Why Law Firms Can’t Measure AI ROI (And What Has to Change First)

Most law firms can’t measure whether their AI tools are actually working, and the reason isn’t the technology. It’s that the financial baseline needed to compare before and after doesn’t exist. Cost per matter, billable hours per attorney, margin by practice area, none of it is tracked in most small firm P&Ls. Without that baseline, “AI is saving us time” is a feeling, not a fact. Here’s what has to be in place before that question can be answered honestly.

Why can’t most law firms measure whether their AI tools are actually working?

Because there’s nothing to measure against. According to Thomson Reuters’ 2026 AI in Professional Services Report, only 18% of firms collect any ROI metrics on their AI tools at all. That’s not a sign of laziness or disinterest. It’s a sign that most firms adopted AI faster than they built the financial infrastructure to evaluate it.

A managing partner can tell you a tool has been activated. Almost none can tell you whether anyone’s work actually changed because of it, or whether that change moved the firm’s numbers. The gap isn’t unique to any one firm. It’s structural.

What financial data does a law firm need before AI ROI measurement is possible?

Three numbers, specifically: cost per matter by practice area, billable hours per attorney before AI was introduced, and gross margin by practice area.

Without cost per matter, you have no way to know if AI actually cut your production cost or just quietly shifted work into your attorneys’ evenings. Without a pre-AI baseline on billable hours, “we’re saving time” is anecdotal, not measured. And without margin by practice area, you can’t tell whether efficiency gains in one type of case are improving profitability or just getting absorbed by other costs.

None of these numbers are exotic. They’re standard CFO-level tracking. Most small and mid-sized firms simply haven’t built them yet, because nobody was asking these questions before AI made the spending visible on the P&L.

Why can’t a bookkeeper or CPA build this baseline for you?

It’s not a matter of skill. It’s a different job.

A bookkeeper’s role is to record what already happened: revenue in, expenses out. A CPA typically engages once a year, at tax time, to confirm the numbers are accurate for filing. Neither role is built to construct forward-looking, per-matter cost structures, or to connect AI efficiency to billable hour math, lockup, or practice-area profitability across a full year.

That work, building a financial baseline specific to how a law firm actually earns money, is a CFO function. It’s not a criticism of bookkeepers or CPAs. It’s simply outside what either role is set up to do.

What does it actually look like when a law firm has the financial infrastructure to measure AI impact?

It looks like monthly reporting that shows cost per matter trending by type, margin by practice area holding steady or shifting, and lockup tracked against any workflow changes AI introduced. With that in place, a renewal decision is based on a number instead of a feeling. A pricing conversation has something to stand on. A headcount decision gets made against real margin visibility, not instinct.

Cathcap builds this as part of the Annual Profit Plan, which establishes baseline cost-per-matter and practice-area profitability in the first phase of the engagement. One attorney client who built this kind of financial visibility into their planning grew from $5.5 million to $18.3 million in revenue over five years. The growth wasn’t about AI. It was about being able to see, clearly, which parts of the business were actually working, and making decisions from there. That’s the same infrastructure that makes an honest AI ROI conversation possible.

What should a managing partner do right now if they can’t answer the AI ROI question?

Stop adding tools, and start building the baseline. The tools were never the problem. The missing infrastructure is.

If your firm is spending $2,000 to $5,000 a month on AI subscriptions with no way to measure what any of it is producing, that’s already close to what a Foundational CFO engagement costs, with none of the visibility. A full-time CFO runs $393,377 a year on average. Cathcap’s Annual Profit Plan builds the same financial baseline at a fraction of that cost.

If you can’t answer the AI ROI question today, that’s a good place to start a conversation.

What to Do Next

1. Pull cost-per-matter data by practice area for the last 12 months, even if it’s rough.
2. Identify each attorney’s billable hours before your first AI tool was introduced, as a comparison point.
3. Check whether gross margin by practice area has shifted since AI adoption, or whether no one has looked.
4. If any of these three don’t exist yet, that’s the real starting point, not another tool evaluation.