results."

Progress Software uses AI agents to synthesize updates from its collaboration, project management, and knowledge management systems into leadership briefings. Its sales teams use agents to pull data from CRM, data platforms, and customer systems to generate quotes faster and more accurately. On the transformational side, AI now handles requests for information and requests for proposals.

"These are extremely time-consuming processes that used to take days and now can be completed in minutes," Keisling said. "It allows us to be more responsive to opportunities."

Scaling remains the hard part, he notes, with data sovereignty, security, and cost holding organizations back. "There's still a big challenge in scaling AI tooling from the individual to a team, and then to a team of teams or an organization," he said. He also cautions against dismissing measurable gains: "If AI takes meaningful time out of a high-volume workflow, increases throughput, improves quality, or frees capacity, that can create significant economic value without being labeled 'transformational.'"

Can small wins build toward transformation?

Tal Carmi, CIO of digital adoption software provider WalkMe, describes a staged path. "Incremental improvements are the steppingstones to transformation," he said. "We have some areas where we started with AI-based solutions such as Copilot and after seeing the success rate, they evolved into auto pilot, transforming how work is done."

WalkMe isn't finished. Systems limitations, face-to-face processes, risk ownership, and employee enablement still hold parts of the company back.

"AI only magnifies your knowledge and data problems," Carmi said. "Telling your LLM to extract data and build a dashboard is easy, making sure it extracts the right data and processes it the right way is a different story."

Legacy systems create another roadblock. "Many of the systems people most want to automate don't have an API and are unlikely to get one anytime soon," he said. "So people improvise, and that's how shadow IT gets in."

The message for CIOs: reimagine the process, not just the tool — and bring the CEO and CFO to the table before the budget cycle does it for you.