Is This AI Strategy Real?

The easiest mistake in AI strategy is confusing visible activity with strategic progress. Companies launch internal chatbots. Functions run pilots. Vendors present agent demos. Board decks show dozens of use cases. Everyone agrees the firm is moving. Sometimes it is. Often the activity is motion without operating change.

The distinction matters because fake AI strategy is expensive in a way that ordinary software theatre is not. It wastes budget, and it also teaches executives the wrong lesson about the technology. Local enthusiasm appears without operating change; disappointment arrives later without diagnosis. A company can say it is experimenting while leaving the hard work untouched — the actual workflow redesign, the operating owner, the accepted risks, and the stop rule when evidence stays weak.

Recent market evidence makes the problem hard to ignore. McKinsey's 2025 State of AI survey reported broad regular AI use, while many organizations were still in experimentation or pilot stages and only 39 percent reported any enterprise-level earnings impact. A 2025 CEO study from IBM reported that only 25 percent of AI initiatives had delivered expected return on investment and only 16 percent had scaled enterprise-wide. Boston Consulting Group (BCG) in 2026 found that only 5 percent of surveyed companies were generating measurable value from AI at scale, while 60 percent were not achieving material value. Deloitte's enterprise generative-AI survey found enthusiasm still running ahead of full scaling, with many experiments unlikely to scale in the near term. These numbers are not an anti-AI brief. They are an anti-theatre brief. The technology is spreading faster than the management discipline around it.

The board should not ask whether the company has AI initiatives. Ask which operating system is being changed, and what proof will survive contact with the P&L.

The Five Tests

An AI initiative is more likely to be strategy than theatre when it can pass five tests.

1. The workflow test

The initiative names the recurring business workflow it will change — not a capability, not a model, not a department's enthusiasm. A workflow. "Use AI in sales" is theatre language. "Reduce the time between qualified inbound lead and first tailored proposal from three days to six hours without lowering win-rate quality" is strategy language. The second sentence forces the team to name inputs, handoffs, decision rights, data access, approval thresholds, and measurement.

Workflow redesign keeps showing up in the better AI-value research. McKinsey's 2025 survey on organizations rewiring for value found workflow redesign to be the attribute with the biggest effect on reported earnings impact from generative AI. BCG's AI-at-work research similarly argued that value requires redesigning work rather than merely distributing tools.

2. The accountable owner test

The initiative has a named business owner who controls the process being changed. A center of excellence can advise. IT can enable. Legal and security can set control points. If nobody accountable for the business process owns the result, the project is structurally weak. These projects often have sponsors but no owner. A sponsor likes the idea. An owner can change staffing, metrics, handoffs, incentives, and stop conditions. Without that authority, the AI project becomes a demonstration of what might be possible in a parallel world where the organization itself does not have to move.

3. The evidence test

The initiative defines the evidence that would prove or disprove the thesis before the pilot starts. Adoption is not enough. Seat activation is not enough. Prompt volume is not enough. Time saved is useful but incomplete if quality, rework, customer trust, risk, and revenue are ignored. A serious evidence plan includes a baseline, a target, a measurement window, a quality check, a cost denominator, and a decision rule. The cost denominator matters because AI systems now bring governance, data, security, review, observability, and change-management costs. A tool that saves time but creates new review burden may still be worth it — the burden has to appear in the calculation.

4. The management-layer test

The initiative can be managed after launch. That means named permissions, logs, review cadence, exception routing, rollback rules, and a way to discover whether the agent or assistant is still operating inside its intended scope. This test separates attractive demos from durable operations. In a demo, the workflow is narrow, the data is curated, and the failure is private. In production, the system meets edge cases, stale records, user shortcuts, security limits, and ambiguous accountability. The management layer is where the cost of real adoption shows up.

AI Needs a Management Layer argued that agent registries, observability logs, ownership controls, provenance rules, and transparency obligations are becoming an operating layer. The same point applies here: if the initiative cannot be governed as normal work, it is not yet strategy.

5. The opportunity-cost test

The initiative explains why this workflow deserves scarce attention now. AI strategy is as much about rejecting tempting use cases as it is about finding good ones. The opportunity cost may be executive time, change fatigue, security review, data cleanup, vendor dependence, or integration work. It may also be the right to ask employees to alter a process that already functions. A mediocre AI project can crowd out a boring operational fix that would produce more value. The strongest AI roadmaps are short. They concentrate on workflows where the upside is large enough to justify the organizational movement.

Pass or fail

Executives can use a simple screen before approving the next AI initiative.

The C-Suite Blind Spot

The most dangerous theatre is not the obviously silly demo. It is the plausible project that flatters every function. Chief executives see transformation. Finance sees potential savings. IT sees modernization. HR sees employee enablement. Marketing sees content velocity. Vendors see expansion. Boards see responsiveness to the market. Nobody wants to be the person who slows it down. That political dynamic is why the pass-fail screen has to be explicit. Without a screen, weak AI initiatives survive because they are individually defensible. With a screen, the question changes from "is this interesting?" to "does this pass the conditions for managed operating change?"

What To Do This Week

Skip the grand AI strategy refresh. Take the ten most visible AI initiatives in the company and score each one against the five tests. Use three labels only: pass, revise, or stop.

Pass means the initiative has a named workflow, accountable owner, evidence plan, management layer, and opportunity-cost case. Fund it properly and protect it from distraction.

Revise means the initiative is plausible but missing one or two conditions. Rewrite the charter before spending more money.

Stop means the initiative is mostly signalling. Kill it politely, record the reason, and put the saved attention on a workflow with a better claim.

The reality test is intentionally unsentimental. Belief in AI is irrelevant. What matters is whether the company will change how work is done, measured, owned, and governed. That is where strategy starts. Everything else is lighting.