IPA Data: Half of Accounting Firms Are Using AI

The debate about whether accounting firms should adopt AI is over. According to new data from the Inside Public Accounting June 2025 Insights Report — drawn from 192 firm respondents to IPA's information technology survey — more than half of firms are now actively deploying AI in some form. The conversation has shifted from "should we" to "why isn't it working yet."

That second question is the one I care about

Forty-four percent of firms reported no measurable impact from their AI investments. That's not a fringe result from a handful of laggard firms. It's nearly half the field, and it holds steady across firm sizes. Large firms, mid-size firms, small firms — the adoption-to-results gap is showing up everywhere.

The Bottleneck Isn't the Technology

The most useful finding in the IPA data isn't the adoption rate. It's this: 55% of firms are using 25% or less of the AI capabilities already built into software they own.

So firms are buying tools, deploying them in a nominal sense, and then leaving most of the capability untouched. Before the next vendor conversation, the more productive question is: what percentage of what we've already purchased are we actually using, and why isn't that number higher?

The answer usually comes down to three things — training, workflow integration, and governance. IPA's data directly confirms the training gap: among firms with over $100M in revenue, 69% offer AI training to staff. Among firms under $10M, that number drops to roughly 25%. Smaller firms are adopting AI with roughly one-quarter the training infrastructure of their larger peers. That's a setup for underutilization, not transformation.

Where Firms Are Seeing Results — and What That Tells You

Among firms that have seen measurable outcomes from AI, the most commonly reported results are shorter turnaround times and improved employee satisfaction. Both of those make sense. AI genuinely accelerates drafting, synthesis, and first-pass structuring. Staff who spend less time on those tasks report a better experience.

What didn't show up in meaningful numbers: decreased staffing needs. Only 4% of firms reported that as an outcome of their AI adoption. This matters because it was one of the most frequently cited expectations heading into 2024 and 2025. The firms that invested in AI anticipating headcount reductions haven't seen them — at least not yet. IPA's Chelsea Simmons noted on the Inside Public Accounting Podcast (June 2025) that "maybe it just hasn't caught up to the investments that have been made," but the gap between expectation and current reality is significant.

The practical implication is straightforward: don't build your AI business case around staffing reductions. Build it around throughput, quality, and staff experience — because that's where the actual returns are appearing.

The Governance Issue Is Bigger Than Most Firms Realize

The top concern among IPA survey respondents — above cost and unclear ROI — was the risk of disclosing private client information, cited by 53%. That's not surprising. What's notable is where that concern concentrates.

Smaller firms cite lack of internal expertise and cost as their primary pain points. Larger firms are wrestling with governance and data privacy as their use of AI scales. Both are legitimate problems, but they require different solutions. A $5M firm and a $150M firm don't have the same AI governance needs, and conflating them produces policies that don't fit either.

The firms getting this right are treating governance as infrastructure, not compliance theater. That means defining which tools are approved and why, establishing clear protocols for what client information can and can't be entered into AI systems, and creating sufficient structure so staff can use AI confidently — rather than defaulting to avoidance because the rules aren't clear.

What to Do With This

IPA's data points toward a few specific actions. First, audit utilization before expanding your AI stack. If more than half the profession is using less than a quarter of existing AI capabilities, the issue is almost certainly not the tools. It's deployment, training, and workflow integration. Second, build the training infrastructure your firm size actually needs — not the one that feels proportionate to what you've spent on licensing. Third, set realistic expectations internally. Shorter turnaround times and better staff experience are real and valuable. Staffing reductions at scale aren't showing up yet.

The firms that come out ahead in the next two to three years won't necessarily be the ones who spent the most on AI. They'll be the ones who built the organizational infrastructure to use what they already have.

Source: Inside Public Accounting June 2025 Insights Report and the Inside Public Accounting Podcast, episode featuring IPA Executive Director Chelsea Simmons. IPA survey data drawn from approximately 192 firm respondents to the 2025 IT survey. Full IT report expected September 2025. Data cited as preliminary.