This is part 3 in a 3 part series.  Part 1 questions Goldman’s Sachs data showing that 44% of of legal tasks could be replaced by Generative AI.  In Part 2, we find some better data and estimate an upper limit of 23.5% of revenue that could be reduced by Generative AI. All of our assertions and assumptions will be discussed in further detail in a free LVN Webinar on August 15th.

The Big Idea:  We apply reductions in hours due to Generative AI to a few matters to determine Generative AI’s potential effect on profitability.
Key Points:
  • We establish a baseline sample matter and compare changes to that sample matter when Generative AI is applied
  • We explore how leverage is affected by Generative AI and how those changes may affect profitability in unexpected ways

Determining Generative AI’s effect on law firm profitability requires a bit more than a “back of the napkin” calculation with rough percentages based on keywords in time entries, as we did when roughly calculating the effect on revenue.

As Toby pointed out at the end of the last post, Generative AI is unlikely to hit all timekeepers equally.

We begin with this assertion.

Generative AI will disproportionately impact non-partner hours.

We are comfortable making this assertion for two reasons:

  1. Generative AI, in its current state, is most likely to replace or shorten the time to complete lower complexity and lesser specialized tasks that should be performed at the associate or paralegal level.
  2. Any time legal work hours are reduced, Partners tend to protect their own hours.

With that in mind, Toby began a profitability analysis, beginning with a baseline sample matter that does not factor in any use of Generative AI. We will use this baseline to compare against our AI adjusted matters.

Baseline M&A Sample Matter Data

Our baseline sample matter is loosely modeled on an M&A transaction and includes 5 timekeepers:

  • an Equity Partner
  • a 17th year service partner
  • 10th, 7th and 3rd year associates
TK Hours Rate Realization Revenue Expense Profit
EP 80 $1,000 88% $70,400 $15,200 $55,200
SP17 100 $895 88% $78,760 $48,500 $30,260
10yr 125 $735 88% $80,850 $48,750 $32,100
7yr 90 $660 88% $52,272 $32,400 $19,872
3yr 55 $595 88% $28,798 $18,150 $10,648

Estimated Annual Profit Per Equity Partner (PPEP) – $1,851 X 1400 hrs = $2,591,400

Leverage – 60% Non-Partner Hours

There are, of course, a number of assumptions in this baseline data that could greatly change from firm to firm, including the billable rates, the realization rate, and the expense for each timekeeper. However, we will keep this baseline data consistent across all of our examples in order to make a fair comparison. With different rates, realization, and expenses you will get different results. We strongly encourage every firm to perform a similar calculation for themselves.

Baseline Matter Analysis

The total hours billed are 450. The total revenue is $311k and the total profit in dollars is $163k.

Our model then translates the profit on this one matter into an estimated PPEP number for the firm. This is so we can determine profit margin impact separate from profit dollars.

In this baseline model, the PPEP number is ~$2.6m; meaning that if all work at this firm were staffed and billed like this one matter, the firm average PPEP would be about $2.6m.


There’s an old adage in economic circles: “Workers Work. Owners Benefit.”

For many people, especially in law firms, this can be somewhat counter-intuitive. Partner’s hourly rates are higher than non-partners, so it seems only logical that their billable hours must also generate greater profits for the firm. However, the reality is exactly the opposite. Partners charge a lot for their time, but they also cost the firm a lot. Non-partners bill less per hour, but their work still generates greater profits for the firm. In fact, this margin on non-partner timekeepers is where most law firm profitability comes from.

The percentage of non-partner hours to partner hours is called leverage. In our baseline example above, the leverage is 60%; meaning that roughly 60% of this matter’s hours are billed by a non-partner. This metric will become very important once we start adjusting our sample matter data away from the baseline.

Adding Generative AI to the Baseline

Conservative Gen AI Scenario – 5%/20% Reduction in hours

Our first alternative scenario is somewhat conservative, in that we believe that these reduction in hours will be substantially eclipsed in the long-term, but are quite possible, even likely in the next 12-24 months. This scenario assumes a reduction in partner billable hours by 5% and a reduction in non-partner billable hours by 20%.

These reduced hours result in the following changes to our baseline example:

  • Total hours reduced by 63 (-14%)
  • Total Revenue reduced by ~$40k (-13%)
  • Profit in Dollars down by ~$17k (-11%)
  • PPEP down by ~$173K to $2.4M (-6.68%)

Leverage (a.k.a. non-partner hours) in this scenario is reduced from 60% down to 56%.

Better Leverage Scenario

If we assume our baseline scenario remains at 450 total hours billed but these hours are better leveraged at 63%, then the estimated PPEP is increased +$107K to $2.7M or roughly +4% higher than baseline. A 3% increase in leverage results in 4% increase in profitability without any additional hours being billed to the client. This is a clear demonstration of the power of leverage.

Better leverage equals greater profit.

However, applying the same conservative effect of Generative AI that we applied to the baseline (-5% to partners /-20% to associates) to this better leverage model, results in an overall decrease in PPEP of -$193K or -7.46% to ~$2.5M.

In other words, for firms, matters, or practices that have better leverage already, Generative AI usage could reduce profitability more than it does for those with worse leverage. In our example, our conservative AI scenario reduces estimated PPEP in the better leveraged practice by 7.46% versus only 6.68% in the decently leveraged baseline practice.

Do not take from this that worse leverage is beneficial when using AI.

Worse Leverage Scenario

Our baseline model starts with leverage of 60%. When Generative AI is added in our first scenario above, it results in leverage of 56% and a PPEP of $2.4M. If our baseline instead started with a leverage of only 55%, adding Generative AI would result in an adjusted leverage of 51% and a PPEP of $2.07M. So, a 5% lower starting leverage results in 13.75% less profitability on the same work even when both practices are using Generative AI.

The AI-Pocalyptic Scenario

But what if our 5%/20% scenario is too conservative? Or what if we look a bit further out to a time when Generative AI is even more capable and replaces more complex lawyer tasks?

To test this scenario we ran a 20%/40% scenario, where 20% of partner hours and 40% of non-partner hours are displaced by Generative AI. Here’s how that scenario would affect our baseline matter:

  • Total hours reduced by 144 (-32%)
  • Total Revenue reduced by ~$95k (-30%)
  • Profit in Dollars down by ~$42k (-28%)
  • PPEP down by ~$274K to $2.3M (-10.57%)

Leverage (a.k.a. non-partner hours) in this scenario is reduced from 60% down to 53% and profit on this single matter is down 28%.

Bottom Lines

Now is the time to explore this technology

For the next 6 to 12 months, invest in education around Generative AI. Get access to AI tools as inexpensively (and safely) as possible, and deploy them for educational purposes and internal uses only. Emphasize that these tools are not (yet) to be used with client confidential data and that each lawyer is personally responsible for the verification and validity of all work product they produce whether created in collaboration with AI or not. Then give your lawyers and staff, especially your profitable associates, some basic training on how to prompt the AI and what to expect in return, and allow them time to experiment with the various capabilities of these tools.

Do not blindly invest in expensive Generative AI solutions

Before investing heavily in Generative AI solutions, you will need to understand how deploying these tools affects your bottom line. If our primary assertion is correct and Generative AI will disproportionally displace associate time, then your most profitable practices may be negatively impacted by the use of Generative AI.

You have to improve leverage

This is not new, but it does take on greater urgency as new technology begins to do lower level work. Toby has a saying, “I don’t care what the question is, the answer is better leverage.” This is his way of saying firms should have a strong focus on building and improving their leverage model. Gen AI makes that need even greater.

Partner compensation models need to change

Most partner compensation methods do not reward more profitable behavior. All of the compensation models we know reward Partner “production” (i.e., the number of hours a partner bills) above other metrics. In a world where lawyers can do more work faster, associates should be performing the vast majority of that work, while Partners as owners focus on building their business. Models that continue to incentivize moving work to higher cost resources (partners) will continue to raise legal costs for clients, even while diminishing firm profitability. That’s a lose/lose scenario.

Flat Fees and Alternative Pricing

It’s a dirty word in some legal circles, but now is the time to consider legal work that could be done at a flat rate, or using value based pricing, and put those models in place. Generative AI will expose the dirty little secret of a lot of practices, that they bill top dollar for a decent amount of associate busy work that can now be done by a computer at a fraction of the cost. Finding a way do this work quickly, for your clients at a lower (but still profitable) cost will increase client satisfaction, improve stickiness, and make it more likely the firm will win the higher value hourly work that the computer cannot (yet) perform.

Don’t get caught up in the hype

Generative AI is at the peak of inflated expectations on the Gartner Hype curve and we’re at a point in time when no one can really say for certain whether this technology will be a short lived, tech investor-driven, flash-in-the-pan phenomenon, or whether it will truly revolutionize everything that comes after. However, the smart money is on something in between.

Magic does not exist.

Faster is not always better.

More [or less] work does not equal greater profit.

Know what you’re buying and where you’re deploying it and how it will affect your businesses bottom line before pulling the trigger on that shiny new Generative AI tool that promises to do everything you want and more.

Please join us for a free LVN Webinar on August 15th in which we’ll discuss all of this in further detail, take questions from the audience, and viciously defend our most outrageous assertions and sloppily-reasoned assumptions.