AI ROI depends on people, not tech

Like many great business ideas, it began as some scribbles on a napkin. 

At a Cambridge, Massachusetts, cafe in 2019, Laks Srinivasan, a former COO with decades of experience in data and analytics, and Tom Davenport, bestselling author, professor, and enterprise technology expert, mapped out a plan for a new research-driven organization with a singular mission: to help companies convert their AI investments into long-term, bottom-line results. 

Five years later, it’s no surprise why their growing advisory firm, the Return on AI Institute, is in heavy demand. After two frenetic years experimenting with generative AI, companies are shifting from experimentation to execution and maximizing the business value of AI tools and applications. But most companies remain at the beginning of that learning curve. 

Many are asking the same question that Srinivasan and Davenport mulled at that cafe: Why do only a sliver of businesses—5-10%, by Srinivasan’s reckoning—manage to extract meaningful value from AI investment, while the vast majority see their AI pilots never take off (or, worse, crash and burn)?

One answer, gleaned from extensive research and interviews the firm conducted with leaders of more than 100 global businesses, was that the elite few didn’t solely outsource AI to outside consultants and vendors. Instead, they nurtured internal capabilities and talent within their organizations to generate business value from AI and deployed company-wide strategies to use the technology to solve business problems. 

“It’s not about how much data you have or how big a model you build,” Srinivasan says. “It’s about human behavior. How do you change the company, the way they think, the way they behave? That’s how to leverage AI for meaningful value.”

The human factor

Human, not technological, transformation drives the RoAI Institute’s methodology in working with organizations. The framework includes a couple of key pillars: 

• AI readiness. First, they assess the company’s core capabilities and readiness for AI, from talent to infrastructure. Is there strategic alignment across leadership, talent, and culture? What return governance models do they have? Do they have sound data and enterprise technology infrastructure?

• Boosting organizational knowledge. Based on those insights, the institute maps out a 12- to 18-month advisory program to accelerate progress. It includes primer sessions on what AI is and isn’t, workshops, and individual coaching. 

“The goal is forming consistent habits,” says Srinivasan. “We give them a foundation of knowledge about AI, plus ongoing simulations and social learning to make sure these behaviors stick at the individual and organizational level.”

Over five years, nearly 1,000 business leaders from the healthcare and life sciences, financial services, industrial manufacturing, and technology sectors have gone through the institute’s AI primer sessions. Those who have benefited the most, Srinivasan adds, end up reaping the greatest ROI from AI. And they are not just open to, but rather ready for the organizational and strategic changes that AI requires. 

“They already understand how important AI is,” he says. “They don’t know exactly what it is, but they know it’s important, and they’re afraid of getting left behind. They have what I call a ‘transformation trigger,’ meaning the urgency is there for change. If you don’t have that, it’s very hard. It doesn’t matter whether you’re a small company, a big company, even an industry leader. If you don’t transform, you’re going to become Kodak.”

In the end, ROI comes back to people, not technology, Srinivasan says. “I’ve known companies that have invested huge sums in data warehouses and data infrastructure, and after five years they’re saying, ‘Where’s the beef? Where’s the return?’ They’re stuck because they haven’t put in the mental work.”

Define outcomes first, not last

Generative AI has infused urgency into companies’ AI investments because of the massive hype around it, by competition, and direct pressure from boards or their CEOs. 

“Executives and leaders say AI is life or death for them,” Srinivasan notes, “but as important as CEOs say AI is, I don’t see it getting the airtime and management focus it needs.” 

Instead, when it comes to setting AI strategy and prioritizing use cases, too many CEOs offload the responsibility to others in the organization, letting data science or AI teams decide on pilots and sell them to the business. The reason? “They’re just not comfortable thinking in new ways due to lack of AI fluency,” Srinivasan observes. 

It’s a critical mistake that too many companies make: The allure of new technology drives the business use, rather than business strategy. 

“From my vantage point, when it comes to AI, it becomes solutions chasing problems,” he says. “Some 90% of the AI projects fail to deliver on expectations because they start with AI first and outcomes next.” 

To get real return on AI investment, business leaders need to identify concrete business problems that the technology alone can solve.

One success story Srinivasan points to is a large wireless carrier whose most urgent business problem was customer churn. They piloted an AI solution that reduced churn by 3-5% in one market. Small potatoes? He notes that in rolling it out across all markets, the telecom ended up generating hundreds of millions of dollars of incremental value—while reversing costly erosion of its customer base.

“It’s easy to come up with ideas and chase them because you get excited,” says Srinivasan. “But defining the problem requires critical thinking, research alignment, and talking to customers. It’s a hard process. Companies making the mistake of AI-first are not clearly defining the agenda. That’s happening with gen AI right now, and it’s wasting a lot of time and money. Successful companies say: ‘This is what AI should do for us. This is our North Star.’”

Case in point: Airbus, the global aircraft manufacturer, has fully baked AI into its processes and products, enabling predictive maintenance, fuel optimization, and “smart” factory lines. Founded in 1970, Airbus began its embrace of AI in 2005 and has kept innovating with it over nearly two decades to not only keep but sharpen its competitive edge.

Dual-track deployment

When considering how to ensure that AI strategy maximizes returns, Srinivasan says companies should follow parallel paths: 

1. Tactical deployment. This means identifying where AI can improve existing processes, such as an HR team using ChatGPT to craft job descriptions or customer service agents using a copilot to respond to common queries. This route ends up improving cost savings and revenue, and it’s where many—if not most—companies are looking for value from AI today.

2. Strategic deployment. A second tack is equally critical, despite the fact that it is more difficult, more ambitious, and more challenging to measure the short-term value in terms of investment. Strategic deployment means adopting a “digital native” mindset and exploring new AI-driven business models with an eye toward becoming a disruptor, a potential Uber or Airbnb in a bustling sector.

Based on Davenport’s research, companies that are all-in on AI—meaning they’re trying to change the business, innovate business models, and create new markets and new products—are 2.7 times more likely to improve their competitive position in their sectors.

It’s not easy, Srinivasan admits. 

“It requires CEO and board-level commitment, an experimentation mindset, more employees fluent in AI, long-term commitment, and appropriate levels of investment,” he says. “Your senior management team needs to double down. They need to have conviction.” But to get the most value out of AI over the short and long term, companies need to launch both tactical and strategic initiatives, Srinivasan advises. 

And his final insight for business leaders looking to get ahead of the ROI curve?

“Have a clear philosophy around how you’re going to measure value,” he suggests. “With some projects, it’s easy. You can do a clear A/B test and show with and without AI how it’s incrementally adding value. But in other cases, especially with gen AI and productivity, it’s hard to assess how much of the value is AI and how much is human decision-making.” 

Successful companies will have a clear philosophy and defined processes already in place around measuring the ROI of any new initiative or technology, he notes. They can and should apply that same fiscal discipline to how they measure the ROI of AI. 

“You don’t need an algorithm to tell you the value of an investment,” he says. “Measuring meaningful value has more to do with culture than it does with AI.”