A few weeks ago, I wrote that for law firms billing by the hour, the return on investment from AI was really a loss on investment. LOI, not ROI. If AI does what the vendors say it does, it does work that associates and paralegals used to bill. That would mean fewer hours and less revenue. It’s simple math. And as us lawyers all know, math sucks.
A few weeks before that, I also wondered whether all of us, me included, might wind up being wrong about AI killing the billable hour. The hard data kept stubbornly pointing the other way.
Now Clio has weighed in with its 2026 Legal Trends Report. And at first glance, it looks like AI use may indeed result in more billable hours, not less. Hmmm. Me, completely wrong? Not so fast.
Let’s look at the numbers.
The Numbers
Clio projects that average utilization, the share of an eight-hour day recorded as billable, will hit 42.2% in 2026. That’s up from 39.1% in 2025. If it holds, the 3.1-point jump will be the biggest single-year increase in the report’s decade-long history.
And the firms using AI aren’t billing less. They’re billing more. Clio compared firms that adopted its AI tools with similar firms that didn’t. The adopters recorded about 8% more revenue per responsible lawyer, roughly $1,772 a month.
Spread that out over 12 months and 10 lawyers, and you’re talking real money. And their utilization ran 1.2 points higher. Their collection rates ran 2 points higher. Firms using AI on both the legal work and the business side did even better: 10% more revenue and 6.5% more billable hours.
Clio even has a name for it: the AI paradox. Individual tasks get faster, yet firms record more billable time overall.
So is AI actually increasing billable hours? Maybe. But you have to ask why the hours went up. Clio, to its credit, offers three explanations. They don’t necessarily all point in the same direction.
Lawyers Are Getting To More Work
I talked with Josh Lenon, Clio’s lawyer in residence, at ClioCon26. His take: firms aren’t necessarily taking on new matters, and they aren’t billing existing clients more. They’re getting to work they otherwise would have put off to next week or next month. That’s work that would have previously been done if only they had had more time.
The data seems to back him up. AI adopters recorded billable work across 6% more of their existing hourly caseload each month than comparable nonadopters. New matters at those firms were about 5 points more likely to see their first billable work within 14 days.
That rings true. I often asks rooms full of lawyers when I present whether they have time to do everything they’d like to do for their clients. Blank stares. No hands go up. Every practice has a backlog. So do most lawyers. If AI lets lawyers chip away at it, more hours get billed, and clients get served sooner. That’s value. But I wonder how long that can be sustained before the backlog of work is gone.
Boring AI Is Doing The Heavy Lifting
The second explanation is less glamorous but makes sense. AI is cutting the administrative drag that keeps lawyers from doing billable work.
As might be expected, the biggest utilization gains went to solos using both Clio’s legal-work AI and its practice-management AI: 9% higher than comparable nonadopters. Solos and lawyers in small firms often have to do their own admin. Take some of it away, and there’s more time to bill.
Lenon gave me an example. The most-used AI feature in Clio’s practice-management software? Expense entry. You take a picture of the receipt, and the AI handles the rest. Half of nonlawyers use it. So do 34% of lawyers. If I were practicing, I’d use that one in a heartbeat.
Lenon also told me that much of this year’s gain came before Clio’s main substantive AI tool had been out long. So that should mean next year’s jump could be bigger. But think about what it implies. A portion, and likely a large portion, of this year’s ROI came from AI that handles chores, not AI that does legal work. There’s only so many chores.
Better Timekeeping
Here’s another explanation and one that gave me pause. Clio acknowledges that some of the difference may reflect work lawyers were already doing. But now they record the time on that work more consistently. Because the work and the timekeeping live in the same system. Timekeeping is becoming more automated; time leakage is getting plugged.
That’s a genuine benefit. Every lawyer who has reconstructed a week of time entries late on a Friday afternoon wanting to get home knows how much slips through the cracks. But it isn’t AI making lawyers more productive. It’s AI making them better timekeepers. And once all that previously unrecorded time is captured, the gains from better timekeeping will run their course.
Some Caveats
Clio is quick to point out that the Legal Trends Report is not just a survey. Lenon rightly stressed that the operating data comes from tens of thousands of firms. The AI analysis compared adopting firms against a matched control group. And it tracked each against its own pre-adoption performance. That’s a hell of a lot more rigorous than most vendor surveys, some of which don’t even tell you how many people responded.
Still, as a litigator, I know it’s important to know the caveats. And there are some that Clio candidly admits:
• Clio is measuring its own customers using its own products.
• The analysis is observational. Clio’s methodology says plainly that it doesn’t establish causation.
• The 2026 utilization figure is a full-year projection based on data through the second quarter.
Lenon was transparent on another point: Clio has only about a year of data on its AI tools. That isn’t yet enough to say, for example, that drafting time per matter has dropped while the number of tasks has gone up. The data, he said, just isn’t statistically significant yet.
And, of course, Clio’s data cannot capture every factor affecting performance like management practices, demand, staffing, pricing, or other technology firms use outside of Clio.
Was I Wrong About LOI?
Buried in the report is an important finding. In a separate survey analysis, holding the extent of AI adoption and other factors constant, firms billing by the hour were the least likely to report that AI improved revenue.
Firms using alternative fees were the most likely. Clio’s own conclusion is that AI can produce gains under hourly billing, but not automatically. All of which is to say firms have to redeploy the freed-up time or change how they price the work. That’s going to go down hard for many firms.
In September, I offered three possibilities: 1. We were all wrong. 2. We underestimated the billable hour. 3. Change just hasn’t arrived yet. The answer looks like the second and third together. Right now, law firms are trying to deal with the billable hour the way they always try to do with disruptive change. They try to absorb it without letting it disrupt what they care about most. Things like where their money comes from.
My Above the Law colleague Joe Patrice recently suggested that AI has become the billable hour’s bodyguard. Clio’s data suggests it may become its bookkeeper.
Clients Are Doing The Math Too
Here’s another issue. Clio also surveyed consumers, and they’re viewing AI very differently.
Fifty-seven percent say the same work delivered faster should cost less. Only 13% would pay more for it. Forty-seven percent say AI reduces the value of legal services. And when asked the single most important reason to hire a lawyer on a routine matter, speed, the very thing AI can supply, was ranked last. Dead last at 3%.
Corporate clients are no different. Last week I wrote about in-house counsel: Fifth-nine percent hadn’t noticed any savings from their outside firms’ AI use, and only 7% had seen lower matter costs.
And a lawyer quoted in the report stated the obvious: if they can get a case to settlement in 30 hours instead of 100, “just do the math.” They’re right. And their clients are doing it too.
Lenon offers a counterpoint. He sees collection rates as a rough proxy for client satisfaction. If clients pay without fighting the bill, they’re probably happy. AI adopters, remember, collect 2 points more.
Fair enough. But paying today’s bill isn’t the same as accepting tomorrow’s price. Firms are billing more hours at the very moment clients are starting to conclude AI should mean fewer. That can’t last forever.
The Next Question
Lenon told me Clio wants to ask a new question next year: do lawyers find themselves working nights and weekends? Then it will compare AI users with nonusers. That may be what finally solves the paradox. Is AI freeing up lawyers’ time, or just filling it?
So is AI killing the billable hour? At least for now, it isn’t. So I may have been wrong about AI and the billable hour. For the time being. But AI is really just making the hours easier to find. The real question is how long clients will keep paying for what it finds.
Full disclosure: Clio paid my travel expenses and the registration fees for me to attend ClioCon26. But they certainly didn’t try to tell me what to write or when. I can assure you if they did, it wouldn’t have done any good.
Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.
The post Clio Says AI Is Boosting Billable Hours: Maybe I Was Wrong. Or Maybe Not. appeared first on Above the Law.
A few weeks ago, I wrote that for law firms billing by the hour, the return on investment from AI was really a loss on investment. LOI, not ROI. If AI does what the vendors say it does, it does work that associates and paralegals used to bill. That would mean fewer hours and less revenue. It’s simple math. And as us lawyers all know, math sucks.
A few weeks before that, I also wondered whether all of us, me included, might wind up being wrong about AI killing the billable hour. The hard data kept stubbornly pointing the other way.
Now Clio has weighed in with its 2026 Legal Trends Report. And at first glance, it looks like AI use may indeed result in more billable hours, not less. Hmmm. Me, completely wrong? Not so fast.
Let’s look at the numbers.
The Numbers
Clio projects that average utilization, the share of an eight-hour day recorded as billable, will hit 42.2% in 2026. That’s up from 39.1% in 2025. If it holds, the 3.1-point jump will be the biggest single-year increase in the report’s decade-long history.
And the firms using AI aren’t billing less. They’re billing more. Clio compared firms that adopted its AI tools with similar firms that didn’t. The adopters recorded about 8% more revenue per responsible lawyer, roughly $1,772 a month.
Spread that out over 12 months and 10 lawyers, and you’re talking real money. And their utilization ran 1.2 points higher. Their collection rates ran 2 points higher. Firms using AI on both the legal work and the business side did even better: 10% more revenue and 6.5% more billable hours.
Clio even has a name for it: the AI paradox. Individual tasks get faster, yet firms record more billable time overall.
So is AI actually increasing billable hours? Maybe. But you have to ask why the hours went up. Clio, to its credit, offers three explanations. They don’t necessarily all point in the same direction.
Lawyers Are Getting To More Work
I talked with Josh Lenon, Clio’s lawyer in residence, at ClioCon26. His take: firms aren’t necessarily taking on new matters, and they aren’t billing existing clients more. They’re getting to work they otherwise would have put off to next week or next month. That’s work that would have previously been done if only they had had more time.
The data seems to back him up. AI adopters recorded billable work across 6% more of their existing hourly caseload each month than comparable nonadopters. New matters at those firms were about 5 points more likely to see their first billable work within 14 days.
That rings true. I often asks rooms full of lawyers when I present whether they have time to do everything they’d like to do for their clients. Blank stares. No hands go up. Every practice has a backlog. So do most lawyers. If AI lets lawyers chip away at it, more hours get billed, and clients get served sooner. That’s value. But I wonder how long that can be sustained before the backlog of work is gone.
Boring AI Is Doing The Heavy Lifting
The second explanation is less glamorous but makes sense. AI is cutting the administrative drag that keeps lawyers from doing billable work.
As might be expected, the biggest utilization gains went to solos using both Clio’s legal-work AI and its practice-management AI: 9% higher than comparable nonadopters. Solos and lawyers in small firms often have to do their own admin. Take some of it away, and there’s more time to bill.
Lenon gave me an example. The most-used AI feature in Clio’s practice-management software? Expense entry. You take a picture of the receipt, and the AI handles the rest. Half of nonlawyers use it. So do 34% of lawyers. If I were practicing, I’d use that one in a heartbeat.
Lenon also told me that much of this year’s gain came before Clio’s main substantive AI tool had been out long. So that should mean next year’s jump could be bigger. But think about what it implies. A portion, and likely a large portion, of this year’s ROI came from AI that handles chores, not AI that does legal work. There’s only so many chores.
Better Timekeeping
Here’s another explanation and one that gave me pause. Clio acknowledges that some of the difference may reflect work lawyers were already doing. But now they record the time on that work more consistently. Because the work and the timekeeping live in the same system. Timekeeping is becoming more automated; time leakage is getting plugged.
That’s a genuine benefit. Every lawyer who has reconstructed a week of time entries late on a Friday afternoon wanting to get home knows how much slips through the cracks. But it isn’t AI making lawyers more productive. It’s AI making them better timekeepers. And once all that previously unrecorded time is captured, the gains from better timekeeping will run their course.
Some Caveats
Clio is quick to point out that the Legal Trends Report is not just a survey. Lenon rightly stressed that the operating data comes from tens of thousands of firms. The AI analysis compared adopting firms against a matched control group. And it tracked each against its own pre-adoption performance. That’s a hell of a lot more rigorous than most vendor surveys, some of which don’t even tell you how many people responded.
Still, as a litigator, I know it’s important to know the caveats. And there are some that Clio candidly admits:
• Clio is measuring its own customers using its own products.
• The analysis is observational. Clio’s methodology says plainly that it doesn’t establish causation.
• The 2026 utilization figure is a full-year projection based on data through the second quarter.
Lenon was transparent on another point: Clio has only about a year of data on its AI tools. That isn’t yet enough to say, for example, that drafting time per matter has dropped while the number of tasks has gone up. The data, he said, just isn’t statistically significant yet.
And, of course, Clio’s data cannot capture every factor affecting performance like management practices, demand, staffing, pricing, or other technology firms use outside of Clio.
Was I Wrong About LOI?
Buried in the report is an important finding. In a separate survey analysis, holding the extent of AI adoption and other factors constant, firms billing by the hour were the least likely to report that AI improved revenue.
Firms using alternative fees were the most likely. Clio’s own conclusion is that AI can produce gains under hourly billing, but not automatically. All of which is to say firms have to redeploy the freed-up time or change how they price the work. That’s going to go down hard for many firms.
In September, I offered three possibilities: 1. We were all wrong. 2. We underestimated the billable hour. 3. Change just hasn’t arrived yet. The answer looks like the second and third together. Right now, law firms are trying to deal with the billable hour the way they always try to do with disruptive change. They try to absorb it without letting it disrupt what they care about most. Things like where their money comes from.
My Above the Law colleague Joe Patrice recently suggested that AI has become the billable hour’s bodyguard. Clio’s data suggests it may become its bookkeeper.
Clients Are Doing The Math Too
Here’s another issue. Clio also surveyed consumers, and they’re viewing AI very differently.
Fifty-seven percent say the same work delivered faster should cost less. Only 13% would pay more for it. Forty-seven percent say AI reduces the value of legal services. And when asked the single most important reason to hire a lawyer on a routine matter, speed, the very thing AI can supply, was ranked last. Dead last at 3%.
Corporate clients are no different. Last week I wrote about in-house counsel: Fifth-nine percent hadn’t noticed any savings from their outside firms’ AI use, and only 7% had seen lower matter costs.
And a lawyer quoted in the report stated the obvious: if they can get a case to settlement in 30 hours instead of 100, “just do the math.” They’re right. And their clients are doing it too.
Lenon offers a counterpoint. He sees collection rates as a rough proxy for client satisfaction. If clients pay without fighting the bill, they’re probably happy. AI adopters, remember, collect 2 points more.
Fair enough. But paying today’s bill isn’t the same as accepting tomorrow’s price. Firms are billing more hours at the very moment clients are starting to conclude AI should mean fewer. That can’t last forever.
The Next Question
Lenon told me Clio wants to ask a new question next year: do lawyers find themselves working nights and weekends? Then it will compare AI users with nonusers. That may be what finally solves the paradox. Is AI freeing up lawyers’ time, or just filling it?
So is AI killing the billable hour? At least for now, it isn’t. So I may have been wrong about AI and the billable hour. For the time being. But AI is really just making the hours easier to find. The real question is how long clients will keep paying for what it finds.
Full disclosure: Clio paid my travel expenses and the registration fees for me to attend ClioCon26. But they certainly didn’t try to tell me what to write or when. I can assure you if they did, it wouldn’t have done any good.
Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.
