Rethinking legal services value in an AI-enabled world

As he considers the evolution of billing models and how law firms continue to define value, Mark Andrews takes the view that a time-based view of value is likely to be a major impediment to the adoption of AI.

In short:

• In a work world that is enabled by AI, the way we think about value is critical to how we shape work into the future.
• Law firms should increasingly think about AI’s impact on value from a client perspective, rather than in terms of its impact on the billable hour.
• Firms need to identify the levers and motivators that will drive behavioural change in the age of AI.

A brief history of billing

Earlier this year, I asked a conference audience of about 70 people, primarily from law firms, to put up their hands if they did not record time. Five people indicated that they or their firms did not record time.

While it is only a small sample size (and I suspect the true number may have been lower) it nonetheless revealed that time recording was still being used by the overwhelming majority of individuals or firms represented at the session.

In days past, a scrivener, or scribe, was someone who made a living from their ability to write and copy material. In an age when literacy levels were low, the scrivener played a critical role. That role exists in various forms today and is very much a volume-based role in that an individual makes money based on the volume of work they do, whether it be related to word count or document numbers. There is no direct link between scriveners and lawyers, of course, but the idea of fees being based on volume can be traced back to the scrivener role.

Jumping forward in history, time-based billing (i.e. volume based) was not the initial preferred model for billing. The use of fee schedules based on task or document type, annual retainers and contingency fees were all billing models that date back far earlier than time-based billing. The 1960s and 1970s were the decades in which time-based billing really started to take hold. In a February 2025 article, published posthumously (How law firms ended up with the billable hour model - Thomson Reuters Institute), James Jones, of Georgetown University Law Center, documented the emergence of the billable hour in the United States. Interestingly, an early example of its use was for the Boston Legal Aid Society, where the introduction of time recording improved case throughput by 65% and reduced the average case cost.

If we review literature of the past 20 years, the end of the billable hour has been widely predicted, and yet it still dominates. We do see many alternatives to time-based billing, including fixed fees, risk sharing, contingency, success fees and a range of other pricing and billing options, but they are not the dominant models.

Internally, the drive for the billable hour remains strong in most law firms, and many of the incentive and reward systems use it as a key input. In this way, regardless of the billing model in place, the billable hour has value to individual lawyers and there is clear motivation to maximise the number of billable hours.

The promise of AI

The main claim on the AI tin is that AI will increase productivity and innovation. To go a little deeper, it will perform tasks at greater speed and scale, and with a greater degree of accuracy than a human. AI is designed to make us more efficient, to relieve us of tasks better performed by machines, and to release time back to us so we can focus on other things (whether that be more work, or more time, to do other things).

What if we think about the promise of AI a different way, though. If we based our definition of value around billable hours, then AI is a value killer. Of course, there is more to the story and I will explore it further in the rest of this article, but calling out this tension is important because it plays to one of the less successful approaches to driving AI adoption.

Defining value

At core, the value of lawyers is through providing representation, facilitating access to justice, advocating and providing ethical frameworks. At the individual lawyer level, value might be from intellectual challenge, representing the unrepresented, simplifying complexity, facilitating a deal, overcoming major hurdles, prestige, reputation, or perhaps money. From a client perspective, we could simplify value to being commercially pragmatic actionable advice and outcomes, at a price the client is prepared to pay.

While value, like beauty, is in the eyes of the beholder, when we think about AI we need to think about value from a client perspective, rather than in terms of the impact on the billable hour. To expand on a point I touched on before, if we define value around billable hours and try to push AI adoption from a pure efficiency perspective – i.e. look how much time you can save – the outcome will be poor as we are asking lawyers to invest time to change their work practices in a way that will result in them having fewer billable hours – certainly not a win-win approach.

Rethinking value

The world of lean management provides what I think is a very useful set of tools and mindset around value, specifically value-stream analysis and value-stream mapping. Value-stream analysis is a discipline focussed on identifying inefficiency and eliminating waste and cost. A value-stream map is the primary tool supporting value-stream analysis.

A full value-stream map will consider, among other things, cycle time, lead time, uptime, changeover time, percent complete and accurate, availability, resource requirements and waste, including defects, overproduction, waiting, talent, transportation, inventory, motion and extra processing. In the law firm context, these waste types can be thought of as follows:

• defects – time spent fixing errors (e.g. the number of redrafts where the redraft is necessary due to basic quality issues, or poor understanding
• overproduction – spending too long on work, resulting in write-downs and write-offs
• waiting – delays caused by a lack of responsiveness from colleagues or clients, or diaries being blocked out with other matters, resulting in waiting time to be able to see colleagues
• talent – insufficient training, excessive utilisation, or chronic underutilisation
• transportation – too many information hops, excessive email traffic, or not using the most efficient means of communication
• inventory – bench size; the degree to which you are carrying capacity just in case it is needed
• motion – workspace organisation (e.g. insufficient quiet spaces, so people need to spend time searching for space), stationery and supplies not being available quickly; not having appropriate equipment
• extra processing – any activity that does not directly contribute to the end product. This tends to be something of a catch-all, but it is important to consider.

Fundamental to value-stream analysis is identifying value-added time and non-value-added time.

An oversimplified example of the process for consuming a sandwich is shown below. This considers value from the customer perspective, as for the customer there is value in the 20 seconds of selecting, the five seconds of ordering (although that is arguable) and the five minutes of eating. Our total value-added time is 5min 25sec. The non-value-added time is the minute queueing and the three minutes waiting for the sandwich to be made. With the non-value-added time, we could then analyse waste and see what we could do to improve. For example, kiosk ordering systems might reduce the queue time, while changing our production process could improve the making time.

Clearly, the value-stream maps in our context cannot be compared to this simple example, but the concepts and techniques are very relevant. When we think about AI and value-stream mapping, the message is clear – minimise time on non-value-adding aspects of the process and look to ways to maximise the time on value-adding aspects. The value-stream map gives us a very different lens to simply thinking that AI is great because it will make you more efficient.

As we rethink value, we need to tune in to write-downs (essentially wasted work that cannot be recovered) and write-offs (wasted work due to not being able to bill). This more disciplined approach is key to obtaining greater adoption of, and benefit from, AI. To over simplify, I would suggest that AI intensity is likely to increase with greater value-based billing and to reduce the more time-based billing.

Behaviour

Daljit Singh, in a March 2025 article for this publication, talked about designing work in the age of AI and called out challenges that AI adoption poses. As Daljit states, avoiding a technocratic approach to AI adoption is certainly one element of the solution. Another element is whether we define value as the number of billable hours, or approach value more from what differentiates us and results in clients receiving more pragmatic commercial solutions.

The client lens is not going to drive AI adoption for every individual in your firm. It is also important to consider a range of other potential motivators and requirements such as:

• ethical and professional – duty to clients, duty to the courts, professional rules, confidentiality, conflicts of interest
• adversarial approaches – a 'win at all costs’ mentality, zero-sum game and the pressure to succeed
• other factors – workplace culture, personal values, uncertainty, mental health, intellectual stimulation, helping clients.

The promise of AI may be to increase productivity and innovation, but if we are to properly drive adoption, we need to rethink value and identify the levers and motivators that will drive behavioural change.

Mark Andrews is Director – Global IT Service Delivery at Baker McKenzie. He has a varied background, including time in the public and private sectors, along with considerable professional services experience. He has held roles ranging from HR to management consulting and has previously been a guest lecturer in the business faculty of the University of Technology, Sydney - teaching at both Bachelor and Masters (MBA) level.