Three Approaches to Measuring and Managing AI ROI
Companies struggle to quantify the real returns on their AI investments. Assess your current efforts and take steps to improve.
Topics
News
- Anthropic's Mythos Finds Critical Flaws in Classified US Systems: Report
- Qatar Launches AI-Focused Scholarship Program to Build Future Digital Workforce
- China's LineShine Tops TOP500, Becomes World's Fastest Supercomputer
- Japan's Sakana AI Unveils Fugu, Claims Edge Over Claude Fable 5 in Coding
- Oracle Cuts 21,000 Jobs as AI Silently Reshapes Its Operating Model
- Nvidia’s New Cooling System Cuts Data Center Water Use to Near Zero—But Not AI’s
Matt Harrison Clough/Ikon Images
After several years of AI experiments and pilot initiatives, a crucial question remains open for most companies: How much of a return — and what kinds of returns — are we getting from all of this AI investment? To many executives, AI ROI still often feels more like art than science: elusive, imprecise, and industry-dependent.
Surveys and benchmarks paint a confusing picture about current returns. Much of the guidance also remains focused on measuring inputs — encouraging organizations to invest, experiment, and build capabilities (“You should invest in …”) — rather than on outputs and how to assess impact (“Here’s how to measure results”). Today, few companies apply the same financial discipline to artificial intelligence as they would to a new factory or piece of machinery.
Our interviews with more than 30 CEOs and senior leaders across various industries confirm that measuring AI ROI is anything but standard practice: Two companies making nearly identical investments may define success in entirely different ways. Yet companies that fail to identify an explicit approach to AI ROI — or that simply roll out generic AI tools and hope for productivity gains — rarely realize credible, lasting returns.
ROI measurement differs by the type of AI technology being used. Analytical AI projects, which are typically based on established machine learning techniques like prediction and optimization, often produce more directly attributable financial returns but tend to be applied to targeted, well-defined use cases. Generative AI, in contrast, is broadly applicable, given its ability to perform a range of knowledge work tasks previously done by humans. A GenAI tool often creates improvements in speed, quality, or volume of work, requiring deliberate translation into financial impact. And some companies combine both analytical and generative AI solutions in a customized manner.
AI ROI also depends heavily on industry context. In the consumer goods sector, companies streamline their supply chains by using analytical AI, enhancing demand responsiveness. A B2B marketing agency using generative AI may focus instead on creative throughput and ideation, proposal win rates, or lead conversions — a different definition of “return.”
Three Pathways to Tangible AI ROI
Based on our interviews with executives, we identified three practical approaches to measure and manage AI ROI. These approaches reflect a range of AI maturity levels among companies, and varying strategic intents.
By comparing your organization’s current approach against this framework, you can identify where you are and what it will take to move forward. The overarching goal for leaders: to ensure the translation of AI activity into verifiable business results.
Measuring and Managing AI ROI: Three Approaches
Companies often start with the function-focused approach and work up to the enterprise portfolio approach over time.
| Function-focused approach | Coordinated approach | Enterprise portfolio approach | |
|---|---|---|---|
| Core idea | Focus on one business function or a small number of functions or processes. Use tailored AI solutions and metrics. | Coordinate the deployment of broadly applicable AI tools and function-focused initiatives. | Engage in enterprisewide governance of the AI portfolio. |
| Typical metrics used | Function-specific KPIs, such as response time or error rates. | A mix of broad operational metrics and function-specific KPIs in selected high-impact AI initiatives. | Investment portfolio value, NPV/IRR, business case ROI. |
| Potential pitfalls | Siloed metrics and no shared view across the organization. | Limited comparability and fragmented portfolio-level oversight. | Risk of excessive bureaucracy that may constrain early-stage or exploratory initiatives. |
| Next steps for improvement | Start scaling metrics toward a companywide AI ROI playbook. | Apply consistent financial translation and measurement logic across all AI initiatives. | Use financial and strategic metrics. Allow early bets without full ROI measurement. |
1. Function-focused approach
Who it serves: Companies trying to build credible proof points before scaling.
With this approach, you select one or a small number of business functions, such as customer service, marketing, production, or HR, as the starting point for focused AI tool deployment. In each function’s case, you build or acquire tailored AI solutions and equip people with rigorous, function-specific performance metrics. This means tracking outcomes such as shorter response times, fewer errors, improved quality, or reduced unit costs. For leaders, the logic is “If we can demonstrate credible ROI here, we can justify broader deployment elsewhere.”
Function-focused AI initiatives often deliver some of the most tangible ROI, especially when paired with deliberate workflow redesign. In customer service, organizations that deploy GenAI-driven agents and decision-support tools have reduced handling times and call volumes — often automating a high percentage of routine customer requests — and translated those gains into lower service costs and improved customer satisfaction.
For instance, Unilever redesigned its early-stage recruitment process around AI-based candidate assessment, reducing HR’s reliance on external recruiters while shortening time to hire and lowering recruitment costs. In other companies, finance units have experienced similar dynamics, where AI-based forecasting, pricing, or fraud detection systems embedded into core decision workflows have improved accuracy, reduced losses, and delivered measurable cost benefits.
As function-level proof points accumulate, leaders can gradually move toward a shared AI ROI playbook.
The function-focused approach to AI ROI is particularly effective for building organizational confidence in AI investments. The plus side: By limiting scope and maintaining clear ownership, organizations can create credible proof points that are easier to measure, explain, and defend. The negative side: Because specific needs and contextual factors shape function-specific ROI, different success stories might be difficult to compare or aggregate as AI adoption expands.
Your next move: If you’ve already done several function-specific AI initiatives, it’s time to begin laying the groundwork for the next stage: coordination. As function-level proof points accumulate, leaders can gradually move toward a shared AI ROI playbook with consistent definitions, financial logic, and data instrumentation standards. Start by standardizing metrics that can be transferred across functions and aligning financial assumptions across key use cases. As one CEO said, “We’re iterating toward a more structured model, linking AI impact to planning, budgeting, and playbook development; it’s an ongoing loop of learning.”
2. Coordinated approach
Who it serves: Companies trying to make AI ROI comparable across functions or units.
With this approach, you’re managing a growing set of AI initiatives across the organization. Concurrently with function-specific deployments, or even earlier, you’re also rolling out some general-purpose AI tools and shared AI capabilities that touch multiple teams and workflows. The defining challenge here is coordination: maintaining broad visibility into AI activity while selectively focusing on the metrics that have the most significant business and economic impact. Ideally, this approach facilitates shared learning, reuse of proven metrics and assumptions, and faster replication of successful AI use cases.
Organizations taking a coordinated approach often use shared AI platforms and capabilities to manage initiatives spanning multiple teams. At JPMorgan Chase, an internal AI platform called LLM Suite has been deployed to more than 200,000 employees across legal, research, client services, operations, and other functions. This gives people broad access to generative and analytical AI tools while requiring coordination mechanisms to ensure consistent value creation. At Amazon, the evolution of internal AI systems resembles an AI flywheel, whereby innovations — such as recommendations or robotics — that begin in isolated teams spread and are reused across the organization through shared machine-learning platforms and practices.
In both cases, value comes from coordinating how results are interpreted, compared, and scaled across the organization. At this stage, generative AI tools are often used both inside and across business functions, heightening the need for coordination. Analytical AI tools deliver results that are often easier to compare, via clearer links to operational and financial outcomes.
The logic and business motivation for coordination are straightforward: “We’ve invested in many AI initiatives, and we need a way to stay on top of them all.”
However, especially in larger organizations, coordination without clear standards can result in a patchwork of ROI methods, making it difficult to align priorities, compare outcomes, and decide what to scale.
Your next move: During this phase, it’s important to continue prioritizing and standardizing. Identify where deeper ROI instrumentation is warranted, and apply consistent financial logic across the full set of AI initiatives, regardless of whether they involve broad tools and capabilities or targeted deployments. Standardizing how results are translated into financial terms enables meaningful comparison and scaling across initiatives. As one CEO put it, emphasizing the need for a common baseline, “If an AI initiative claims to replace the work of four employees, I want to know who they are; otherwise, it’s not real savings.”
3. Enterprise portfolio approach
Who it serves: Companies that are ready to govern AI ROI at scale.
This stage represents the highest level of ROI maturity and is where you’re applying rigorous financial logic across the entire portfolio of AI initiatives. An AI initiative is treated like any other significant investment: It is governed through forums similar to those for capital projects and is evaluated with business cases, financial models, and portfolio metrics such as net present value and internal rate of return. This approach emphasizes funding projects that create measurable value as quickly as possible.
At Morgan Stanley, for example, AI initiatives are assessed through a structured evaluation framework that tests each use case against real-world criteria before deployment. This approach enables disciplined enterprise-level oversight and scaling of AI tools. In comparison, one equipment manufacturing company we studied applied strict financial discipline: Both analytical and generative AI initiatives were allowed to run for a limited trial period and were routinely terminated if they failed to demonstrate measurable value within six months. This ensured rigor but risked premature rejection of promising efforts.
A professional services firm pursued another option: It separated two kinds of AI initiatives — those that built mandatory foundations for generative AI adoption, where ROI was not enforced upfront; and targeted, tailored applications, where clear financial returns were required. Effective enterprise AI ROI management depends on deliberate and company-specific choices about timing, risk tolerance, and evaluation rigor.
At the portfolio level, both analytical and generative AI are evaluated as part of the investment mix, but often under different expectations. Analytical AI work fits naturally into traditional financial models, whereas generative AI initiatives may require staged evaluation and adapted governance. For example, milestone-based funding or phased ROI thresholds may be needed to save worthwhile initiatives from premature rejection when those projects have indirect benefits, delayed adoption, or value creation driven through learning.
Apply lighter ROI tracking to early-stage AI experiments and introduce more scrutiny as projects scale.
The enterprise portfolio approach to AI ROI offers clear benefits. You can compare AI initiatives side by side, compare them with other technology investments, track portfolio-level value creation, and make more confident decisions. As AI initiatives begin to reshape the operating model, however, initiative-level ROI comparisons become less informative; leaders should then rely more heavily on enterprise-level performance indicators to assess systemwide impact.
Your next move: If you choose to take an enterprise portfolio approach, it’s important to decide how strict you want to be. Fully enforced ROI can kill breakthrough AI bets too early if you overlook the value of new capabilities, learnings, and spillover benefits. The goal is to balance financial discipline with strategic patience: Apply lighter ROI tracking to early-stage AI experiments and introduce more rigorous scrutiny as projects scale. Consider creating a separate unit or governance track for more radical AI initiatives. As one executive told us, “You don’t need to measure everything from day one; start with clear KPIs for each area, then layer in more rigor as solutions mature.”
Getting AI ROI Right: Three Takeaways
Many organizations will move through all three approaches to AI ROI over time. Here are three parting takeaways from the executives we interviewed:
- Prioritize high-value, scalable AI use cases. ROI becomes most visible and meaningful when AI is applied to high-volume, high-leverage work. Whether through enterprisewide deployment or targeted use cases, even small productivity gains in large-scale activities can compound into significant value.
- Lead decisively. AI ROI doesn’t materialize by accident. The benefits come only when you provide direction, follow through, and rethink how work gets done.
- Remind yourself that your company and AI technology will keep evolving. To navigate ongoing changes, avoid both overengineering and under-measuring.
As your organization accumulates AI maturity, use the three approaches to track your progress and see your ROI grow.
