Why does AI ROI stall even when the technology works?
KPMG answers the question finance leaders must resolve in May 2026: why does AI ROI stall even when the technology works? KPMG discusses how ROI stalls when adoption lags—so leaders must treat change management, role‑specific use cases, and hands‑on practice as core infrastructure for scaling AI in finance.
Why does AI ROI stall even when the technology works?
AI in finance has moved beyond pilots. In the next 18 months, 93% of US companies expect to be deploying or scaling AI in their finance functions, and half are planning to orchestrate or develop multi‑agent AI systems across workflows. That pace creates a predictable problem: the toolset can be ready before the organization is.
KPMG’s 2026 AI in Finance survey shows most leaders are seeing returns—46% say ROI is meeting expectations and 28% say it is exceeding them. The more revealing signal is what happens when ROI disappoints: the primary barrier is not model performance or platform selection. It is slow organizational adoption and change management.
Why It’s Harder Than It Looks
AI ROI depends on daily behavior, not a one‑time deployment. Finance teams can “have AI” while still operating in legacy ways—using AI outputs as optional inputs, keeping work in old queues, or limiting usage to a small set of early adopters.
The adoption gap is reinforced by a practical enablement gap. Leaders trying to build day‑to‑day understanding of AI cite two main obstacles: a lack of clear, role‑specific use cases (64%) and a lack of hands‑on practice environments (61%). When teams do not know exactly where AI fits in their role—or cannot practice safely with real workflows—usage stalls, and ROI stalls with it.
The Evidence
1
2
3
4
KPMG’s Answer
ROI stalls when AI is treated only as a technology rollout instead of an operating model change.
Leaders can buy tools, integrate data, and deploy models, yet still miss value if the organization does not adopt new ways of working at scale.
The fix is straightforward but not easy: make adoption measurable and role‑specific. Finance leaders should define where AI is expected to change decisions and workflows, then equip teams with practical environments to practice those new behaviors. The survey’s obstacles—unclear use cases and limited hands‑on practice—signal that training must be tied to real roles, real tasks, and repeatable routines, not generic awareness sessions.
If leaders do not treat adoption as core infrastructure, AI remains confined to pockets of success. That is how organizations end up with strong demonstrations and weak enterprise ROI.
Build a short list of role‑specific “AI must‑use” moments. Choose the decisions and workflows where faster, predictive insights matter most, and define what “using AI” means in those moments.
Fund practice, not just platforms. If 61% of leaders cite missing hands‑on environments, treat sandboxes, coached pilots, and repeatable training loops as part of your AI budget, not a nice‑to‑have.
Explore more
Get in touch
Start the conversation
Connect with our team today to learn how we can help you realize the full potential of GenAI.
