Nearly every small accounting firm is using AI in some capacity. Very few can prove it’s actually working, but the reason probably isn’t what you’d expect.
In most small firms, the AI adoption story doesn’t read like a neat-and-tidy case study. Instead, it reads more like a diary (and a messy one at that).
There was no steering committee. No pilot program with clearly defined success metrics. No change-management consultant armed with a fancy slide deck and a 10-step implementation plan. In most cases, leadership simply said something like, “Hey, I think we should be using this.” And then everybody wandered off to figure it out on their own, one chatbot window at a time.
And oddly enough, that worked. Well, sort of.
Based on Financial Cents’ recent survey of 486 accounting and bookkeeping professionals across North America, nearly every small firm is at least on the AI adoption curve—a staggering pace of change for a profession that still openly adores spreadsheets.
But as it turns out, “using AI” and “getting something out of AI” are two very different things. And the gap between them is wider than most firms would like to admit.
But, let’s start with the good news.
The Surprising Speed of Implementation
The headline number from the survey was hard to argue with: 95% of firms are actively using AI. Only 5% are still sitting it out.
That’s a pretty remarkable split for an industry that tends to greet new technology with a healthy dose of skepticism (some might even say disdain). And it happened largely organically, without any dedicated budgets, firm-wide mandates, or detailed rollout plans. Small firms simply decided, more or less independently, that AI was worth a try.
95% of firms have adopted AI in some capacity, but only 11% are “running” with it.
Where would you put your firm on the AI adoption curve?
When you look a little closer at that curve, though, the picture gets more complicated. Only 11% of respondents describe themselves as “running” with AI (meaning it is woven into daily workflows across the firm). The rest are walking, crawling, or still lacing up their shoes. So while initial adoption appears overwhelmingly common, pushing beyond surface-level AI usage is a much rarer accomplishment.
The Payoff Problem
Almost everyone is using AI; that much is abundantly clear. But is it actually working? This is where the picture gets a whole lot fuzzier.
Sort of? Maybe? For most firms, the current answer is essentially a shrug.
Only 1 in 5 firms can point to a clear, measurable return on their AI investment. A smaller cohort of about 4% say it hasn’t paid off at all (quantitatively or qualitatively). The most common answer to the ROI question, with a vote of 52%, is that the value of AI is “promising but hard to quantify.”
Just 1 in 5 firms can point to measurable ROI from AI.
They’re not necessarily disappointed, but they aren’t logging their AI investment as a win yet, either. In some ways, the ambiguity is tougher to face than a flat-out failure. If they knew for sure that AI wasn’t panning out the way they hoped it would, they could take action (e.g., remove or change tools, adjust processes, or press pause). But ambiguity just lingers, silently consuming budget and attention while they wait to see if it amounts to anything.
Which raises the obvious question: what’s standing between all these firms and a return they can actually measure?
The answer might surprise you.
The Time Tax
If you’ve ever had an honest conversation with your peers about why their firms aren’t getting more out of AI, then you’ve probably heard all the reasons bandied about the industry most frequently: it’s too expensive, the team won’t get on board, or we’re not convinced it’ll actually pay off.
The data tells a different story.
When asked what’s really holding firms back, very few survey respondents called out cost or team resistance. And “unclear ROI” (i.e., the fear that AI might not be worth the trouble) came in dead last, selected by just 2% of respondents.
The runaway winner? Time. Selected by 41% of survey-takers, time was the top-cited AI barrier by more than double.
Time is the #1 barrier to getting more value from AI (41%). “Unclear ROI” is the last (2%).
What’s the biggest barrier to getting more value from AI at your firm?
It’s the central irony of AI in small firms right now. The technology is supposed to give people their hours back, but very few accounting professionals can spare an hour to learn and adopt it in the first place.
“Things are evolving SO FAST that it is almost impossible to keep up,” one firm owner told us. “It would be a full-time job.”
So as firms struggle to stay on top of their current workload, AI stays half-implemented, and the full return stays just out of reach. Except, that is, for the top echelon of firms that have found a way through.
What the 11% Do Differently
Divided by adoption stage, the ROI data tells a more useful story. Among firms that are still “crawling” with AI (meaning a few people use it on an informal, occasional basis), just 4% report measurable ROI. Among firms that are “running,” however, that number jumps to 56%. That’s a nearly 15-fold increase, and it extends to time savings as well: “running” firms are far more likely to save seven-plus hours a week than their peers who are only dabbling.
Firms “running” with AI are nearly 15x more likely to report measurable ROI than firms where only a few people dabble.
ROI and time savings by adoption stage — Crawling
The takeaway isn’t simply that firms should “use AI more.” It’s that depth pays in a way surface-level AI use doesn’t. And depth, the data shows, comes from structure.
Firms that have invested in organized skill-building (e.g., formal training, an internal champion, or vendor-led sessions) report clear ROI at more than double the rate of firms that haven’t focused on internal education. Firms that have a formally documented AI policy are even more likely to have seen a return.
The tools firms select also appear to factor into their overall AI value realization. Right now, 96% of AI users lean mainly on general-purpose chatbots, while only 20% use anything purpose-built for accounting work. That means most AI is happening in a separate tab, off to the side of the systems a firm actually runs on—which is exactly where value tends to get lost.
Firms seeing the greatest returns are bringing AI into the workflow itself instead of bolting it on from outside. And more and more accounting tools are starting to offer their own AI features. Financial Cents, for example, has a suite of built-in AI agents integrated directly into the core practice management platform. When AI lives where the work already happens, with no extra tool to manage, leveraging it consistently becomes more doable.
But even the most advanced firms approach AI with a healthy dose of caution.
The Nonnegotiable Human Element
For all the optimism in the data—and there’s plenty, with more than 7 in 10 professionals feeling positive about AI’s role in the profession—one conviction is nearly unanimous.
Ninety percent of respondents agree that human judgment matters more, not less, as AI spreads. It’s the single most agreed-upon statement in the entire survey report, consistent across every role, firm size, and adoption stage.
90% of accounting professionals agree that human judgment matters more, not less, as AI spreads.
The numbers back up the sentiment. Only 19% of firms trust AI output enough to use it with limited review, and that holds true even for firms “running” AI across the whole organization. Close human oversight isn’t a phase firms expect to outgrow as the technology progresses. Instead, they are designing their work around it.
“If your GPS tells you to drive into a lake, do you drive into the lake?” one firm owner asked, referencing blind acceptance of AI output.
Another reinforced that accounting professionals, not their tools, remain responsible for any work with their name on it—regardless of what tools assisted them with the work.
“The liability always stays with the human,” said that firm owner.
A third pointed out that a tool can provide a wrong answer just as confidently as a right one.
“AI can produce answers that sound extremely confident, but confidence doesn’t always equal accuracy,” that firm owner explained.
AI Adoption Does Not Equal AI Mastery
Which brings us back to the diary.
The AI story in small firms isn’t one of resistance or reluctance. Most hopped on the AI train early and enthusiastically, largely on their own initiative. But few have built the foundational structure—the training, policies, and workflows—to turn that enthusiasm into measurable value.
And that, more than any single tool or trend, is what separates the “running” firms from those that are stuck crawling.
Curious what else these nearly 500 survey respondents had to say about the current AI environment in accounting? The full data report dives into much greater detail, including:
- Role-by-role breakdowns for owners, service line leaders, ops leaders, and admin staff;
- Concrete ways AI will change the industry in the next decade; and
- What accounting professionals are most worried about.
Read the full findings in The State of AI in Bookkeeping & Accounting: 2026 Report.
Check out the full report →
