TD Bank Group has high expectations for its AI initiatives. But like many other businesses bringing AI into the fold, measuring AI's success financially is a challenging priority.
In a September 2025 investor day conference call, TD Bank Group President and CEO Raymond Chun said the Toronto-headquartered financial company is "targeting $1 billion in annual value from AI. Half through revenue uplift, and half through cost savings, with concrete plans already delivering clear outcomes."
Since that conference call, TD has reported progress toward its objective. In its Q4 2025 results presentation, the company reported that approximately 75 AI use cases implemented in 2025 generated $170 million in value and said its AI use cases are expected to generate $200 million in value in 2026.
Like many companies, TD has been a longtime user of machine learning and predictive analytics, said Ted Paris, head of analytics, intelligence and AI for TD Bank. The bank has historically used AI for campaign optimization, next-best product recommendations and crediting decisions. It has used intelligent geospatial mapping to identify optimal locations for its storefronts and natural language processing to read operational documents.
Building on that foundation, TD has been advancing its AI agenda over the past decade. In 2018, it acquired AI company Layer 6, which is now the bank's AI research center of excellence. In the last few years, it has deployed generative AI, agentic AI and advanced AI capabilities in an increasing number of processes. The objective is to transform workflows to hit that $1 billion in annual value, Paris said.
"We're enhancing [AI] capabilities to support our colleagues, so they evolve their roles into things that are more value-generating," he said. "And we're improving experiences for clients."
TD sets its sights on AI for revenue uplift
Chun's public commitment to target $1 billion in annual AI value set a north star for ensuring the bank's AI efforts deliver returns, Paris said. "Thinking about ROI gets us focused on the right priorities," he added.
TD is intentionally identifying business opportunities where AI can increase revenue in measurable ways, Paris said. For example, the bank launched TD AI Prism, a predictive AI foundation model, in June 2025. TD uses the model to understand customers and advise them on which products best suit their needs.
The model's goal is to provide more personalized, customer-centered banking experiences, Paris said. Early testing showed that TD AI Prism was 20% to 30% better at identifying customer needs than the models the bank had been using, according to a TD announcement about the model.
Paris also cited the bank's use of AI in marketing to generate hyper-personal content for existing and prospective clients, delivered through various channels, as another example of how AI is improving top-line numbers. He said TD is also using AI to improve its sales and operations execution (SOE) on AI platforms such as ChatGPT, Claude and Gemini as consumers shift away from traditional search.
Thinking about ROI gets us focused on the right priorities.
Ted ParisHead of analytics, intelligence and AI, TD Bank
In another case, announced in May, TD is using agentic AI to automate the application process for mortgages and home equity lines of credit, reducing the average application time from 15 hours per client to just under three minutes.
Executives have identified metrics for the bank's AI use cases to ensure their initiatives deliver on both bottom-line and top-line objectives, Paris said. Specific top-line growth metrics include measures of customer acquisition, customer retention and engagement and other revenue-oriented metrics. However, Paris declined to provide specific figures on the metrics used to measure ROI for any AI initiative.
Businesses must get creative to see AI ROI
Many businesses have yet to successfully shift to using AI for financial growth, meaning that any ROI they're getting from their AI investments still comes exclusively from efficiency and productivity gains.
Deloitte's January 2026 "State of AI in the Enterprise" report identified that while companies are seeing efficiency and productivity gains from AI, other benefits are more aspirational. "In particular, revenue growth largely remains an aspiration, with 74% of organizations hoping to grow revenue through their AI initiatives in the future compared to just 20% that are already doing so," the report said.
Other research paints similar pictures. For example, "Building The AI-Native Enterprise," a July 2026 report from business and technology consulting firm West Monroe, found that 88% of companies are using AI in some form, but only 6% are seeing clear financial returns.
There are multiple reasons why businesses are not seeing ROIs from their AI initiatives, said Connor Augustyn, a partner at West Monroe and managing director of CFO advisory. Some organizations are focused on using AI rather than identifying business problems or business opportunities that AI could address. Some skipped the foundational work, such as data modernization, needed for successful AI deployments. Others ignored the change management required for transformations to stick.
Most firms started their AI journeys by using the technology to reduce costs and boost productivity, but are now pivoting to more transformative projects, Augustyn said. "A lot of those [initiatives] are failing to deliver the ROI that executives were hoping for, so they're opening up the aperture."
Augustyn saw that approach pay off for one client, a B2B company. Like others, the company had been using AI to drive back-office efficiencies but has since shifted to identifying ways AI can help drive revenue growth. It is now using AI to analyze internal and external signals to identify which customers are at risk of terminating or not renewing their contracts with the company. The tool includes a dashboard that shows clients at risk and uses GenAI to propose steps salespeople can take to retain each client.
The company has incorporated the AI tool's outputs into its weekly sales meeting agendas. "That is influencing the top line because they're reducing customer churn and preserving some of their revenue," Augustyn said.
Measuring AI's value
Brian Hopkins, vice president of emerging technology and a principal analyst at Forrester Research, said he sees many executives struggle to measure the value of their AI deployments.
In an April 2026 report, Forrester introduced its AI Value Matrix: A Framework For Measuring What Matters. The matrix is built on three financial outcomes -- top line, bottom line and risk management -- and three value mechanisms: productivity, engagement and strategy. Using AI to enhance productivity delivers short-term gains that are easy to measure, while using AI for customer engagement produces midterm gains that are harder to measure. Gains around strategy are the hardest to measure and take the longest to see.
Despite these challenges, businesses can and should pursue those top-line and strategic gains and identify the metrics that will show whether they're succeeding, Hopkins said.
Some organizations doing this have implemented value management practices. The goal, Hopkins added, is to determine "the big metrics your organization needs to move with AI to create a substantial advantage over your competitors."
Some businesses might not know whether they're affecting their bottom line or top line because they haven't identified which metrics to use to make such determinations, Augustyn added.
It's challenging to know what to measure regarding AI's effect on revenue. Using gross margins or gross sales is unlikely to accurately measure AI's impact because many factors can affect those figures, Augustyn said. However, some longstanding metrics, such as days payable outstanding and days sales outstanding, could be used when AI is introduced in singularity to determine whether it is having a positive effect on the company's finances.
West Monroe's Augustyn explained that his B2B company client knows that its AI initiative is working by measuring its effectiveness. The company opted against using broad metrics, such as client terminations, and instead introduced a new metric, one that measured the churn rate of clients at risk, to measure the tool's value.
Mary K. Pratt is an award-winning freelance journalist with a focus on covering enterprise IT and cybersecurity management.
