Accounting Firms Are Buying AI — But Only 1 in 3 Is Making Money From It. Which Stage Are You In?

Published: March 14, 2026 | By: The Crossing Report | 7 min read

---

Summary

The March-April 2026 issue of Today's CPA documents the central problem facing accounting firms right now: most have AI tools, but only about one in three has reached the stage where those tools are generating measurable financial return. The gap isn't capability — the tools work. The gap is adoption stage. This article presents a 5-question diagnostic to identify which of three stages your firm is in, and a specific next step for each.

---

The Adoption Gap Is Not What You Think

The AI in accounting market is projected to reach $10.87 billion in 2026, growing at 44.6% CAGR. Firms are spending. Tools are being purchased. Demos are being run.

And yet: AICPA/CIMA survey data from February 2026 found that only 19% of accounting professionals use AI daily. The Today's CPA analysis puts it bluntly — most of that spending is landing at Stage 1, and staying there.

The issue isn't that the tools don't work. It's that most firms treat AI adoption as something that happens when someone curious tries a new tool. It doesn't scale on its own. Nobody told the rest of the team. There's no workflow. The curious person moves to the next thing. Six months later, the firm has three AI subscriptions it doesn't actively use.

The Today's CPA framework describes three stages of accounting firm AI adoption. Where you are determines what you should do next.

---

The Three Stages

Stage 1: Practical Automation

One or two people are using AI tools — usually for personal productivity. They drafted a few emails with Copilot. They used ChatGPT to write a summary. It worked fine. They'll do it again when they remember.

The defining feature: there is no defined workflow. "Using AI" means "one person chose to use a tool today." Nobody else knows what was tried or what worked. There is no standard process, no second person who can replicate the result, and no way to measure whether it's saving time or generating revenue.

Most firms reading this are in Stage 1.

Stage 2: Workflow Integration

AI is embedded in at least one specific, repeatable task. More than one person on the team uses it. The workflow has a defined structure: input, AI action, human review step, output. Someone can explain it to a new hire.

The defining feature: it's measurable. You could calculate how much time the AI workflow saves per week, per month, or per client. You haven't necessarily done that math yet — but you could.

Firms in Stage 2 are pulling ahead of Stage 1 peers on speed and capacity. They're not yet repricing or restructuring — but they have the foundation to do so.

Stage 3: Strategic Advantage

AI is integrated into how the firm delivers services, not just how individuals work. The efficiency gains are reflected in how the firm is priced, staffed, or positioned to clients. A Stage 3 firm can close books faster, take on more clients without proportional headcount growth, or offer a class of service competitors can't match at the same price point.

According to Today's CPA, firms that reach Stage 3 report a 73% rate of strategic advantage — measurably better outcomes than peers. The AI in accounting market's growth is concentrated in this tier.

---

The 5-Question Diagnostic

Answer these honestly. Your answers tell you which stage you're in.

Question 1: Does at least one person on your team use an AI tool on every workday?

• Yes → proceed to Question 2
• No → you're in early Stage 1. Start with meeting summarization (Fathom, free at fathom.video) — one person, one tool, one week.

Question 2: Can you describe one specific AI workflow — not "we use AI," but a defined sequence of steps — that your team runs on a repeatable basis?

• Yes → proceed to Question 3
• No → you're in Stage 1. Individual tool use without a workflow doesn't scale. Go to the Stage 1 next step below.

Question 3: Has more than one person on your team used AI on client work in the past 30 days?

• Yes → proceed to Question 4
• No → you're in Stage 1, even if one person is using AI regularly. The workflow exists only in one person's head, not in the firm. Go to the Stage 1 next step.

Question 4: Can you measure the time or revenue impact of your AI use — even roughly?

• Yes → proceed to Question 5
• No → you're in Stage 2. The workflow exists, multiple people use it, but you don't have data on the return. Go to the Stage 2 next step.

Question 5: Has AI changed how you price, staff, or structure your services?

• Yes → you're in Stage 3. Go to the Stage 3 next step.
• No → you're in Stage 2, upper end. You have the evidence you need to make structural changes. Go to the Stage 2 next step.

---

Next Step by Stage

Stage 1: Define One Workflow, Make It a Team Standard

The problem is not that AI doesn't work. The problem is that no workflow exists. Fix the workflow before adding more tools.

Pick one task that your team does repeatedly, consumes time, and has a consistent structure. Transaction categorization review, client email follow-up, meeting note documentation, engagement letter drafting — any of these works. Define the task: what goes in, what does AI do, what does a human review, what comes out.

Run it yourself for one week. Then show one other person. Have them run it. That second person is the test. If they can do it from your description, you have a workflow. If they can't, keep clarifying until they can.

The tool recommendation for Stage 1 accounting firms in 2026:

• Fathom (free) — meeting summarization. Every client call gets a summary. The summary becomes the first draft of your follow-up, your time entry, and your engagement update. It's the easiest visible win.
• Microsoft Copilot ($21/user/month, M365 add-on) — email drafting in Outlook. Start with one type of email (monthly engagement update, tax deadline reminder, question response). Draft it, review it, send it. Repeat until it's the default approach.

Stage 2: Measure the Workflow, Then Reprice or Restructure

You have a workflow and multiple people using it. The question now is: what is it worth?

Spend 30 days tracking one metric for your primary AI workflow. Hours saved per week. Client files processed per day. Time entries captured per attorney per month. Any single measurable number.

At the end of 30 days, you'll have one of two answers:

1. The data shows significant time savings or capacity gain — in which case, you have the evidence to reprice, take on more clients, or redirect staff to higher-value work.
2. The data shows minimal impact — in which case, you have identified that this particular workflow is not your leverage point, and you should try a different one.

Either answer is useful. Most Stage 2 firms skip this step and never get to Stage 3 because they don't have the data to justify structural changes.

The tool recommendation for Stage 2 accounting firms in 2026:

• Karbon ($59/user/month) — if client communication management is your bottleneck. Measurable metric: time from client inquiry to response. Build an AI-assisted response workflow and track before/after.
• Ramp Accounting Agent — if your firm uses Ramp for client expense management. Measurable metric: time spent on transaction categorization per month. Early customer data: 90%+ auto-coding accuracy, 3x faster month-end close, 40+ hours/month recovered.
• Canopy ($45-66/user/month) — if document management and client portal workflow is the time sink. Measurable metric: client document processing time per engagement.

Stage 3: Reprice and Position the Efficiency as a Differentiator

You have measurable AI efficiency. You are faster, or more accurate, or can serve more clients. Most of your competitors are still in Stage 1.

Two things to do now:

First, reprice. Eighty percent of US accounting firms plan to raise prices in 2026, according to Ignition's 2026 pricing benchmark. The firms doing this successfully are the ones whose cost structure isn't growing with revenue — because they've replaced some of the growth they would have hired with AI workflow capacity. If your Stage 3 AI use is genuinely freeing up team capacity, you can absorb more clients or raise prices without a proportional cost increase. Run the math before your next client renewal cycle.

Second, name the advantage. When a client asks why your firm processes their close faster, or how you turned around a deliverable in 48 hours instead of a week, you don't have to say "we use AI." You can say: "We've invested in building workflows that let our team focus on the judgment work — the things that actually require an experienced accountant's eye." That is a true statement. It is more compelling than a tool name. It is the differentiated positioning that Stage 3 enables.

---

One More Thing About Stage 1

There is a temptation to evaluate your stage generously. The Today's CPA analysis suggests most firms do.

The clear-eyed test: if the person on your team who uses AI most was not at work for a week, would anyone notice any change in how client work gets done?

If the answer is no — if the AI use lives entirely in one person's habits and nothing would change for clients — you're in Stage 1, regardless of how many subscriptions you have.

That's not a critique. Most firms are there. The gap between Stage 1 and Stage 2 is not a technology gap. It's a documentation and delegation gap: you need a written workflow, a second person who can follow it, and a human review step that ensures quality. That's it.

Build one, run it for a month, measure it. That's how the 1 in 3 firms that are making money from AI got there.

---

Related Reading

• Accounting Firms Are Replacing Hires With AI — And the Math Is Starting to Make Sense
• Basis AI's $1.15B Valuation: What Autonomous Accounting Means for Small Firms
• ChatGPT for Accounting Firms: The Workflows That Actually Save Time
• 88% of Accountants See AI as Transformative — But Only 8% Are Ready: The AICPA/CIMA Data