The Legal Stack
Independent LegalTech Analysis
← Research Briefings
Research BriefingNo. 071 · June 24, 2026 · 10 min read
Legal Economics · Research Report

The Legal AI Retention and Billing Realization Report 2026: How AI-Assisted Matter Delivery Is Actually Affecting Client Retention, Realization Rates, and Origination Credit Disputes at AmLaw 200 Firms

Clients expect AI savings while hourly billing still dominates. How ABA Opinion 512 limits capturing efficiency, and what Am Law 200 firms can measure.

Research briefing — a review of public sources. The Legal Stack did not conduct a survey, benchmark or interviews for this piece.

Executive summary

This report's title asks how AI-assisted delivery is affecting realization rates, client retention and origination-credit disputes at Am Law 200 firms. On all three, the public evidence is thin. Firms do not publish realization rates by practice area or by AI use. Retention data is private. Partner compensation disputes stay inside the partnership. We found no published study that ties any of these metrics to AI deployment, and earlier versions of this briefing should not have implied otherwise.

What can be documented is the surrounding economics. Hourly billing still dominates. Clients expect savings and mostly have not seen them. Firms are raising rates and recording strong profits. And ABA Formal Opinion 512 constrains how firms may capture AI efficiency. This briefing sets out that record, then offers labelled analysis of how the pressures are likely to show up in realization and compensation.

The market backdrop: hourly billing, rising rates, record profits

The State of the US Legal Market 2026 report from the Thomson Reuters Institute and Georgetown Law's Center on Ethics and the Legal Profession, as covered by Legal IT Insider in January 2026, found:

  • Average law firm profit grew 13% in 2025, and demand had its best year of growth since the global financial crisis.
  • Technology spend grew nearly 10%, and talent costs rose 8.2%.
  • About 90% of legal dollars still flow through standard hourly arrangements, according to Thomson Reuters Legal Tracker data.
  • Most clients do not know whether or how their outside firms use generative AI. Corporate legal departments want firms to propose AI-informed billing, while firms say clients still judge them on hourly rates.

The Thomson Reuters Institute's November 2025 analysis of its Law Firm Rates Report 2026 gives the most specific public data on realization. Some firms make annual standard-rate increases of 10% to 15%, while others make increases of 3% to 5%. Firms with what TRI calls "high compression" keep worked realization as high as 95%. "Low compression" firms can see pre-work realization fall to 75% to 80%. Collected revenue per hour averages $553 to $580 across the configurations TRI describes. TRI does not attribute any of these patterns to AI, and neither do we.

Analysis. The realization story in 2025–26 is mainly a rate story. Firms are pushing standard rates up quickly, and realization reflects how much of that clients accept. AI is a plausible future pressure on realization: if a matter takes fewer hours but invoices still look like before, clients will push back. But the public data does not yet isolate that effect.

What clients say they are seeing

The best client-side evidence comes from the Association of Corporate Counsel and Everlaw's 2025 GenAI Strategic Value for Corporate Law Departments survey of 657 in-house professionals in 30 countries, released 14 October 2025:

  • 59% reported no generative AI savings from their outside firms yet.
  • Only 24% were satisfied with how outside counsel are adopting generative AI for cost-effectiveness.
  • 61% plan to push for changes in how legal services are delivered and priced.
  • 64% expect to rely less on outside counsel, and 50% expect lower outside counsel costs.

Bloomberg Law's coverage of the survey added two details. Among those who had seen benefits, 13% reported fewer billable hours for tasks like document review and drafting. And more than 60% said it was too early in the adoption cycle for cost reductions to appear.

Separately, the Thomson Reuters 2025 Generative AI in Professional Services report found that 71% of corporate legal clients did not know whether their outside firms used generative AI, according to LawSites.

Analysis. This is the nearest public evidence to the "retention" question. Clients have not yet seen AI savings, most do not know whether their firms use AI, and a majority intend to push on pricing and bring work in-house. That points to retention risk being driven less by AI use itself than by whether firms can show clients where the efficiency went. It is an inference, not a measured outcome.

How much efficiency is at stake

Estimates of AI time savings are forward-looking and self-reported. The Thomson Reuters Future of Professionals report (June 2025, 2,275 respondents across professions) found that respondents expect AI to save them 5 hours per week within the next year, up from 4 hours predicted in 2024. Only 22% said their organisation had a visible, defined AI strategy. Organisations with one were twice as likely to report AI-driven revenue growth as those adopting AI informally.

Clio's 2024 Legal Trends Report, covering a broader and smaller-firm-weighted market, estimated that up to 74% of hourly billable tasks are potentially exposed to automation. It also found that 71% of clients prefer paying a flat fee for an entire case, according to LawSites. Clio's client data reflects consumer and small-business clients, not Am Law 200 corporate clients.

The ethics constraint on capturing efficiency

Firms do not have a free hand to keep AI efficiency as undisclosed margin. ABA Formal Opinion 512 (29 July 2024), as summarised by the ABA's Law Practice Today:

  • Lawyers billing hourly "must charge only for actual time spent," even when AI makes them faster.
  • A flat or contingent fee may be unreasonable if the lawyer can finish the work much more quickly with an AI tool.
  • Lawyers generally cannot bill for time spent learning an AI tool, unless the client asked for that tool.
  • Lawyers may charge direct costs of AI use but should not pass on general overhead, such as a "technology surcharge," without prior disclosure and agreement.

The opinion also requires disclosure of AI practices when a client asks, or when the engagement agreement or the client's outside counsel guidelines require it. The Law Practice Today author argues for a more flexible reading of Model Rule 1.5 that could allow a disclosed "tech fee." That is her position, not the ABA's.

Analysis. Under hourly billing, the rules push AI efficiency toward the client: fewer hours, lower bills. Firms that want to keep some of that efficiency as margin have to change the fee structure, usually to fixed or value-based fees agreed in advance, rather than quietly keeping hours constant. The billing-system and pricing questions are therefore linked, and "absorb first, pass through selectively" is a riskier posture than earlier commentary suggested.

Origination credit: an honest gap

We found no public data on origination-credit disputes connected to AI. Compensation systems are confidential, and disputes rarely become public unless they reach litigation or arbitration. The following is analysis, not reported fact.

For illustration, consider a firm that moves a due-diligence practice to fixed fees supported by AI review. Hours recorded by junior timekeepers fall. If origination and "working attorney" credit are computed from hours or fees worked, the partners who designed the workflow and the partners who own the client relationship may each feel under-credited. Firms with formula-heavy compensation systems are the most exposed to this tension. Firms that already credit pricing, innovation or matter profitability are less so.

Measurement gaps firms can close themselves

  • Matter-level AI tagging. Without a record of which matters and tasks used AI, firms cannot attribute realization changes to AI rather than rates, staffing or client leverage.
  • Pre/post baselines by task type. Hours per task category before and after deployment are the minimum evidence for pricing conversations.
  • Client-perception data. Given the ACC/Everlaw and Thomson Reuters findings, ask key clients directly whether they know how AI is used on their matters and whether they have seen value from it.

Recommendations (analysis)

  1. For firm leaders: decide deliberately, matter type by matter type, whether AI efficiency will be passed through as fewer hours or priced into fixed or value-based fees agreed in advance. Avoid the undisclosed middle ground that Opinion 512 makes hazardous.
  2. For pricing teams: build AI tagging into time entry before trying to explain realization to clients or partners.
  3. For compensation committees: review whether credit formulas penalise the partners who reduce hours through AI-enabled workflows.
  4. For GCs: ask firms how AI is used on your matters. Opinion 512 obliges them to answer. Put pricing expectations for AI-compressible work into outside counsel guidelines and RFPs.

Sources

Filed under Legal Economics → · The Legal Stack accepts no vendor funding for its research.

More Research

View all →
No. 113
10 min
What Published USPTO Guidance, Federal Circuit Opinions, and PTAB Decisions Through October 2026 Actually Say About AI-Assisted Patent Prosecution, Inventorship Disclosure Obligations, and the Duty of Candor When AI Tools Generate Claim Language
10 min
No. 112
10 min
What Published NLRB Guidance, ALJ Decisions, and Board Orders Through October 2026 Actually Say About the Use of AI-Assisted Monitoring, Discipline, and Work Allocation Tools in Unionized and Union-Eligible Workplaces
10 min
No. 111
10 min
What Published Delaware Court of Chancery Opinions, Docket Orders, and Judicial Commentary Through October 2026 Actually Say About AI-Assisted Legal Filings, Discovery Productions, and Counsel Disclosure Obligations in Corporate Litigation
10 min
© 2026 The Legal Stack — Independent LegalTech Analysis