Survey: 59% of Firms Deploy AI But Only 35% See ROI

A new global survey has put hard numbers on something many executives have quietly suspected: most companies are using AI, but very few are making real money from it.

The study, conducted by Harvard Business Review Analytic Services and sponsored by process automation firm Appian, surveyed 385 business decision-makers across North America, Europe and Asia-Pacific. The results are a reality check for boardrooms that have poured resources into AI over the past two years.

59% of organizations already have AI running in production. Yet only 35% report improvements in ROI, and just 30% say they've seen AI create new revenue streams. The gap between "AI is deployed" and "AI is working" is wide.

The Productivity Trap

Most companies are measuring the wrong things. 64% say AI has improved productivity, and 58% report better operational efficiency. Those are real wins. But they're not the same as business growth.

MIT Report: 95% of Enterprise AI Pilots Fail to Deliver ROI

MIT research shows 95% of enterprise AI pilots fail ROI. Asian organizations see wider gaps between productivity gains and actual business impact.

Mission MediaMission Media

Think of it this way: making your team faster at their current tasks is useful, but it's different from finding a new customer, winning a new contract, or opening a new market. Right now, most AI deployments are achieving the first thing, not the second.

Only 16% of organizations say they've realized a high degree of measurable value from AI. Another 36% say the value has been slight. 8% say there's been no measurable value at all.

Despite this, 86% of respondents say they still expect more from their AI investments. That expectation gap, between what's been promised and what's actually delivering, is the defining tension in enterprise technology right now.

The Integration Problem

The survey reveals a clear dividing line between companies that are getting results and those that aren't. It comes down to where AI sits in the organization.

Only 18% of companies have AI primarily integrated into their core workflows. The majority (34%) use it as a standalone tool sitting alongside their processes, rather than inside them. Another 34% use a mix of both approaches.

That positioning makes an enormous difference. Among organizations that have embedded AI directly into their core processes, 71% report substantial or moderate value. Contrast that with the broader average of just 16% reporting high measurable value.

"Organisations are adopting AI, but many haven't integrated it into the core processes that drive business outcomes," said Alex Clemente, Managing Director of Harvard Business Review Analytic Services. "Those that successfully embed AI into workflows will be better positioned to realise meaningful value."

Matt Calkins, CEO of Appian, framed the stakes even more directly: "The true potential of AI can only be realised when it moves from a standalone tool to an embedded worker that drives revenue."

Legacy Infrastructure Is Blocking Progress

Across Asia-Pacific, the barriers are compounded by older technology systems. 69% of survey respondents say legacy systems are limiting their ability to scale AI across their organizations. IDC research identifies legacy architecture as the primary blocker, with 43% of APAC firms specifically hindered by it.

AI Isn't the Bottleneck. Your Organization Is.

Organizations failing with AI are missing organizational alignment, not tools. Top performers using the SRM Maturity Index generate 73% revenue growth.

Mission MediaMission Media

Other barriers cited in the survey: siloed or poor-quality data (34%), lack of integration across systems (31%), and a shortage of AI skills (30%). In a region as technologically diverse as Asia-Pacific, from Singapore's digital-native firms to manufacturers in Southeast Asia running decade-old systems, there's no single playbook that fits every organization.

PwC's 2026 AI Performance Study finds that 74% of AI's economic value is being captured by just 20% of organizations. The rest are getting left behind.

The Governance Gap Nobody Is Talking About

As more companies start exploring AI agents (software that can take actions on its own, not just answer questions), a new problem is emerging. 25% of organizations are already using agentic AI, and 62% are considering it. But the guardrails aren't keeping pace.

92% of respondents agree that AI agents need rules-based controls to operate safely. Yet only 48% of those already using or exploring agentic AI have actually defined those rules. Deloitte finds only 21% of organizations have a mature governance model for agentic AI.

For industries where reputation is everything (communications, PR, financial services), deploying autonomous AI without proper guardrails isn't a technology risk. It's a brand risk.

What the Leaders Are Doing Differently

The survey also points to where organizations are focusing to close the gap. Half of respondents say they're working on better defining the rules that AI must follow. 49% are standardizing processes and workflows across functions. 47% are improving coordination between departments.

The $11B Market Betting on AI Orchestration Infrastructure

Orkes raises $60M Series B for AI workflow orchestration platform. Only 6% of enterprises see AI ROI; 89% of AI agents never reach production.

Mission MediaMission Media

Gartner confirmed in April 2026 that AI projects in infrastructure and operations are stalling before delivering meaningful ROI. The pattern is consistent across virtually every major research firm.

The organizations pulling ahead aren't necessarily spending more. They're integrating more deeply, governing more carefully, and measuring outcomes that actually connect to revenue rather than just activity.

The question for every executive in 2026 isn't whether to invest in AI. It's whether you've built the business conditions that let AI actually perform.