Posted on

The GenAI Divide: Why 95% Get Zero ROI (and What the 5% Do Differently)

GenAI divide
AI + Operations Series
Based on MIT NANDA Research (2025)

The GenAI Divide: Why 95% of Organizations Get Zero ROI

Enterprise spend on GenAI is huge, adoption is high, and yet most organizations report no measurable P&L impact. The “GenAI Divide” explains why only a small group extracts real value—and how to join them.

Quick takeaway

The gap isn’t model quality. It’s approach: workflow fit, learning capability (memory + adaptation), and operational integration determine outcomes.

What the research found

The report calls it the GenAI Divide: many organizations adopt general-purpose tools, but very few turn AI into measurable business performance.
A key claim is stark—most organizations see no measurable return, while a small minority extracts meaningful value at scale.

High
Adoption

Teams try ChatGPT/Copilot quickly because the interface is familiar and flexible.

Low
Transformation

Most efforts stop at productivity improvements—not P&L impact.

Rare
Scale

Custom tools often stall due to brittle workflows and poor fit in day-to-day operations.

The real reason most GenAI initiatives stall

According to the report, the core barrier is not infrastructure, regulation, or talent. It’s learning:
many GenAI systems don’t retain feedback, don’t adapt to context, and don’t improve over time. In real operations,
that creates friction instead of reliability.

What the 5% do differently

  • They start with a specific process (not a generic “AI program”).
  • They measure outcomes (cycle time, error rate, external spend reduction).
  • They demand workflow fit (integration with existing systems and real user behavior).
  • They choose tools that learn (memory + feedback loops).

A simple “Monday morning” playbook

  1. Pick one repeatable workflow that touches revenue, risk, or delivery (approvals, ticket routing, document handling, forecasting prep).
  2. Map it: trigger → inputs → decision → handoffs → outcome.
  3. Remove ambiguity: define required inputs and rules (and what counts as an exception).
  4. Deploy AI where it removes friction (summaries, routing, extraction, drafting, classification).
  5. Track 2–3 metrics weekly and iterate for 30 days.

Want to cross the GenAI Divide in your organization?

We’ll identify a high-ROI workflow, build the measurement plan, and choose the right implementation path.

Talk to WSI