While companies are enthusiastically investing in artificial intelligence, evidence that CFOs aren’t yet seeing a bountiful return within their own departments continues to pop up.
The latest such information comes from Zuora, a provider of quote-to-cash software, including AI-powered tools, for subscription-based businesses. According to a survey of 321 U.S. finance and accounting decision-makers, performed on behalf of Zuora by The Harris Poll, only 28% of those whose finance teams have invested in AI are currently seeing a measurable financial impact.
That 28% does represent a lot of companies, considering that 92% of those polled said their teams are using AI tools.
However, the gap between investment and outcomes is even more pronounced at larger companies, Zuora noted in its survey report. More respondents from companies with at least 1,000 workers have invested in AI (100% of them, compared with 81% of smaller companies), yet the larger ones also were more likely to say they haven’t seen a measurable financial impact (72%, vs. 48% of smaller firms).
The larger companies are also less optimistic that AI will begin delivering measurable financial results soon — 35% said it will take a year or more, compared to 15% of the smaller companies.
A large proportion of those polled (87%) said there are gaps between AI’s promise and its reality in finance. The top three cited causes are difficulty integrating AI outputs into finance workflows (41%), difficulty embedding AI in processes that require multiple teams or cross-functional handoffs (39%) and instances where AI insights don’t match data in core finance systems (35%).
“When AI falls short in practice, it creates doubt,” Zuora wrote. “In finance, trust isn’t an abstract concept. Instead, it’s earned through precision, traceability and control. Systems must produce outputs that can be validated, explained and reconciled within strict audit and regulatory frameworks.”
If AI cannot meet that standard, Zuora said, it introduces risk rather than reducing it.
Ninety-one percent of survey participants have concerns about using AI for core financial processes. The top three are cybersecurity risks and data privacy (45%), lack of appropriate human oversight (38%) and data quality, integrity and reliability issues (36%).
Zuora posed the question: What does “trustworthy AI” actually look like? And the answer was “a desire for AI embedded directly in embedded systems.” More than half (53%) of the survey participants said they would trust that setup the most, compared to an AI solution built by their organization (25%) or an AI-native solution (13%) when evaluating AI tools for finance.
