Getty Images/iStockphoto
AI changes the ROI equation. Here’s how some have found success
A significant portion of enterprises struggle to demonstrate clear ROI from AI initiatives, but some have found areas where AI can provide productivity gains and even revenue growth.
Like many companies, Alight Solutions, a benefits administrator, has been searching for ways AI could improve operations and its bottom line.
While not a new technology to the company, Alight began assessing a different kind of AI tool in 2025. In September of that year, Alight became a beta user of a fraud detection recruiting agent from Phenom, an HR technology vendor. The company wanted to see if the agent could help identify fraudulent candidates early in the recruitment stage.
It didn’t have to wait long for results. During the testing process, the agent identified a candidate who applied for an open position twice, using a different name and email address each time.
“Even by catching that one person that we ultimately didn't hire, that was enough for us to say, ‘It's going to work for us,’” said Julie Eagy, talent acquisition operations manager at Alight.
While the agent showed where it might help save the company money, especially if it could prevent it from paying for an unnecessary background check, the greater value is time savings, Eagy said.
Alight provides an example of how measuring a return on AI investments can get messy. For many companies, the return on investing in AI tools might not be monetary gains but rather productivity gains or the ability for the AI agent to provide a missing piece within their organization.
That lack of clarity has led some enterprises to let go of some initiatives. A Pizza Hut franchisee, for example, sued the franchise in May claiming that an AI delivery management platform was counterintuitive in that it cost it millions of dollars rather than providing it with any kind of savings.
Despite AI’s mixed results, enterprises have continued to push forward with the technology, often leaning on productivity as a metric. Meanwhile, some small businesses, including SMBs, have been able to track monetary gains and returns from their AI projects.
A complicated situation
The reality for enterprises is that AI ROI can be nuanced and even opaque, according to Arun Chandrasekaran, an analyst at Gartner.
“It’s a very complex discussion,” Chandrasekaran said. “Broadly speaking, I would argue that a lot of customers are still struggling to articulate clear business value from AI.”
Domains such as software engineering, customer service and back-office operations in finance or HR have been able to automate core tasks, resulting in productivity gains, but companies struggle to tie that back to tangible business value, such as revenue growth or cost reduction, Chandrasekaran said.
“Alignment between productivity gain and these value-driven metrics that businesses care about -- that’s still a work in progress, I would argue, across a lot of organizations,” he added.
The struggle might be more difficult, depending on the size of the organization. For some SMBs, making the connection has proven to be the best way they can come out ahead of the competition and stay ahead.
The OBI creative business value
One such business is OBI Creative, a 25-year-old advertising agency based in Omaha, Nebraska, with customers including TiVo and HD Radio. OBI Creative has spent the last few years experimenting with AI tools as it seeks to grow the business.
“Agentic AI has become a really great efficiency opportunity for us to scale some of our offerings from one industry to another,” said OBI Creative’s CEO and founder, Mary Ann O’Brien.
The company, which has fewer than 50 employees, built a couple of agentic AI tools in pursuit of its goals. One tool monitors the health of the websites its digital team builds and ensures they are up to date. The advertising agency also used models, including Gemma and Qwen, to code prompts into the tool to help generate reports on the website's health. It also used OpenAI GPT models and Anthropic Claude models to code prompts into its tools.
Another tool enables the creative team to check whether the campaign it is working on aligns with what the client wants.
“We built it as a prototype to help [with] our brand strategy team and creative team alignment, and then it became something our clients wanted to use, so we started selling/licensing it to them,” O’Brien said.
She added that the value gained has turned into more gross margins for her agency.
“We don’t have to go back to the client five times for changes,” she said.
Plus, she added, by using agentic AI tools, the agency “will have at least 20% year-over-year growth.”
With the growth comes added expenses.
“We have definitely seen an increase in our overhead costs, but it’s quickly been balanced out by some of the efficiencies,” O’Brien said. She added that the agency began by experimenting with AI models that had fixed costs and offered subscription services, enabling her employees to work within a set budget.
In addition to investing in the technology, O’Brien made it clear to employees that the use of agentic and generative AI tools wouldn’t jeopardize their jobs. In fact, she challenged one employee to become an ambassador for efficiency and to provide the results as part of his performance.
“There are still people in my agency who are very afraid of the tools, and so it's just really getting people comfortable and being able to trust that the thing between their ears is still the most valuable to our clients,” O’Brien said. “But the tools are there, and why wouldn't you use them?”
The Cornell approach
While Cornell University is not an SMB, it has taken a multifaceted approach to finding ROI with its AI investments.
“We focus on this technology as a tool,” said Ayham Boucher, head of AI innovations for Cornell Information Technologies. He added that early on, Cornell allowed faculty to access any frontier model in a secure, private environment, but does not mandate their use. It uses Microsoft Azure, where users can create, manage and deploy their agents, as well as user-facing tools including Microsoft Copilot and Claude Desktop.
The main objective of agentic AI for Cornell is its ability to expand human capabilities, Boucher said.
“We are not superb creatures,” he said, referring to human capabilities. “We have limited memory. ... We have been using technology to expand what we do as humans for a long time, and AI is just one more powerful tool to expand our capabilities.”
He added that Cornell is also not focused on measuring value with tokenmaxxing, the recent trend of organizations partly measuring employee performance by how many AI model tokens they consume.
“Our outcomes are aligned with the mission of the institution,” he said. In other words, the university measures value by the number of students served and the number of scientific discoveries made. His department measures value by supporting those endeavors rather than by counting the number of tokens burned.
Some companies have begun experimenting with AI but have found little traction or struggle to transition an AI pilot project into production. According to a 2026 report from Gartner, around half of generative AI projects were abandoned after the proof of concept for reasons such as poor data quality, escalating costs or unclear business value.
“Because we deployed and started to build with AI so early, we didn’t fall into the trap of FOMO [fear of missing out],” Boucher said. “We did not have to. We are ahead of the curve, which allows us to maintain a balanced approach. We are not trying to catch up."
A measure of value
For those trying and experimenting and still not finding value, they might consider a different metric other than revenue or productivity, Gartner’s Chandrasekaran said.
“We have to tie more operational metrics to more value-oriented metrics,” he said. For example, in software engineering, the metric should not be lines of code but velocity, such as the new features or capabilities being delivered to an application.
Cost reduction or revenue growth can limit a company’s understanding of how AI is delivering value, Chandrasekaran continued.
For enterprises that have not been able to achieve ROI, the key might be to first establish what return is needed from a select application where generative AI or agentic AI might be useful, Chandrasekaran said.
“We don't want to be implementing use cases and then start thinking about how we're going to measure value,” he said. “If we can't see a very clear line of sight in terms of value, we don't even pursue those use cases.”
However, enterprises should be cognizant that the application they are experimenting with will determine how long it takes to pilot an AI project.
“The ROI period is really dependent on what type of use cases you're pursuing,” Chandrasekaran said. “I would generally say for 80% of enterprise use cases, you want to get to an ROI within a year.”
However, the fact that business value remains a challenge for AI use within the enterprise also suggests that vendors still have a job to do, he said.
“The vendors need to do a better job of not just selling technology, but selling outcomes and selling value to the enterprises,” he continued. “Enterprises need to buckle up and do a better job of leading with value as a metric when they choose use cases and be very relentless in terms of focusing on value realization for their AI projects.”
Esther Shittu is an Informa TechTarget news writer and podcast host covering AI technology.
