Nov 10, 2026

The AI Efficiency Paradox Reveals What Law Firms Need to Get Right

By Mike Raposa, Chief Revenue Officer, Williams Lea

Law firms are investing in technology designed to make their people more efficient. And at the same time, many continue to measure productivity, profitability, and individual performance through the number of hours worked. There is a contradiction here that the industry can no longer afford to overlook.

At the Thomson Reuters Institute’s 25th Annual Law Firm COO & CFO Forum in Washington, DC, this tension surfaced repeatedly in discussions about AI investment, pricing, client expectations, and the future of law firm operations.

I moderated a panel on The Convergence Era, examining what happens when AI moves across finance, technology, operations, and risk. The discussion reinforced something I have been hearing from law firm leaders for some time: the challenge is no longer simply deciding which AI tools to deploy. It is determining how to change the business around them.

The Forum also raised a broader question. If AI enables firms to complete work faster, how do they translate that efficiency into stronger financial performance, better client service, and a sustainable operating model?

The billable hour is becoming an uncomfortable measure of efficiency

For decades, law firms have relied on billable hours to measure productivity, determine compensation, and assess matter profitability. The model is familiar, relatively straightforward to administer, and deeply embedded in firm economics. AI now complicates that equation.

Consider a task that traditionally requires several hours of associate time. If technology reduces the time needed to complete it, the firm has created capacity. But under an hourly billing model, it may also have reduced the revenue associated with that task.

The firm still carries the cost of its professionals, technology investments, training, and infrastructure. Meanwhile, clients increasingly expect efficiency improvements to be reflected in how legal services are delivered and priced.

That does not mean hourly billing will disappear. The Forum discussions made clear that alternative fee arrangements remain complicated, particularly when matters involve unpredictable variables or changing scope. The more immediate issue is whether firms are using the right measures to understand the value of their work.

AI ROI needs to start with the business problem

The difficulty of measuring AI returns was another recurring theme at the Forum. Firms can track licenses, usage, and technology spending, but those figures reveal relatively little about whether the underlying business is improving. Even adoption statistics can be misleading. One discussion highlighted how high reported usage can conceal significant differences in how professionals actually engage with AI. Some are incorporating it into substantive workflows, while others use it only occasionally.

Neither activity necessarily demonstrates a financial return.

A more effective approach is to define the operational challenges and target desired outcomes before selecting a solution. Take time entry and billing, as an example. Firms may have opportunities to reduce the administrative effort associated with recording time, improve the completeness of entries, and accelerate downstream billing processes. Those improvements can be measured through submission times, correction rates, billing delays, and the effort required to resolve exceptions.

Document production presents another opportunity. Rather than measuring how quickly an AI tool generates an initial draft, firms should examine the entire process, including preparation, formatting, quality control, revisions, and final delivery. This is where the distinction between task efficiency and operational efficiency becomes important. Making one step faster does not necessarily improve the workflow if the remaining steps remain unchanged.

It also means firms need to account for the full cost of AI, including integration, ongoing maintenance, data preparation, governance, and human validation. An impressive demonstration is not a business case.

Operational accountability cannot stop at departmental boundaries

Before the Forum, I wrote about the challenge of determining who owns work when AI crosses multiple departments. The panel discussion brought that issue into sharper focus.

AI-enabled workflows rarely sit neatly within a single function. A solution that improves document processing may depend on technology integration, information security, knowledge management, professional training, and operational delivery. A billing automation initiative may require coordination between finance, IT, practice leadership, and the people responsible for matter administration.

Each function has a legitimate role, but fragmented accountability creates a familiar problem: individual departments can report progress while the overall workflow remains inefficient. The discussion explored different approaches to governance, including centralized decision-making and shared accountability across business and technology leaders. The important distinction is between approving a technology investment and owning the result it is supposed to deliver.

For every significant AI initiative, firms should be able to answer three questions: Who is accountable for the business outcome? Who has authority to coordinate changes across the workflow? And who is responsible for monitoring performance and risk after implementation?

The third question is becoming particularly important. AI platforms are changing rapidly, and new capabilities can introduce additional requirements around data access, security, integration, and oversight. Governance cannot remain a one-time procurement exercise. It needs to operate alongside the technology throughout its useful life.

The workforce question is also an economic question

The Forum’s discussions about AI adoption and talent development exposed another tension. Firms want professionals to work differently, but their training, career development, and incentive structures have not necessarily changed to support that expectation. As AI takes on more routine tasks, firms will need to reconsider how junior lawyers develop professional judgment and how business support teams build the skills required for more analytical work.

There is also a more immediate challenge. Employees are being asked to learn new tools while maintaining existing workloads, often without a clear understanding of how their responsibilities may change. Training alone will not resolve that uncertainty.

Firms need to involve the people performing the work in identifying opportunities for improvement. They should also recognize that successful adoption requires changes to processes, expectations, and sometimes performance measures.

A professional who helps redesign a workflow may create more value for the firm than someone who simply records more hours. The operating model needs to recognize that contribution.

What should COOs and CFOs do in the next 90 days?

The Forum offered no single answer to the challenges facing law firms, nor should we expect one. Different firms have different practice mixes, client requirements, technology environments, and financial priorities. But there are practical steps leadership teams can take now.

First, identify one recurring workflow where the business case is clear. Look for processes that create measurable delays, unnecessary administrative work, or recurring client frustration. Avoid starting with a technology demonstration and searching for a problem afterward.
Second, establish the true cost of that workflow. Include professional time, operational support, technology, quality control, and rework. Without a credible baseline, claims about efficiency will remain difficult to substantiate.
Third, assign accountability for the complete process. Technology, finance, risk, and operations should contribute their expertise, but someone must remain responsible for the outcome.
Finally, measure the improvement and decide what happens next. If the process becomes faster, determine whether that creates additional capacity, improves client service, reduces cost, or strengthens profitability. Those are different outcomes, and leadership needs to know which one it is achieving.

The legal industry will continue investing in AI. The tools will improve, costs will evolve, and new capabilities will challenge established ways of working. The firms that benefit most will not necessarily be those with the largest technology budgets. They will be those that understand how their businesses operate, identify where change creates value, and have the discipline to turn investment into measurable results.

You could argue that this is the real efficiency challenge facing law firms. And it is one that COOs and CFOs are increasingly responsible for solving.