Calculating true ROI when hours worked and hours billed diverge

A midsize litigation associate uses an AI drafting tool to produce a first-pass summary judgment brief in 40 minutes instead of 4 hours. On an hourly engagement, that "efficiency win" can cost the firm money. On a flat-fee matter, it is pure margin. Same tool, same output, opposite financial outcome. If you evaluate AI return on investmentreturn on investmentReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → (ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition →) without asking "which fee structure sits on top of this workflow," you will get the sign wrong.

This lesson builds a simple model across the three dominant law firm fee structures so you can reason about AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → the way finance and practice-group leaders should: fee-structure-first, not tool-first.

Why fee structure changes the sign of ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition →

Law firms bill clients in roughly three ways:

• Hourly billing: the firm bills for time spent, typically in six-minute increments. Revenue is a direct function of hours worked.
• Flat fee: the firm agrees to a fixed price for a defined scope (a standard NDA, an uncontested trademark filing, a routine merger closing). Revenue is fixed regardless of hours worked.
• Contingency fee: the firm is paid a percentage of the outcome (common in personal injury, mass tort, some employment litigation) win or settle only.

AI drafting and review tools (contract analysis platforms like Harvey or Ironclad, deposition summarizers, litigation research assistants like Casetext CoCounsel, now part of Thomson Reuters) all do the same fundamental thing: they compress the hours needed to produce a work product. The financial effect of that compression depends entirely on what happens to the "freed" hours.

The hourly case: efficiency can shrink revenue

Under hourly billing, time is the product. If a partner bills $650/hour (a plausible 2026 estimate for a senior partner at a large US firm; actual rates vary widely by market and practice) and AI cuts brief-drafting time from 4 hours to 1 hour, the firm's *billable revenue* on that task drops from $2,600 to $650, unless the firm bills for the AI-assisted hour at a premium or repositions pricing.

Simple worked example:

• Task: first draft of a 20-page contract
• Pre-AI time: 5 hours at $500/hour = $2,500 billed
• Post-AI time: 1.5 hours at $500/hour = $750 billed
• Revenue difference: minus $1,750 per contract, all else equal

This is the classic hourly billing paradox: the more efficient the lawyer becomes, the less the firm earns per matter, unless volume rises to absorb the freed capacity. Clients, especially sophisticated corporate clients and their in-house counsel, are increasingly aware of this. Many now ask firms directly whether AI was used and expect a rate adjustment. The Thomson Reuters Institute's annual State of the Legal Market reports have tracked rising client pressure on this exact point.

For AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → to be positive under hourly billing, the firm needs one of:

1. Volume growth: freed hours get redeployed to more billable matters (capacity, not efficiency, drives ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition →).
2. Rate repricing: the firm shifts part of the task to flat fee or a blended rate that captures value delivered, not time spent.
3. Cost reduction: fewer associate hours means lower staffing costs relative to revenue, improving margin even if top-line revenue softens.

The flat-fee case: AI is a direct profit center

Under flat fee, revenue per matter is fixed. Every hour AI saves is pure cost reduction, and margin expands directly.

Worked example:

• Flat fee for a standard commercial lease review: $1,200
• Pre-AI cost: 3 associate hours at a $150/hour fully loaded internal cost = $450 cost, $750 margin
• Post-AI cost: 1 associate hour plus AI subscription cost allocated at $20/matter = $170 cost, $1,030 margin
• Margin improvement: +$280 per matter, roughly a 37% margin increase on that matter

This is why flat-fee-heavy practice areas (routine transactional work, immigration filings, standardized estate planning, high-volume contract review for corporate legal departments) are the earliest and clearest AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → winners. The tool directly converts time saved into margin, with no revenue offset.

The contingency case: AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → depends on case velocity and selection, not hours

Contingency fee firms (common in US personal injury and mass tort practice) get paid only on outcomes, as a percentage of settlement or verdict, often around 30 to 40% (a commonly cited industry range, treat as an estimate that varies by state and matter type). Hours worked are largely invisible to revenue.

Here AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → shows up in two places:

1. Case throughput: AI-assisted medical record review, demand letter drafting, or deposition summary tools let a firm handle more cases with the same staff, increasing total contingency fee revenue across the portfolio.
2. Case selection quality: AI-assisted analytics on case value and litigation risk (tools used for early case assessment) help firms decline weak cases sooner, reducing sunk cost on matters that will not pay out.

The ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → math is portfolio-level, not matter-level. A tool that saves 10 hours per case is worthless if it does not increase the number of cases resolved or improve which cases are taken. This is a critical evaluation point: vendors selling into contingency firms should be asked for throughput and selection metrics, not just "hours saved."

A simple framework for evaluating vendor ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → claims

When a vendor says "our tool saves 70% of drafting time" (a type of claim common in 2025 to 2026 legal AI marketing), translate it through fee structure before believing it changes your bottom line:

IF fee_structure == "hourly":
    roi = (freed_hours_redeployed_to_billable_work * rate) - subscription_cost
    # if freed_hours are NOT redeployed, roi can be negative

IF fee_structure == "flat_fee":
    roi = (hours_saved * internal_hourly_cost) - subscription_cost
    # roi is close to guaranteed positive if adoption is real

IF fee_structure == "contingency":
    roi = (incremental_cases_resolved * average_fee) - subscription_cost
    # roi depends on throughput and case-selection improvement, not hours

This is not code you would run in production, it is a reasoning checklist for partnership meetings and vendor demos.

What this means for adoption strategy

Firms with mixed fee structures (most large firms have all three, in different practice groups) should not run one AI adoption strategy. Practical implications:

• Flat-fee and high-volume transactional groups: push AI adoption hard and fast. Margin gains are close to mechanical, and the main execution risk is quality control (a hallucinated clause in a flat-fee contract review is an error cost, not a billing problem).
• Hourly litigation and advisory groups: pair AI rollout with a pricing conversation. Either convert discrete deliverables (first-draft memos, initial contract review, deposition digests) to flat or capped fees, or be explicit internally that ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → comes from capacity reallocation, not per-matter revenue.
• Contingency practices: measure ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → at the portfolio level over quarters, not per case. Track case throughput and settlement cycle time, not hours saved.

Bar associations and regulators are also watching billing transparency here. The American Bar Association's Model Rule 1.5 on reasonable fees, and several state bar ethics opinions issued since 2023, address whether firms must disclose AI use or adjust bills when AI reduces actual time spent. This is an evolving compliance area, not settled law, and firms should treat client billing transparency as part of the ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → calculation, not an afterthought.

🎬 [VIDEO: "How AI Is Changing the Billable Hour" — youtube.com — search for recent legal-industry panel discussions (Thomson Reuters Institute, Clio, or ILTA conference sessions) analyzing AI's effect on law firm billing models]

Key Takeaways

• AI drafting and review tools compress hours, but the financial effect of that compression flips sign depending on fee structure: hourly billing can turn efficiency into revenue loss, flat fee turns it into direct margin, contingency turns it into a portfolio-level throughput question.
• Under hourly billing, positive ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → requires redeploying freed capacity to new billable work or repricing the deliverable, not just adopting the tool.
• Under flat fee, AI ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → is close to mechanical: hours saved times internal cost, minus subscription cost, usually nets positive quickly.
• Under contingency, evaluate AI on case throughput and case-selection quality across the portfolio, never on hours saved per matter.
• Always ask which fee structure sits beneath a given practice group before believing a vendor's "hours saved" ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → pitch, and check emerging bar guidance on billing transparency when AI reduces time actually spent.