Finance Teams Struggle with AI Adoption: Key Insights for CFOs

Let’s be clear about something. When a finance department buys a new piece of software, the job is not done. It hasn’t even really started. This is the quiet, uncomfortable truth settling over corporate finance in 2026. We’ve spent years talking about artificial intelligence as a wave of automation that would sweep through back offices, yet the tide seems perpetually stuck just offshore. The data, from sources like the State of AI in Finance 2026 report by CFO Connect, tells a story of hesitant progress. Yes, over half of finance leaders now report their teams use AI in some fashion, a significant jump from just a few years ago. But dig a layer deeper and you find finance solidly in last place among major business functions for actual, integrated deployment. A mere 17% of teams use AI inside their core daily workflows. Nearly half are still running limited, small-scale pilots—experiments that never seem to graduate to the operational mainstream.

This isn’t a story about faulty technology. The tools, by and large, work. The gap between purchase and genuine use is a people problem. It is the single biggest variable determining which finance departments capture real value from their transformation investments and which ones remain stuck in a perpetual state of trial, years after the initial fanfare. In a way, it’s the most predictable failure mode imaginable. The earlier phases of any transformation—the operating model redesign, the vendor selection, the governance committees—can be planned and executed by a relatively small cadre of senior leaders. Getting hundreds of accountants, analysts and controllers to fundamentally change their daily work is a different beast entirely. It’s a problem of human behavior, and it’s the one most transformation playbooks are least equipped to handle.

Why does this human element so often get shortchanged? The pressure to show a return on a major technology investment can lead organizations to focus on the tool itself, not the people who must wield it. Deloitte’s research on finance workforce strategy puts a stark number on this misstep. Organizations that took a tech-centric approach to AI adoption were 1.6 times more likely to fail to achieve their expected return compared to those that put people at the center from the start. Deloitte’s broader Human Capital Trends research frames a core tension it calls “stagility”—the push and pull between leadership’s desire for agile responsiveness and the workforce’s need for enough stability to properly learn and commit to new systems. This isn’t abstract. In a separate survey of 1,800 finance professionals globally, Deloitte found half reported ongoing employee engagement issues during tech rollouts, with nearly as many citing outright resistance and a glaring shortage of the skills needed to use new tools effectively. These aren’t the grumblings of a skeptical minority. They are the mainstream experience inside transforming finance teams.

So, how are the successful teams breaking through? Case studies point to two very different, yet equally deliberate, paths. At payments company Adyen, the approach is rooted in co-creation and trust. Finance professionals—controllers, FP&A analysts—rotate directly into the technical teams building AI tools. They don’t just provide requirements. They sit alongside engineers, co-designing use cases. The logic, as one finance leader involved in the CFO Connect research noted, is simple: adoption is a trust issue. If a finance team doesn’t understand how an AI model arrives at its output, they will not rely on it, no matter how impressive its technical specs. They need to see the seams.

Automation platform Zapier represents the other end of the spectrum. Its approach to closing the adoption gap could be described as direct, or even blunt. The company made the use of its chosen AI tools effectively non-optional across the finance function. One executive involved framed it plainly: transformation requires mandates, not nudges. The result is a staggering 98% regular usage rate among finance employees, a figure that highlights just how anaemic pilot-stage adoption is elsewhere. Critically, Zapier measures success not just by adoption metrics but through a mix of objective data—cycle times, work volume—and subjective surveys on how employees feel about the tools. Both models, building trust through immersion or enforcing use through mandate, share a critical commonality. Neither assumes adoption is automatic. Both treat it as an outcome that requires sustained, deliberate management attention long after the software is installed.

For the finance teams still spinning in pilot mode, the roadblocks are now well-documented. The CFO Connect research highlights a consistent set of issues. Sixty-eight percent of CFOs cite not knowing which processes to target first, a paralysis that keeps experiments small and contained. There is a widening skills gap. While Deloitte’s workforce research finds finance leaders wisely favor upskilling existing teams over hiring new AI-fluent talent, this strategy takes time—far more time than simply purchasing a software license. Trust and data security concerns linger, though these are easing as enterprise-grade tools with robust governance become standard. Perhaps most telling is what the research calls the “tinkerer” trap. Many finance professionals are experimenting with general-purpose AI like ChatGPT for discrete tasks, manually copying outputs between systems. This creates the illusion of progress without the governed data infrastructure needed for true, end-to-end workflow automation. None of these barriers requires a larger technology budget. They require the kind of sustained focus on people that a software rollout alone does not provide.

This leads to an uncomfortable conclusion for any CFO who has championed a multi-million dollar transformation. The technology stack, the operating model, the governance framework—all of it is necessary. None of it is sufficient. A tool that employees do not trust, do not understand or do not feel obligated to use will remain in permanent pilot mode. Its business case becomes a museum piece. The finance functions pulling ahead in 2026 are not necessarily those with the most advanced algorithms. They are the ones that treated getting their people to actually use the technology as seriously as they treated the act of buying it. They built a deliberate mechanism—be it co-design, mandate or another model—to ensure adoption happened. They stopped assuming it would. For a function whose entire future depends on its people working differently, that focus may be the most non-negotiable line item of all.

• Human behavior is a major factor in technology adoption.
• Successful teams prioritize co-creation and trust.
• Creating mandates can drive usage rates up.
• Organizations need to focus on employee engagement.
• Understanding systems leads to higher adoption.
• Continuous management is required post-deployment.