AI activity is no longer the scarce resource. Focus, evidence and operating change are.

Boston Consulting Group's July 2026 CEO research makes the gap visible. In the survey, 64% of CEOs said their companies pursued AI pilots, while only 26% said AI was embedded as part of broader business transformation. Higher performers were roughly seven times more likely to redesign workflows and reshape the business end to end with AI.

The useful question is therefore not how to run more pilots. It is how to turn a small number of well-chosen opportunities into measurable, owned business change.

The deeper diagnosis is covered in why AI pilots fail to create P&L value. The practical issue here is what leadership does next.

Four moves behind scaled AI value

BCG identifies four transformation disciplines that distinguish higher performers:

  1. The CEO sets direction and orchestrates the portfolio; business and P&L leaders own outcomes.
  2. The company focuses AI on a few high-value areas that can change the business.
  3. Initiatives are designed so the value path can be tracked from the start.
  4. People, roles and change management receive the same attention as technology.

These moves reinforce one another. Focus without measurement creates attractive priorities without an investment case. Measurement without real operating evidence leaves the decisive assumptions untested. A proven capability without workflow, ownership and role change remains a local improvement rather than a changed business.

For leadership teams, the sequence can be translated into four practical decisions.

1. Choose where AI can materially change performance

Start with the outcomes leadership already owns: growth, margin, working capital, service, quality, capacity or risk. Then inspect the workflows and decisions that determine those outcomes.

The aim is not to populate a use-case catalogue. It is to identify a small number of value domains where a different combination of human judgement, AI capability, data and workflow could change performance enough to matter.

AI Strategy should leave leadership able to state what it will back, why those choices matter and what it will stop or defer. For the detailed output, see what an AI strategy consulting engagement should produce.

2. Define the value path before committing investment

A named opportunity is still a hypothesis. Before funding a pilot or programme, define:

  • the current economic and operating baseline;
  • the workflow behaviour expected to change;
  • the operational measures that will show that change;
  • the financial or strategic outcome those measures can influence;
  • the owner, costs, constraints and assumptions that could invalidate the case.

An AI Opportunity Assessment makes that value path explicit and recommends whether to back, sharpen, defer or stop the opportunity. The mechanics belong in a live value model; how to measure AI P&L impact explains the causal chain in more depth.

3. Prove the critical assumptions in real work

A technical demonstration can establish that a model performs a task. The investment decision usually depends on more: real users, representative work, exceptions, controls, adoption, operating cost and whether the workflow actually changes.

AI Proof of Value builds the smallest credible capability that can test those conditions. It measures enough reality to support a scale, revise or stop decision. The goal is evidence, not a polished miniature of the final system.

This is why a Minimum Viable Capability differs from an AI proof of concept. It brings the technology, human judgement, workflow and measures into the same test.

4. Redesign the business around what proves itself

Positive evidence changes the job. Scaling is not simply making the tool available to more people.

The end-to-end workflow may need fewer steps, different handoffs or new exception routes. Decision rights and outcome ownership need to be explicit. Roles, measures, incentives, controls and funding must reinforce the new way of working. Old activity has to be removed, not left running beneath the new capability.

AI Transformation carries that redesign around what has earned further investment. The deeper operating questions are covered in the AI operating model: ownership, measures and decision rights.

A leadership test before the next funding decision

Before increasing investment, leadership should be able to answer:

  1. Which material business outcome will this change?
  2. Why is this opportunity more valuable than the alternatives?
  3. What is the baseline and value path?
  4. Which assumptions could still overturn the case?
  5. What evidence has come from representative work?
  6. Who owns the outcome, not merely the delivery?
  7. What workflow, role, measure or control will change if the evidence is positive?
  8. What result would cause us to scale, revise or stop?

If those answers are weak, more activity will not create confidence. The next move is to improve the decision, not enlarge the programme.

Source

Boston Consulting Group, CEOs Are Starting to See Value from AI. Now Comes Execution., 22 July 2026.