Advanta is currently undergoing final system calibration ahead of launch. Selected infrastructure and experiences may still be in active refinement.

advanta

Framework / ROAI

Productivity is one quadrant of four.

ROAI is the four-quadrant return framework for AI investment in legal functions. Productivity is the smallest. Defensibility is typically the largest. Institutional and category positioning are the multi-year compounders. Single-quadrant cases underfund the disciplines the other three require.

Authored by

Nishant Bhaskar — Founder + Editor-in-chief, Advanta Legal Tech

Version 2026-Q3-v1 · Binding canon

Executive Summary

ROAI is the four-quadrant return framework for AI investment in legal functions. The traditional conversation about return, focused on time saved per task and cost per matter, captures one quadrant and misses three. Functions that justify AI spend on productivity alone underweight the investment relative to its institutional value, build governance posture poorly because the funding case did not require it, and fail to position competitively against peers evaluating AI on a fuller frame. The four quadrants are productivity value, Defensibility value, institutional value, and category positioning value. Productivity is the most easily measured and the most easily overstated. Defensibility value — the reduction in regulatory and incident exposure that AI governance provides — is typically the largest single quadrant by avoided cost. Institutional value, the function's standing inside the organisation, determines budget over a multi-year horizon. Category positioning value, the function's standing relative to peer functions, determines talent attraction and client outcomes. The four together constitute the case a legal function presents to the board. Single-quadrant cases lose to fuller-frame cases at every funding cycle.

Why Productivity Is Not Enough

The productivity-only case underweights the investment.

Most conversations about AI return in legal functions are productivity conversations. The function deploys a contract-review tool. The tool saves twelve hours per contract. The function multiplies hours by loaded rate by volume and presents the result to the board. The board approves, or does not, based on whether the number exceeds the licence cost.

The conversation is not wrong. The mathematics of productivity is real. The conversation is incomplete. Productivity savings are typically the smallest of the four return categories that legal AI investment actually produces. A function that builds its AI funding case on productivity alone underweights the investment by a factor of three or four. Worse, the function does not invest in the operational disciplines that the other three categories require, because the funding case did not name them.

Two years later the productivity savings have materialised, the regulatory exposure has compounded because no one funded the Defensibility framework, and the function is structurally behind peer functions that built the full frame from the start.

The Four Quadrants

What legal AI investment actually returns.

The four quadrants together constitute the case a legal function presents to the board. Single-quadrant cases lose to fuller-frame cases at every funding cycle. Percentage ranges below are illustrative — empirical anchoring comes from the Annual Legal AI OS Index from 2028 onward.

Quadrant Q1

~15–30% of total ROAI

Productivity value

Time saved per task, cost per matter, capacity freed for higher-value work.

The quadrant the existing return conversation already covers. AI tools that augment lawyer tasks reduce time per task; the saving compounds across the function's volume. Real and measurable, but the risks of measuring are well-documented: vendor demos overstate, pilot results do not scale, lawyers absorb saved time into existing work. A function that measures rigorously against pre-AI baselines with cohort-controlled methodology produces credible numbers. A function that accepts vendor claims at face value cannot. Typically the smallest of the four quadrants when measured honestly.

Quadrant Q2

~40–60% of total ROAI — typically the largest

Defensibility value

Reduction in regulatory exposure, incident cost, and disclosure preparation.

The avoided cost of governance: a regulator inquiry answered in twenty-four hours with a current Defensibility Posture Statement costs a fraction of the same inquiry assembled retrospectively over six weeks. An incident with decision traceability costs an order of magnitude less than one requiring full reconstruction. An EU AI Act conformity assessment completed in advance is cheaper than one completed under regulator deadline. Defensibility value is the quadrant the Defensibility framework operates against. A function that invests in AI without investing in Defensibility captures the productivity quadrant and forfeits this one.

Quadrant Q3

~15–25% of total ROAI

Institutional value

Standing inside the organisation as a strategic capability partner.

Determines budget over a multi-year horizon. A function that adopts AI thoughtfully — with governance, with measurement, with documented methodology — becomes the function the CFO calls when the organisation acquires a target with AI exposure, the function the CEO calls when a sectoral regulator raises AI guidance. These calls compound into budget, headcount, and influence. The quadrant requires the function to talk publicly about its AI posture through industry forums, advisory engagements, and documented governance.

Quadrant Q4

~10–20% of total ROAI — most steeply rewarded by early action

Category positioning value

Standing relative to peer functions for talent, clients, and influence.

Legal functions compete for talent, for client confidence in firm contexts, and for influence within their organisations and the profession. AI maturity has become a primary axis of that competition. The most variable quadrant: a function ahead of its peer cohort captures most of the available positioning value; a function behind captures very little regardless of absolute investment. The compounding nature of positioning rewards early movement and penalises late movement steeply.

How to Apply

Per investment. Across the portfolio.

The ROAI framework is applied per AI investment and per investment portfolio. For each candidate, the function evaluates expected return across all four quadrants, presents the four-quadrant case to the board, and tracks actual return against the projection across the investment lifecycle.

The portfolio application looks across investments. A function with three AI investments all justified primarily on productivity may have a portfolio that overweights Quadrant 1 and underweights the others. The portfolio view surfaces structural gaps: a function with no Defensibility-quadrant-focused investment in the last twenty-four months is exposed, regardless of how productive its other tools are.

Cluster Interactions

ROAI is not an isolated framework.

ROAI interacts directly with the other anchor essays in the Defensibility cluster. Each interaction shapes how the four quadrants are measured.

Defensibility

The operational system that produces Quadrant 2. A function with no Defensibility framework cannot measure Defensibility value because it has no instrument to detect avoided exposure.

Risk Taxonomy 2026

The framework against which Defensibility value is measured. The nine classes each have an avoided-cost profile that ROAI can quantify. Without the Taxonomy mapping, attribution falls back to generic averages.

Vendor Index methodology

Scores vendors against six dimensions that map directly to Defensibility-quadrant return. Vendors that score well produce Defensibility-quadrant return when deployed; vendors that score poorly produce Defensibility-quadrant cost.

Maturity Stack

Places the function on a five-band ladder where progression correlates strongly with institutional-quadrant return. A function moving from Operational to Integrated captures institutional value through visible posture maturation.

Editorial status

The canonical ROAI essay is in authorship.

The four quadrants above are anchored to canon 2026-Q3-v1 and are structurally complete. The long-form essay extends each quadrant with worked case examples; the Annual Legal AI OS Index will anchor the percentage ranges empirically from 2028 onward.

Subscribe — get the essay when it lands

From frame to funded

Build the four-quadrant case to the board.

Two paths. Run the diagnostic to surface where your AI portfolio sits across the four quadrants today. Or request the Executive Diagnostic for a board-ready ROAI assessment with attested evidence.