Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

5 ways to measure AI’s hidden ROI

Real AI investment ROI comes from impacting your business’s most critical value drivers.

[Photo: Getty Images]

Profile photo of Traci Gusher

Principal, Americas Data AI and Automation Leader · EY

The past two years have seen a frenzy of artificial intelligence experimentation. Across sectors, teams have raced to build pilots, test use cases, and deploy productivity tools, all under the promise of transforming their business. But now, as the initial buzz fades, many leaders are left wondering, why hasn’t all this potential shown up on the bottom line?  

Here’s the inconvenient truth: If your AI investments feel like a game of whack-a-mole—scattered, reactive, incremental—they probably are. And they’re costing you more than time. They’re distracting you from the real return AI can deliver when it’s aligned to the big levers of enterprise value.  

THE PRODUCTIVITY TRAP  

For the most part, organizations have been measuring AI success primarily through one lens: productivity. The thinking goes, if employees use a generative AI tool to write faster, analyze quicker, or respond more promptly, the business wins. But in reality, these personal efficiency gains, while great for individuals, often fail to deliver meaningful, measurable value for the company.  

Why? Because most of these initiatives follow a grassroots, bottom-up approach—referred to earlier as use case “whack-a-mole.”   

Here, every department launches its own individual experiments, regardless of what may be happening elsewhere. Some stick, others don’t. But even the wins are small and disconnected. A marketing team shaving 10% off report writing time or a finance analyst speeding up spreadsheet reviews are unlikely to move the needle on your P&L.  

ROI THAT MATTERS  

Real AI return on investment doesn’t come from streamlining task efficiency, it comes from impacting the business’s most critical value drivers. Yet many organizations are still hoping that a stack of minor productivity boosts will somehow add up to an overall transformational effect. It’s not impossible, but it’s rare.  

The companies seeing meaningful returns are those that elevated AI from a technology initiative to a business imperative. They’re not focused only on measuring how many employees are using internal GPT tools or how quickly reports are being drafted. Instead, they’re targeting AI interventions at financial levers that have always mattered to the enterprise: reducing physical assets on the balance sheet, cutting the cost of goods sold, unlocking working capital, and lowering selling, general, and administrative expenses.   

Used this way, AI stops being a tool and starts being a catalyst. Investor confidence grows, executive alignment strengthens, and the technology becomes embedded in how the business creates value, not just in the way people complete tasks.  

TRACK IMPACT: 5 STEPS FOR LEADERS  

Of course, a complete AI performance framework isn’t just about financials. It will also track value realization, i.e., is the AI solving business-critical problems tied to core strategy?   

Likewise, it will measure the stability, scalability, and efficiency of AI systems. And it will determine how the company maintains responsible use of AI models, limits their climate footprint, and complies with regulations like Europe’s EU AI Act and GDPR, or U.S. state legislation such as the California Consumer Privacy Act.  

As for how leaders can start capturing AI’s hidden (but most valuable) ROI, here are five first steps:   

  • Merge your bottom-up with your top-down: Identify your most obvious profit and loss levers that need improvement compared to your peer set. Seek to reimagine entire end-to-end process areas associated with those P&L levers, and map them to your use case inventory that has been ideated across the business. Also identify foundational technical patterns and create a “force multiplier” effect from your development efforts.  
  • Move from how to what: Many organizations focus on how to use AI tools, without defining what they should be used for. Providing persona- and/or role-specific guidance, such as a finance analyst’s daily AI-enhanced tasks, translates capability into outcomes. Don’t just train on the tools; instead, leverage those tools to train on the process changes.  
  • Create a transformation office: The most successful companies don’t leave AI adoption to chance. Consider developing a formal AI change management team to drive cultural adoption, provide education, and report usage and success insights back to employees.  
  • Establish an AI ethics compass: To succeed with AI, companies need more than policies, they need principles. For example, a healthcare insurance provider I work with recently made a deliberate choice not to use AI to handle calls about death benefits—not because it couldn’t, but because it conflicted with their values. Bereavement is deeply personal, and the company decided those conversations should always be human. That’s a perfect example of an ethics compass in action, guiding organizations in choosing where not to deploy AI for reasons of principle, even if the ROI might look tempting.  
  • Red team your AI: Just like penetration testing in cybersecurity, red teaming is about having a dedicated group of employees who are essentially trying to break your AI. Their job is to expose weaknesses in the system and stress test for issues like bias, drift, and adversarial manipulation. You can also link red team testing to your ethical compass to ensure the AI is going to perform in accordance with your values.  

GO DEEP, NOT WIDE  

In 2022, the emergence of generative AI sparked a wave of innovation as most organizations took the approach of letting a thousand flowers bloom before deciding which use cases to move forward with long term.  

Yet now, things are changing. Leaders are realizing that a thousand mini productivity gains deliver far less value than a single end-to-end transformation of a business-critical process.  

The initial excitement may be wearing off, but the real impact of AI is just getting started. Businesses that use it to go deep instead of wide, really disrupting systems and processes at an organizational level, will unlock unprecedented value at scale.  

Traci Gusher is EY Americas data AI and automation leader.