Breakingviews - AI dooms the billable hour – and Big Law earnings

Reuters

a day ago

LONDON, July 2 (Reuters Breakingviews) - Artificial intelligence promises to save time for white-collar workers. If true, that could be bad news for companies that bill clients by the hour. Law firms, auditors and other professional-services outfits might find ways to mitigate the financial hit. But there's no getting around the fact that automation risks devaluing part of the pricey service they're offering.
The 'billable hours' model dates to Reginald Heber Smith, a legendary managing partner of Hale and Dorr between 1919 and 1956, who 'pioneered the rationalization of the modern law firm', as described, opens new tab by its descendant White Shoe outfit WilmerHale. At heart, billing by the hour means getting staff to meticulously track the time spent on projects so that they can invoice clients accordingly. Beancounters and tax advisers at groups like Deloitte are also heavy users of so-called timesheets. About 82% of U.S. law firm partners' work is charged by the hour, Thomson Reuters Institute research shows, while such revenue makes up 65% of income at U.S. audit firms, according to, opens new tab the Association of International Certified Professional Accountants.
Rates can be eye-popping. The most senior partners at elite firms, like Kirkland & Ellis or Quinn Emanuel Urquhart & Sullivan, can bill up to $3,000 an hour. The rate for junior lawyers can be $400, according to LexisNexis's Sean Fitzpatrick, or sometimes much more at White Shoe firms. It's normal, opens new tab in Big Law to charge out juniors at multiples of their salaries, which can be a nice earner for the top partners.
But AI, particularly so-called AI agents which work autonomously, are now threatening to undermine the time-honoured practice. Goldman Sachs analysts estimated, opens new tab in a 2023 report that 44% of legal tasks in the United States could be automated. It might sound like a good thing that an AI agent could draft a non-disclosure agreement in minutes, or instantly synthesise board minutes for an audit. Yet a perverse outcome of the billable hour structure is that being more productive, all else equal, can mean generating less revenue. According to American Bar Association guidelines, opens new tab published in July, lawyers can only charge for actual time spent on tasks, even if AI allows them to perform them faster.
Compounding the problem is the fact that professional-services firms may face a chunky upfront IT bill to get the new software up and running. Only one-third of tax firms surveyed, opens new tab by Thomson Reuters reckon they can directly pass on generative AI investment costs to customers, implying that developing or buying slick new AI agents will initially eat into profit margins.
There are no painless ways to respond to this double whammy. One extreme option, in theory, would be to let AI agents replace a big chunk of the junior staff. Clients pay partners for their wisdom and personal touch, not the grunt work. The implication is that seniors could keep charging themselves out even if the rest of the firm becomes populated by faceless AI robots. And to the extent that some juniors spend time on work that can't be billed, agents could boost profitability. Associates, who are generally younger members of staff, are already shrinking as a proportion of law-firm headcount – to 40% in recent years compared with 45% from 2005 to 2009, according to Thomson Reuters, opens new tab.
One problem with this option, other than its heartlessness, is that firms need a constant pipeline of juniors to repopulate the partnership. Who else will replace the old guard when they finally cash out to hit the golf course full time? It's also far from clear that the hallucination-prone software is ready for the big time, implying that a horde of associates may still be needed to check AI agents' accuracy.
That points to a different solution: moving away from billable hours. It's already happened at the elite strategy consultancies like McKinsey & Company, Bain & Company and Boston Consulting Group, who often charge flat project fees tied to specific outcomes.
Doing the same would flip the AI equation for law and accountancy firms: productivity improvements could boost margins rather than hurt them. There's a precedent in the legal world too: Allen & Overy in 2002 created a subscription-based business called Aosphere, whose lawyers give advice online to 1,200 clients. 'We don't even do time sheets', its website claims, opens new tab. Buyout shop Inflexion and Endicott Capital agreed to invest in the division in 2023 at an unspecified valuation, suggesting that the model may hold some promise.
But it's a different type of service to advising on a complex deal or piece of litigation. The risk is that it will be tough to systematise pricing across the vast variety of projects. Doing so might be easier for beancounters, since audits can in theory share a common overarching process. But no two lawsuits, for example, are the same. Switching to a project-fee approach puts the onus back on professional-services firms to judge how many resources a clients' work will take.
The bigger problem, however, is that automating tasks makes it harder to charge a margin. Under the classic law-firm model, for example, revenue gets split equally three ways between overhead costs, salaries and partner profit. The implication is that seniors should charge juniors out at a minimum of three times their pay.
Yet clients may balk if Big Law tries to apply the same logic to an AI agent. Why should a White Shoe firm add a markup to software that it just bought from someone else? Corporate clients could argue that they could just get their own AI agents instead. It's a management challenge that Hale and Dorr's attorney mastermind Reginald Heber Smith would probably have relished. Solving it will require moving beyond the billable hour.
Follow Karen Kwok on LinkedIn, opens new tab and X, opens new tab.