The Law Society Has Blown a Hole in the Billable Hour

New guidance on charging for AI-assisted work exposes a problem that more time-recording training cannot fix.

The Law Society has released a new practice note on using artificial intelligence in legal practice. It covers a range of issues that firms need to consider as AI becomes a more familiar part of legal work, but there is one section that I’ve found most interesting: charging for AI-assisted work.

The Law Society says that, where fees are based on time, firms should record and bill only the time actually and properly spent on a matter. If AI has been used, a firm should not charge for the time the task would have taken without it.

Lawyers can charge for appropriate time spent operating an AI system, reviewing and verifying its output, making revisions and applying professional judgement, where that is consistent with the client care letter. What they can’t do is charge for hours that were never worked.

That may seem obvious, but it has significant implications for a profession whose business model is still largely built around the billable hour. If AI helps a lawyer complete a piece of work in less time, and that work is charged by the hour, the fee for that work goes down. The firm may have invested in the technology, trained its lawyers to use it and delivered the service more efficiently. Under an hourly arrangement, the client benefits from the time saved through a lower bill, while the firm receives less for the work.

The Law Society hasn’t banned the billable hour, but it’s made the problem with it much more obvious.

AI makes the billable hour’s weakness harder to hide

Let’s take a simple example: a lawyer previously spent 10 hours on a piece of work and charged £300 per hour. The fee was £3,000.

The firm introduces an AI tool. With appropriate use of that tool, including the lawyer’s review, revisions and professional judgement, the same work now takes six hours. At the same hourly rate, the fee is £1,800. The firm can’t add four hours to the bill because that is how long the work would have taken without AI. Those hours were never worked.

Of course, there are other things the firm can do. It could use the four hours to take on more work (that’s a BD issue for another time), increase its hourly rates or change the way the work is staffed. But none of those possibilities changes the calculation on this matter: at the same hourly rate, fewer chargeable hours mean a lower fee.

This problem didn’t start with AI. A lawyer who becomes more experienced and completes a task in less time has always faced the same problem. AI has the potential to make that problem more widespread by changing how quickly lawyers can complete a much larger range of tasks.

More time-recording training won’t solve the pricing problem

One response I expect to see is more time-recording training. For years, firms have tried to improve revenue by making sure lawyers capture all the time they spend working. The thinking is straightforward: if lawyers record more of the time they actually work, the firm has more time to bill.

Firms that charge by the hour must record their time accurately. But what happens when the work that used to take 30 minutes now takes 10? No amount of time-recording training can recover the other 20 minutes. They weren’t spent on the task. The Law Society’s guidance makes clear that they shouldn't be recorded and billed as though they were.

Firms need to think carefully about where they are putting their training effort. A firm can spend time and money teaching lawyers how to capture every six-minute unit, but why bother putting its training effort into a business model that AI is making less dependable? Training should be teaching them how to understand what a client values, scope work and agree prices upfront.

As AI helps lawyers complete more work in less time, teaching them to record every minute they work will do less and less to address the firm’s pricing problem. Firms need to stop looking to time-recording training for the answer and start teaching lawyers how to understand what clients value, scope work and agree a price upfront.

Does checking AI’s work cancel out the time saved?

A common response to the argument that AI will reduce billable hours is that lawyers will spend so much time checking its output that there will be little or no saving.

That may be true for some tasks—in the short term. An AI tool might produce a poor first draft, miss important context or provide an answer that takes substantial time to check and correct. In those circumstances, using it may save very little time, or even take longer. But let’s be careful about turning that possibility into a general rule. The amount of time saved will depend on the task, the tool, the information provided and the review required. Ultimately, knowing how to use the tool will determine how much benefit you get from it. And, over time, AI tools get better, meaning it can do more with the little you can sometimes give it.

Research suggests that savings are possible, although the results vary. One experiment involving 206 law students found that AI-generated highlighting reduced completion time by 30% on the task studied, while AI-generated summaries did not produce the same improvement. The study does not tell us how much time every practising lawyer will save, and it was published in 2024, which is a long time ago in terms of the development of AI. It does show that we need to look at the work being done rather than assume that checking will always cancel out the time saved.

And AI hasn’t created the job of checking work. Partners review work produced by associates. Associates review work produced by trainees. Lawyers check documents, authorities, facts and advice because they are responsible for the work they deliver. Verification has always been part of the job. To suggest verification is a new thing is more an indictment on how we have worked up to now than what AI may be lacking.

AI can introduce different risks. It can, for example, produce a convincing answer that contains an incorrect statement or an authority that does not exist. Lawyers need to carry out the checks appropriate to the work and the tool they have used. The Law Society is clear that professional responsibility remains with the solicitor.

That said, appropriate verification isn’t a replacement for time that would previously have been spent producing the work. It is work that must actually be carried out. There is already evidence that some lawyers add time to their timesheets that they have not worked. In a 2024 anonymous RollOnFriday poll, 35% of respondents admitted to padding their time, with 13% saying they did it regularly. For anyone who says padding time doesn’t happen, this poll should be enough to take the issue seriously.

The poll didn’t look specifically at AI verification. As AI reduces the time needed for some tasks, though, ‘verification’ must not become a convenient label for time that was never spent. If a lawyer spends two hours properly reviewing AI-assisted work under a time-based arrangement, those two hours may be chargeable where that is consistent with the client care letter. If the review takes 30 minutes, a lawyer cannot record two hours because that is what the task would have taken before AI was used.

Firms need to be clear about how they will work going forward, particularly where lawyers are still expected to meet billable-hour targets that were set for a different way of working.

What does an hourly rate tell the client?

The Law Society’s practice note also asks firms to consider how their commercial models and billing arrangements might reflect changing ways of delivering legal services, including how they communicate the value of AI-enabled efficiencies to clients.

If I tell you that my rate is £300 per hour, you know the price of an hour. You don’t know whether your matter will take five hours, 15 hours or 50 hours. A lawyer can provide an estimate, explain the assumptions behind it and keep the client informed if circumstances change. Those conversations improve transparency, but the hourly rate on its own doesn’t tell the client the likely overall cost. This is why I’ve always found the argument that the billable hour is transparent so strange.

The SRA’s guidance on transparency in price and service makes the case for the end of the billable hour even louder. For certain specified services, firms must publish price and service information. Where a firm publishes a range of costs, it must explain the basis of its charges, including relevant hourly rates and the factors that will determine the final price.

The guidance also points to the wider obligation to provide clients with the best possible information about how their matter will be priced and its likely overall cost. The SRA doesn’t tell firms which pricing model to use, and it doesn’t require every firm to publish a binding quote for every possible matter. What it does is place an obligation on law firms to give clients the best possible information so they can make an informed decision about how to proceed.

The problem with the hourly rate is that it has led to bad behaviours. Poor scoping and communication can mean clients base their decisions on estimates when the final bill is wildly different. A law firm may win the work, but it risks losing the client’s trust.

A clearly scoped fixed fee helps with those conversations. The client knows the agreed price for the work covered before it begins. The firm still has to explain what is included, what is excluded and what happens if the scope changes, but the starting point is more certain and transparent than an hourly rate alone.

Fixed fees and value-based pricing are not the same thing

A fixed fee tells the client what they will pay for an agreed scope of work. It gives them certainty, but it doesn’t necessarily reflect their particular circumstances or what the work is worth to them.

A firm might estimate that a matter will take 10 hours, multiply those hours by an hourly rate and offer the result as a fixed fee. That gives the client more certainty, but the price is still based on time.

A published conveyancing price menu is another example. A firm might offer a standard fixed price for a property within a particular price band, subject to any conditions it publishes. The firm may also list additional charges for particular issues or pieces of work. That menu gives clients useful information, but a standard price for a category of work is not automatically based on what the service is worth to each individual client.

Value-based pricing starts with a different conversation: what is the client trying to achieve? What is at stake? What concerns them? What would a successful outcome or a greater degree of certainty mean to them? The answers help the lawyer understand the client and the work, but there isn’t a formula that produces the right price at the end of that conversation. The lawyer has to use their judgement, taking into account the client’s circumstances, the work involved, the value to the client, and how much they want to take on the work. At some point, you have to decide whether the price feels right. That price can then be agreed upfront as a fixed fee.

This doesn’t mean that every matter can be scoped perfectly from the beginning, or that a firm should promise an outcome it cannot guarantee. Some matters may need to be priced in stages, with clear agreement about what happens when the work changes. The point is to have the conversation about the client and the work before deciding that the clock is the best way to price it.

I explore this in more detail in A Beginner’s Guide to Value Pricing for Lawyers.

Where should firms start?

Firms don’t need to change their business model overnight. They could start by choosing one recurring type of matter they know well and reviewing the information they already hold about it.

Look at past matters and invoices: what work was involved, what was charged, what was written off and what the client ultimately paid. That information can give the firm a baseline for considering how it might price similar work in future. It may also reveal how often the final bill differed from the original estimate and what caused those differences.

AI could help firms collate and analyse that information across far more matters than would be practical to review manually. It could identify patterns in the work, fees, write-offs and changes in scope for lawyers to investigate. The firm would still need to check the analysis and make sure client information is handled appropriately, but it doesn’t have to start with a blank page.

The next step is to talk to clients about what they need from that work, what they value and what level of certainty would help them. The historical data gives the firm a starting point; the client conversation helps it decide what to offer and what price to agree upfront.

Firms should also look at how they measure their lawyers. If they want lawyers to price work around the client and deliver it efficiently, but continue to judge them primarily on hours recorded, they are sending two different messages.

What do law firms want to price?

The Law Society’s guidance leaves firms with a choice about where to put their attention. They can continue investing in training lawyers to capture every minute they work, or they can invest in teaching them how to understand client value, define the work and agree a price before it begins.

Law firms have spent years teaching lawyers how to record more time. As AI changes how legal work is delivered, it’s time they put that effort into teaching lawyers how to price it.

That means learning how to understand what the client values, talk confidently about fees and agree a price before the work begins. These are skills lawyers can learn, but they won’t develop by attending another time-recording course.

If you’re ready to start thinking differently about how you price your work, Pricing for Value: An Introduction to Value Pricing will help you understand the principles of value pricing and how to start having those conversations with clients.