A while back, I joined an ILTA panel about chasing AI ROI. During prep, as I explained how returns could be tracked, someone described my focus as the mechanics of ROI.
The comment stayed with me—not because the mechanics were being dismissed, but because it raised a more interesting question: where do the mechanics end and the harder organizational work begin?
I keep returning to that exchange because the mechanics do matter—just not for the reason I first thought.
The panel spent real time on adoption numbers: seats filled, licenses activated, prompts run. We called them what they are—proof that a tool is being used, not proof that the work has become better, faster, cheaper, or more valuable to the client.
Nobody was arguing for adoption over impact.
The harder question was why firms still struggle to move from one to the other.
The billable hour doesn’t explain everything
The usual explanation is the billable hour.
It has merit. When revenue remains tied to time, efficiency creates complicated incentives. And anyone who has worked with realization data knows how difficult it is to isolate the effect of a single technology investment. Accounting teams make analyzing realization rates look far easier than it is.
But the billable hour cannot be the whole explanation.
Every large firm already has work that does not operate entirely on hourly economics: fixed-fee matters, flat-rate arrangements, and practices that have spent years managing delivery costs and margins.
If compensation were the decisive obstacle, those practices should be clearly ahead. AI-driven efficiency should flow more directly into their economics.
There is movement in areas such as IP and labor and employment, but fixed-fee practices are not consistently leading the transformation. They encounter the same difficulty as the rest of the firm.
Something else is missing.
The signals already exist—they just don’t connect
It would be too simple to say that nobody is looking for opportunities.
People see them constantly.
A relationship partner hears a recurring client frustration. A pricing lead notices that a matter type is becoming difficult to deliver profitably. A BD lead sees demand forming around a new service. An innovation team spots a workflow that could be redesigned. A CTO sees that a technical capability has just become viable.
The problem is not an absence of signals.
It is that they remain scattered across functions, practices, and client relationships, with no repeatable path from signal to opportunity, decision, and funded action.
BD hears the client. Pricing understands the economics. Technology evaluates feasibility.
Nobody holds all three perspectives at once, so the decision about where to go deep is often shaped by whichever voice is loudest or whichever pilot already has momentum—not by someone who can see the whole opportunity.
Without a process that connects those functions, a promising idea produces a conversation, perhaps even a pilot, but not a business investment with a named owner and an intended return.
A CTO can assess feasibility, security, architecture, and integration. But the CTO cannot decide alone which client problem is commercially worth pursuing.
That judgment depends on signals that sit outside IT: client demand, delivery economics, practice strategy, and partner commitment.
Asking the technology function to create business value without the business making the underlying business choice is asking it to perform a role it was never given the authority to perform.
Table stakes and strategic investment are different decisions
One argument against measuring ROI is that AI has become table stakes: firms need to invest whether or not they can attribute a return to each tool.
That may justify the baseline investment. It does not remove the need to make strategic choices above that baseline.
Table stakes means investing broadly: the enterprise license and standard platform rolled out firm-wide because not having them is no longer a serious option.
That investment is necessary. It gives people access and keeps the firm from falling behind.
But broad access is not a strategic choice.
By design, an enterprise rollout distributes capability evenly. No practice receives meaningfully more than the baseline unless someone makes a separate decision.
Pursuing a specific opportunity is the opposite move.
It means choosing one practice, one client problem, or one service idea and backing it more heavily than the rest. It means committing technology, lawyer time, process redesign, training, data work, and pricing support before the return is certain.
A firm can become very good at enterprise deployment without ever becoming capable of making that kind of concentrated bet.
Rolling out access is primarily an infrastructure and procurement decision. Backing a specific opportunity is a judgment call—a different capability entirely.
This is the same argument I made in an earlier blog post, approached from a different direction: firms are built to spread AI investment evenly across practice groups, not to concentrate it.
Table stakes is that instinct made concrete. It is what spreading the bet looks like in budget terms: access for everyone, advantage for no one in particular.
It is also the default because it is fair, defensible, and does not require anyone to say that one practice should receive more attention and investment than the other fourteen this year.
Agreeing that firms should concentrate their investment more than they currently do is the easy part.
The more difficult question is what happens once a firm tries.
Pick one opportunity and go deep has its own failure points, and they do not become visible until the firm moves beyond agreeing with the principle.
Finding the opportunity is only the first decision
Even when a firm connects the signals and identifies a credible opportunity, a second problem appears.
Someone has to act on it.
Suppose a BD or practice leader identifies a service that was not previously economical to deliver but has become viable with AI.
Developing it properly could easily require an investment north of $100,000 once the firm accounts for integration, data preparation, subject-matter expertise, workflow redesign, testing, risk review, and go-to-market support.
For most firms, that is significant enough to require more than one enthusiastic practice leader.
It requires a deliberate decision from someone with budget authority who is willing to place the bet before every uncertainty has been resolved.
Without that decision path, the person who identifies the opportunity simply discovers something the firm is not equipped to pursue.
Repeated often enough, that experience may teach people to stop bringing opportunities forward.
ROI starts before the investment is approved
This is where the mechanics return to the argument.
At Organizing4Innovation, we use the RISE methodology—Returns, Investments, Support, and Expectations—to define what an initiative is intended to deliver and what it will take to deliver it before the initiative is approved.
What return are we pursuing?
What investment will it require?
What support must be in place?
What should we expect, and by when?
Answering those questions up front does not remove uncertainty.
It makes the uncertainty governable.
It also gives the firm a basis for deciding whether to continue, redirect, expand, or stop, rather than allowing even a technically successful initiative to drift because nobody agreed on what success was supposed to look like.
Access is not advantage
Law firms do not mainly lack AI tools, ideas, or even ROI formulas.
They lack a repeatable path for turning a promising opportunity into a deliberately funded bet.
Table-stakes investment proves that a firm can give technology to everyone. It does not prove that the firm can choose one opportunity, back it more heavily than the others, and hold someone accountable for the result.
That second capability—not the first—is what turns AI spending from a shared cost into a competitive advantage.
The signals are already there.
The question is whether the firm can bring them together, make a deliberate choice, and fund it.
Rise Above AI Chaos: A Business Fable About Leading Organizations Through the AI Revolution