Most finance teams using AI are not yet seeing measurable financial impact. Only 28% of finance and accounting decision makers report measurable results from AI investments, while 87% say there is a gap between AI promise and reality, according to Zuora-commissioned Harris Poll research.
The most common problems are operational: 41% of respondents cite difficulty integrating AI outputs into finance workflows, while 33% point to the inability to audit or explain results. These issues limit whether AI outputs can be used in billing, revenue, and audit processes, where decisions must be traceable and consistent with financial controls.
Zuora addressed these constraints in a recent webinar, Supercharging Productivity with AI in Quote-to-Cash: What Finance Needs to Know. The session focused on how finance teams apply AI inside quote-to-cash workflows, where outputs must be reviewed, explained, and executed within existing systems.
Why AI Is Hard to Apply in Quote-to-Cash Workflows
Quote-to-cash workflows, which cover how contracts, billing, revenue, and collections are processed and reconciled, are becoming more complex.
In the webinar, Shealyn Nosal, Senior Product Marketing Manager at Zuora, said finance teams face “more pricing models, more contract changes, more revenue rules,” while team capacity has not grown. That leaves teams spending more time “reconstructing what happened across those invoices, contracts, and revenue before they can act.”
That additional work sits at the point where decisions must be validated and explained.
Kaela Gentry, Billing Manager at Zuora, said, “Finance requires accuracy,” adding that “a small mistake can have financial consequences.” She also pointed to audit and compliance requirements, noting that AI can make it harder “to explain the decision-making to auditors and prove consistent outputs.”
The reason is that those decisions do not sit in isolation. Mallory Foster, Revenue Operations Manager at Zuora, explained that finance workflows are connected from upstream processes through billing and revenue. “Everything is connected,” she said. “It all flows from upstream processes, so you can’t really optimize one piece in isolation.”
Given these constraints, Foster said, “it’s not enough for AI to just be fast or helpful,” it “has to be accurate, auditable, and consistent.” Finance-grade AI must deliver “automation with accuracy, transparency, and control,” because “if we can’t trust it, we can’t use it.”
How AI Is Being Applied Inside Finance Workflows
In practice, adoption is more controlled. Gentry and Foster described teams starting with lower-risk use cases, with additional checks to prove accuracy and consistency, prioritizing outputs that can hold up under scrutiny.
Contract review came up early in the session as one of the more practical ways AI is being used. Foster said it has been “one of the biggest wins” for Zuora’s close process, helping teams “move faster, reduce manual effort, and increase confidence” in outputs.
She also described testing ways to model revenue impact in real time, including making allocation changes directly in the tenant without using a separate sandbox.
It replaces much of the manual preparation work before teams can begin analysis. In beta, Gentry said Zuora AI has supported faster invoice investigations and reconciliations, along with reporting and data queries that “used to take hours and are now just taking minutes.”
Kevin Suer, Director of Product Management at Zuora, showed how those investigations, reconciliations, and queries are performed inside Zuora AI during a demonstration. He generated account summaries, a billing operations dashboard, Excel outputs, and a billing-to-revenue reconciliation query from within the application.
The reconciliation example pulled data across billing and revenue systems to identify invoices from the previous 60 days where records did not align, work Foster said typically requires pulling reports, matching them manually, and building supporting analysis.
Controls were not treated as a separate layer. In the demo, a rule blocked AI from writing off subscriptions above a configured threshold, with the system refusing the action because of an “active business control violation.” Suer said read-only mode is the default and updates require approval, while Gentry called that behavior “huge,” because teams can trust the system “to stop when it needs to stop and get a second check.”
What This Means for ERP Insiders
Workflow integration defines AI value in finance. The primary constraint is whether outputs can be used inside workflows that connect billing, revenue, and reporting. AI that cannot integrate at that level remains peripheral to financial decision-making.
Quote-to-cash is a proving ground for AI. These workflows combine structured data, repeatable processes, and financial consequences, making them an early test case for AI operating inside enterprise systems. Success here signals whether AI can move beyond analysis into controlled execution environments.
AI in finance is refocusing on control systems. Evaluation depends on how it operates within existing controls, approvals, and audit structures, with implementation shaped by system design, governance, and how outputs move through controlled workflows.
