home  /  library  /  harvard-clp-business-models
Analysis & commentary

The Impact of Artificial Intelligence on Law Firms' Business Models

Harvard Law School Center on the Legal Profession · 2025 · HLS Center on the Legal Profession, Knowledge Hub

Billing & feesThe futureAdoptionAnyone who has wondered why partners are quietly resistant
Why it is on the shelf. The clearest available statement of why adoption stalls in firms that have bought the tools and run the training — the obstacle is compensation arithmetic, not attitude.

The Institute's reading

This is the piece to hand to anyone who thinks the adoption problem is cultural. Hourly billing accounts for the large majority of law firm revenue — commonly put at around 80% of fee arrangements — and it has one property AI attacks directly: the fee tracks time rather than result. A firm that adopts a tool cutting research or drafting time substantially, and continues to bill by the hour, has spent money to reduce its own revenue on that matter.

Every partner works this out immediately, and the resulting behaviour gets misdiagnosed with great consistency. What is read as scepticism about the technology, or as generational resistance, is very often a correct calculation about personal compensation. No amount of training addresses it, because training is not the binding constraint.

The analysis also makes clear that the pressure is uneven, which is the actionable part. It falls hardest on high-volume, repeatable, document-heavy work and barely reaches genuinely bespoke matters where time was never really the measure of value. That unevenness is what makes a mixed pricing model the realistic response rather than a wholesale abandonment of the hour.

Key propositions

  • Hourly arrangements represent roughly 80% of law firm fees and will not unwind quickly.
  • Hourly billing creates a direct conflict between adopting efficiency and being paid for the work.
  • Adoption resistance is frequently an incentive problem misread as a technology objection.
  • Pricing pressure is concentrated in high-volume repeatable work and light on bespoke matters.

In practice

  • Compensation metrics have to change before adoption will, and changing them is harder than running training.
  • Segment the book by compressibility and price the segments differently — a mixed model, not a switch.
  • Fixed-fee work is the natural place to start, because the incentive there already points the right way.

Where authorities disagree

A well-supported counter-view holds that the billable hour has survived every predicted death for forty years, and that clients continue to prefer it for its predictability of process and the comparability it allows across a panel. There is also a serious argument that alternative fee arrangements transfer risk to the firm in ways that are underpriced. What is not much disputed is the direction of pressure on specific work types — the disagreement is about how far it travels.

AI adoption conciergeorientation · not legal or ethics advice
Happy to dig into it. What would you like to pressure-test from The Impact of Artificial Intelligence on Law Firms' Business Models: one of its propositions, how it applies to your situation, or where it disagrees with the rest of the shelf?