Before Testing AI, Define the Change You Need

Law firms are experimenting with AI left and right, but many haven’t defined what it’s supposed to improve. Without a clear understanding of the problem, the workflows involved, and the outcomes they want to achieve, firms risk investing in tools that may work technically but fail to deliver meaningful, sustainable value in practice.

A smiling young man with dark curly hair, wearing a blue patterned shirt, exudes confidence as an author for PointOne.

Katon Luaces

A brass drafting compass and ruler on blank paper over a travertine slab in warm low sunlight

In this article

Title

The legal space used to be one that resisted modernity, content with long-lived traditions in how law was practiced. In 2026, 94% of law firms say they are using or exploring generative AI tools, up from 80% last year. For the first time, emerging technology has become the primary focus of law firm investments.

One would think that by now, nearly four years after the modern generative AI boom, firms would already be reporting the results of those investments. One would also assume that this is especially true for boutique firms, where there are fewer infrastructure barriers to new tool integration, so they should be moving faster and seeing results sooner. Instead, only about a third of solo and small firms using AI report an associated revenue increase. Many even said that AI has had no impact on revenue yet or that it is too early to tell.

Approximately 42% of professional service organizations do not measure ROI for their AI tools, and a further 40% don’t know whether they do, leaving just 18% who actually track metrics. Intent informs what gets measured. Some firms are adopting AI to keep pace with the market or to market themselves to clients. Others are adopting simply out of fear of falling behind. Fewer are adopting it to drive genuine innovation.

Bar chart: 42% of professional services organizations don't measure AI ROI, 40% don't know if they do, and 18% track metrics. Source: Accounting Today.

What many firms overlook is that AI is not a software purchase where a license alone will produce results. AI tools deliver value as part of a broader workflow strategy.

Measuring return on investment (ROI) requires a baseline to measure against. Testing an AI tool is an experiment. Experiments need a targeted, measurable hypothesis. A control group. Independent and dependent variables. Testing too many tools at once, without planning how to measure them or checking whether they overlap enough to nullify each other’s impact, makes the results impossible to read.

At the most basic level, firms need to answer:

What are we trying to improve, and how will we know it improved?

Bárbara Gondim da Rocha, Director of Operations at RMS, has worked through this question from the implementation side. Her experience shows that the pressure to adopt AI quickly can work against getting accurate, usable results. To stay competitive, law firms need to decide how much speed they’re willing to trade for accuracy, and that starts with defining the problem they are trying to solve before choosing a tool.

The problem law firms face is rarely a shortage of tools

More tools can lead back to the same unanswered question

Try to flood your firm with too many tools and you will end up back where you started, or even worse off. Attorneys aren’t sure which tool to open for which task, the same work gets done twice in two places, and the firm pays for licenses that overlap. At least the starting point before all that new technology was familiar.

When a new tool appears, the easiest response is to test it. A competitor announces a rollout, a vendor offers a demo, and the firm adds another product to the list. There is a strong desire to explore what’s possible with AI, driven by excitement and, to some extent, the fear of being left behind. But that enthusiasm doesn’t always come with the same level of attention to what will actually make these tools effective.

The problem is that firms are testing tools without first mapping the workflow those tools will sit inside, or in some cases replace. Some saw this coming. At the start of 2026, a law firm partner predicted that buyers would see overlapping products built on the same LLMs, and that actual adoption would not be extensive even after purchases were made.

Even when a tool works exactly as intended, it may add little value if the underlying process remains inefficient or the firm isn’t prepared to incorporate it. We’re seeing firms starting with the technology and then looking for a problem to solve, rather than understanding the problem first and evaluating whether AI is the right answer.

The cost is not just time and wasted resources, but trust as well. When expectations aren’t matched by practical results, frustration grows, making it harder to get people on board with the next initiative. AI solutions require continuous monitoring. When something goes wrong, firms need to identify the issue, understand its root cause, and adapt.

What firms measure shows what they’re aiming for

Firms have a habit of tracking inward results. Among organizations that measure AI ROI, the most common metric is internal cost savings (77%), followed by employee usage (64%) and employee satisfaction (42%). Cost savings are helpful in proving that the tool is making work cheaper to produce and improving margins, but it doesn’t connect how the direct results delivered ties to the revenue growth.

Take AI timekeeping. Counting only the internal hours saved due to the AI tool misses the bigger picture: time that used to go unbilled is now captured, narratives are clearer, and clients can see exactly what they’re paying for. Those changes improve client retention and the chance to win more work from the same client. A tool can save money internally, but its value should always also measure what it does for the firm’s clients and growth. And if the change isn’t clearly defined from the start, usage is the only thing left to count.

Look past the task to the workflow and the outcome

Individual speed doesn’t add up to firm-wide results

The gap between AI activity and real results isn’t unique to law. In a 2026 global survey of business leaders across industries, about 80% said AI improves their productivity, but just 37% reported that it contributes positively to their company’s earnings. Individuals are saving time, but their organizations aren’t changing, because the work around those tasks stays the same.

Bar chart: about 80% of business leaders say AI improves their productivity, while 37% say it contributes positively to earnings. Source: McKinsey 2026 global survey.

In legal, AI use is still concentrated at the task level, particularly in drafting, legal research, document summarization, and contract review. People are saving time, but the workflow around those tasks stays the same.

The three levels must be separated:

Diagram of three levels. Task: one action, like summarizing a court decision. Workflow: the chain of handoffs from when the decision arrives to when someone can act on it. Outcome: what the firm or client gains when that chain works better.

An AI product can improve the speed of a task, but if the surrounding workflow stays the same, the bottleneck will just simply move elsewhere. For example, generating hundreds of time entries faster creates little value if the process for reviewing and releasing them remains unchanged.

The organizations seeing results redesign the work

Most firms fall into the trap of buying AI the way they buy traditional software: sign the contract, install it, train people, and expect results. Firms should take it as a warning sign if AI is introduced but the workflow remains unchanged.

The small group of companies getting real financial returns from AI stands out for one change. Nearly 75% of these high performers report redesigning workflows through AI use, compared with 25% of others. They fundamentally redesign workflows enabled by AI rather than inserting AI into existing ones.

Bar chart: nearly 75% of AI high performers report redesigning workflows through AI use, compared with 25% of all other companies. Source: McKinsey 2026 global survey.

The new workflow must be mapped not after a tool is purchased, but during the evaluation stage, where firms can question whether the workflow itself still makes sense, what should change, who should be involved, and how success will be measured.

At high-volume litigation firms like RMS Advogados, AI makes it possible to analyze thousands of decisions, identify patterns, and turn that information into insights that inform litigation strategy and portfolio management. The evaluation process requires involving legal, operations, data, and technology teams from the start, when the problem is being defined, rather than only at deployment.

The right measurement depends on how the firm creates value

Firms create value in different ways, and that should shape both what they buy and how they measure it. A firm that buys out of fear of missing out or for marketing tends to measure what’s easy to count: how many tools it has, how many people log in, how many announcements it makes. A firm buying to innovate measures what changes in the way it earns revenue, and that depends on its business model.

Hourly billing. When revenue is directly generated by the number of billable hours worked, saving time isn’t automatically a gain. If an attorney finishes a task in half the time, the firm bills half as much unless the saved time goes into other billable work. Hourly firms tend to get the most value from AI that protects revenue, like capturing work that currently goes unrecorded or reducing write-offs.

Per-case and portfolio fees. RMS Advogados works differently. In high-volume litigation, a firm may handle thousands of similar matters for one client, paid per case, per portfolio of cases or at specific stages of a case. Any single case may carry modest stakes, but managing the portfolio carries enormous ones.

Today, identifying trends across these cases often requires manually reviewing decisions and recording details in spreadsheets. AI offers the opportunity to analyze thousands of decisions, identify patterns across courts, judges, and legal issues, and turn that information into actionable insights early enough to inform strategy.

For RMS, success depends on accuracy, consistency, speed, and cost efficiency. Processing thousands of documents can become expensive, particularly when AI extracts information that isn’t relevant. Firms need to be intentional about what data they capture and why. The real measure is whether those insights improve decisions and client outcomes at a sustainable operational cost.

Transactional and fixed-fee work. A contract-heavy practice billing flat fees keeps whatever it saves, so its focus might be margin per deal: comparing terms across hundreds of agreements faster, without adding risk.

Strategy is the strongest predictor of results

The same tool can be a strong investment for one firm and a wasted one for another. Legal organizations with visible AI strategies are twice as likely to experience revenue growth from AI adoption as those taking more informal approaches. Yet only 22% report having a visible, defined AI strategy.

Two statistics: legal organizations with a visible AI strategy are 2 times as likely to see revenue growth from AI, and 22% of legal organizations report having a visible, defined AI strategy. Source: Thomson Reuters via LawNext, 2025.

A strategy doesn’t need to be complicated. It should be the opposite: clear, targeted and measurable.

A simple test applies: if someone asks what the tool is meant to improve and the answer takes a paragraph, the problem hasn’t been defined narrowly enough.

For example, if a firm is struggling with write-offs and evaluating AI timekeeping tools to reduce them, it should track its write-off rate before and after the test. Externally, it should track how often clients question or reject entries.

One central focus should cut through the noise of spillover “wins” that will deflect from the question the firm set out to answer.

Looking to improve your firm’s efficiency?

Book a demo to see how PointOne can help maximize billable time and cut out admin at your firm.

Thank you to Bárbara Gondim da Rocha for collaborating with me on this piece and sharing her perspective on AI implementation in law firms.

Thanks for reading,
Katon Luaces
Co-Founder & CEO of PointOne

PointOne's "Attorney Intelligence" digital product logo, featuring a sleek black abstract symbol and bold text, conveys.

Industry news and insights on the future of legal AI, delivered weekly to your inbox.

The PointOne logo, a sleek black abstract shape resembling a stylized "L" or a data flow, next to the bold text "Attorney.

Industry news and insights on the future of legal AI, delivered weekly to your inbox.

The PointOne logo, a sleek black abstract shape resembling an elongated "P" or a stylized "1," accompanies the bold text.

Industry news and insights on the future of legal AI, delivered weekly to your inbox.

Newsletter

© 2026 PointOne Technologies, Inc.

AI Timekeeping and Firm Intelligence

Newsletter

© 2026 PointOne Technologies, Inc.

AI Timekeeping and Firm Intelligence

Newsletter

© 2026 PointOne Technologies, Inc.

AI Timekeeping and Firm Intelligence