# AI Strategy for Leaders 2026: Be in the 6% | Dexity

> 88% use AI but only 6% are high performers. A data-backed, sequenced AI strategy field guide for executives who want to be in that group.

AI at Work

# AI Strategy for Leaders in 2026: Why Most AI Fails, and How to Be in the 6% That Wins

88% of organizations now use AI, yet only about 6% qualify as "AI high performers" (McKinsey), 95% of GenAI pilots deliver no measurable P&L impact (MIT), and just 5% of companies are "future-built" (BCG). This is a data-anchored, sequenced strategy field guide for executives who want to be in the small group that actually captures value in 2026.

17 August 2026

10 min read

Author

[Shashi Bhushan Kumar](https://www.linkedin.com/in/shashi-bhushan-coder/)

Director - Product Management, Dexity

![An executive presents a seven step AI roadmap and a 6 percent high performer figure to leaders around a boardroom table](/assets/intel-ai-strategy-for-leaders-hero-55daf1.webp)

On this page

1.  [Why do most AI investments fail to deliver measurable ROI in 2026?](#why-do-most-ai-investments-fail-to-deliver-measurable-roi-in-2026)
2.  [The data: three research houses, one paradox](#the-data-three-research-houses-one-paradox)
3.  [Why do 95% of AI pilots fail to deliver ROI?](#why-do-95-of-ai-pilots-fail-to-deliver-roi)
4.  [What separates AI winners from laggards?](#what-separates-ai-winners-from-laggards)
5.  [The CxO operating playbook: seven sequenced moves](#the-cxo-operating-playbook-seven-sequenced-moves)
6.  [How should executives measure AI ROI in 2026?](#how-should-executives-measure-ai-roi-in-2026)
7.  [Who owns AI strategy now?](#who-owns-ai-strategy-now)
8.  [Related reading](#related-reading)
9.  [Build the capability, not just the strategy](#build-the-capability-not-just-the-strategy)

## Why do most AI investments fail to deliver measurable ROI in 2026?

Because most organizations pilot AI but never rewire the business around it. **88% of organizations regularly use AI** in at least one function and 72% now use generative AI (up from 33% in 2024), yet only about **6% qualify as "AI high performers"** attributing more than 5% of EBIT to AI, per McKinsey's _The State of AI in 2025_ (via [CX Today](https://www.cxtoday.com/ai-automation-in-cx/mckinseys-state-of-ai-the-scaling-gap-is-now-cxs-problem/)). The gap is even starker at the pilot level: MIT's Project NANDA found that despite $30-40B in enterprise GenAI spending, **~95% of pilots delivered no measurable P&L impact** — only about 5% extract real value ([MIT GenAI Divide, via Fortune/Yahoo Finance](https://finance.yahoo.com/news/mit-report-95-generative-ai-105412686.html)). And BCG classifies only **5% of companies as "future-built"** against 60% laggards ([BCG](https://www.bcg.com/press/30september2025-ai-leaders-outpace-laggards-revenue-growth-cost-savings)). AI adoption is nearly universal; value capture is rare. This guide is the sequenced operating playbook for being in the small group that captures it.

## The data: three research houses, one paradox

Three independent 2025 studies triangulate the same story — near-universal adoption, scarce value.

| Metric | Finding | Source |
| --- | --- | --- |
| Organizations using AI | **88%** (72% use gen AI, vs. 33% in 2024) | McKinsey, State of AI 2025 |
| Organizations reporting any EBIT impact | **~39%** | McKinsey, State of AI 2025 |
| "AI high performers" (>5% of EBIT from AI) | **~6%** | McKinsey, State of AI 2025 |
| GenAI pilots with no measurable P&L impact | **~95%** ($30-40B spent) | MIT Project NANDA |
| Companies that are "future-built" | **5%** (35% scalers, 60% laggards) | BCG, Widening AI Value Gap |
| Orgs hoping to grow revenue via AI vs. already doing so | **74% aspire / ~20% achieve** | Deloitte, State of AI 2026 |
| Executives self-describing as AI-"mature" | **~1%** (92% increasing AI spend) | LSE Exec Education (citing McKinsey) |

The bottleneck is not model quality or budget. Spending is rising almost everywhere — 92% of organizations expect to increase AI spend (LSE, citing McKinsey), and 43% of C-suite respondents now prioritize AI and technology investment ahead of product innovation or customer experience (Conference Board 2026 C-Suite Outlook). The differentiator is what leaders do with the organization around the technology.

## Why do 95% of AI pilots fail to deliver ROI?

Because organizations invest in tools and skip the workflow and people changes that turn tools into outcomes. MIT's Project NANDA found **~95% of GenAI pilots delivered no measurable P&L impact**, and the failure pattern is specific and fixable ([MIT GenAI Divide](https://finance.yahoo.com/news/mit-report-95-generative-ai-105412686.html)):

-   **Build-vs-buy misjudgment.** Buying from specialized vendors succeeds about **67% of the time**; internal builds succeed at roughly one-third that rate. Most enterprises default to building.
-   **Budget aimed at the wrong function.** The biggest ROI showed up in **back-office automation**, yet more than half of GenAI budgets go to sales and marketing tools.
-   **The learning gap.** Pilots stall because tools do not retain context or improve inside real workflows — they are bolted on, not built in.

McKinsey's data explains the deeper cause: **only ~21% of adopters have fundamentally redesigned any workflow**, even though workflow redesign is the single strongest predictor of EBIT impact. High performers are **2.8x more likely to redesign workflows (55% vs. 20%)**. Deployment is easy; rewiring the work is the actual job — and most organizations skip it.

## What separates AI winners from laggards?

Winners treat AI as an organizational transformation, not a technology purchase. BCG's _Widening AI Value Gap_ (a survey of **1,250 senior executives** across 9 industries and 41 AI capabilities) sorts companies into three tiers, and the performance spread is large ([BCG](https://www.bcg.com/press/30september2025-ai-leaders-outpace-laggards-revenue-growth-cost-savings)):

| Tier | Share of companies | What they do |
| --- | --- | --- |
| **Future-built** | 5% | Scaled AI across core functions; plan to upskill more than 50% of employees |
| **Scalers** | 35% | Expanding beyond pilots into multiple functions |
| **Laggards** | 60% | Stuck in isolated pilots |

Future-built firms achieve **1.7x revenue growth, 3.6x higher total shareholder return, and 1.6x EBIT margin** versus peers. The mechanism behind that gap is captured in BCG's most citable framework — the **10-20-70 rule** ([BCG Leaders' Guide](https://www.bcg.com/featured-insights/the-leaders-guide-to-transforming-with-ai)):

-   **10%** of the value comes from the **algorithms**
-   **20%** from the **technology and data**
-   **70%** from **people and process** — redesigning roles, managing change, upskilling

Organizations that invert the 10-20-70 ratio — over-spending on tools and under-investing in workforce and process change — consistently fail to generate measurable ROI. If your AI budget is 90% software and 10% change management, you are funding the 30% that does not move EBIT.

## The CxO operating playbook: seven sequenced moves

The research points to a repeatable sequence. Run it in order — each step de-risks the next.

### 1\. Diagnose your maturity tier

Locate yourself honestly on the BCG ladder: laggard (isolated pilots), scaler (multiple functions), or future-built (scaled across the core). Only ~1% of executives call their organization "AI-mature" (LSE, citing McKinsey), so most leaders are earlier on the curve than their board decks suggest. The target state is concrete: the revenue, TSR, and margin deltas of future-built firms.

### 2\. Pick value-first use cases

Start with the business problem and a measurable outcome, not the model or the vendor. Build a **portfolio**: pair quick back-office automation wins — where MIT found the biggest ROI — against a small number of transformational bets. Resist the pull to spend the majority of budget on customer-facing tools simply because they are visible.

### 3\. Escape pilot purgatory

Use MIT's diagnostics directly. Favor **buying from specialized vendors (~67% success) over internal builds (~33%)** unless AI is genuinely core IP. Sequence **back-office-first** to bank measurable P&L wins early. And close the learning gap by embedding tools inside real workflows rather than bolting them alongside.

### 4\. Rewire workflows, not just deploy tools

This is the step 79% of adopters skip. Redesign the actual sequence of work around AI — the strongest predictor of EBIT impact per McKinsey. Deployment adds a tool; redesign changes how the work gets done and where the headcount and cycle-time savings land.

### 5\. Fund the 70% — people and process

Make the 10-20-70 rule an actual budget line. Future-built firms plan to **upskill more than 50% of employees**. Deloitte's _State of AI in the Enterprise 2026_ (3,235 leaders across 24 countries) names **insufficient worker skills as the biggest barrier** to integration — education/AI fluency (53%) and upskilling/reskilling (48%) top the list. Protected learning time and structured AI-fluency programs are strategy, not HR overhead.

### 6\. Govern agents before you scale them

Agentic AI is the next value layer — AI agents account for ~17% of AI value in 2025, projected to reach ~29% by 2028 (BCG). But only about **one in five companies has a mature governance model for autonomous AI agents** (Deloitte). Stand up board-level oversight and responsible-AI guardrails before agents touch production decisions, not after.

### 7\. Score ROI on EBIT attribution

Deloitte's headline gap — **74% of organizations hope to grow revenue through AI but only ~20% are already doing so** — is a measurement failure as much as an execution one. Define KPIs and benefit-realization mechanisms up front, and hold AI initiatives to EBIT attribution the way you would any capital investment.

## How should executives measure AI ROI in 2026?

Tie every initiative to EBIT attribution and a named business outcome before it launches, not after. The Deloitte 74%-aspire-vs-20%-achieve gap exists because most organizations measure activity (pilots launched, tools deployed) rather than value (EBIT moved, cycle time cut, cost removed). A workable scorecard:

-   **EBIT attribution** — what share of margin can you trace to AI? High performers clear 5%; ~6% of organizations get there.
-   **Workflow-redesign coverage** — how many core workflows have been fundamentally redesigned, not just augmented?
-   **Value concentration** — is spend flowing to the functions with proven ROI (back-office) or the ones that merely feel strategic?
-   **Buy-vs-build success rate** — track it; the ~67% vs. ~33% gap is a portfolio-management signal.
-   **Workforce coverage** — what percentage of employees are being upskilled? Future-built firms target more than 50%.

## Who owns AI strategy now?

The CEO and the board — not the CIO alone. AI strategy has moved from a delegated IT initiative to an enterprise-wide, board-level concern: 31% of C-suite leaders cite enhancing AI expertise as a top priority and 27% emphasize cultural readiness (Conference Board 2026 C-Suite Outlook). Notably, **38% of US CEOs see AI as potentially negative** for their business in 2026 (vs. 30% globally) — a signal that leaders increasingly recognize the execution risk of getting this wrong. LSE frames the executive mandate as five capabilities: strategic value identification, governance and responsible-AI oversight, leading organizational change, AI-supported decision-making, and cross-functional alignment. All five are ownership responsibilities, not technical ones.

## Related reading

-   [AI Leadership in 2026](/intel/ai-leadership-2026)
-   [AI Governance in 2026](/intel/ai-governance-2026)
-   [AI Agents for Product Managers in 2026](/intel/ai-agents-for-product-managers-2026)

## Build the capability, not just the strategy

The research is unambiguous: the 70% that decides AI ROI is people and process, and insufficient skills is the top barrier every study names. If you are equipping leaders and teams to actually execute an AI strategy — not just approve one — explore Dexity's AI for Leaders course to build the operating capability behind the playbook above.

_Sources: McKinsey, The State of AI in 2025 (via CX Today); MIT Project NANDA, "The GenAI Divide: State of AI in Business 2025" (via Fortune/Yahoo Finance); BCG, "The Widening AI Value Gap: Build for the Future 2025" and "The Leaders' Guide to Transforming with AI" (10-20-70 rule); Deloitte, State of AI in the Enterprise (2026); The Conference Board, 2026 C-Suite Outlook; LSE Executive Education (citing McKinsey)._

## Frequently asked questions

What percentage of companies actually capture value from AI?

By McKinsey's measure, only about **6% are "AI high performers"** (>5% of EBIT from AI) and ~39% report any EBIT impact. MIT found **~95% of GenAI pilots** delivered no measurable P&L impact, and BCG classifies just **5% of companies as "future-built."** Adoption is near-universal (88% use AI); value capture is rare.

Should we build or buy AI capabilities?

For most functions, buy. MIT's Project NANDA found purchasing from specialized vendors succeeds about **67% of the time versus roughly 33% for internal builds**. Reserve internal builds for cases where AI is genuinely core intellectual property.

Where does AI actually create the most measurable value?

MIT found the **biggest ROI came from back-office automation**, not the sales and marketing tools that receive more than half of GenAI budgets. Sequence quick back-office wins first, then fund transformational bets.

Why is workforce upskilling central to AI strategy?

Because BCG's 10-20-70 rule shows **70% of AI value comes from people and process** change. Deloitte names insufficient worker skills as the top barrier to integration, and future-built firms plan to upskill **more than 50% of employees.** Tools without skills do not move EBIT.

What is "pilot purgatory" and how do we escape it?

It is the state where AI stays stuck in isolated pilots that never scale — where 60% of companies sit (BCG). Escape it by buying over building, sequencing back-office wins to prove P&L impact early, and redesigning workflows rather than bolting tools onto existing ones.

How mature are most companies at governing AI agents?

Not very. Only about **one in five companies has a mature governance model for autonomous AI agents** (Deloitte), even as agents grow from ~17% of AI value in 2025 toward ~29% by 2028 (BCG). Governance should precede scaled agent deployment.

## More in AI at Work

[All Intel →](/intel)

[AI at Work

### AI Agent Frameworks in 2026: LangGraph vs CrewAI vs AutoGen vs OpenAI Agents SDK, Compared

There's no single "best" AI agent framework — the right pick depends on your stack and how much control you need. In 2026 the main options are LangGraph (graph-based, most control, strongest for production and complex multi-agent), CrewAI (role-based "crews," fastest to prototype), AutoGen/AG2 (conversation-driven multi-agent), and the OpenAI Agents SDK (lightweight, handoff-based, Python + TypeScript), plus LlamaIndex Workflows, Google ADK, Pydantic AI (type-safe), and the Microsoft Agent Framework (the AutoGen + Semantic Kernel successor). Many teams skip frameworks and build agents with the model SDK plus a loop — Anthropic and Microsoft both recommend starting there. This guide compares them on language, abstraction, multi-agent support, state/memory, and best-fit, with version and maturity claims attributed and dated (they drift).

17 September 2026 · 12 min read](/intel/ai-agent-frameworks-2026) [AI at Work

### How to Build an AI Agent in 2026: A Practical Guide for Engineers

You build an AI agent by wrapping a large language model in a loop: the model gets a goal, decides an action, calls a tool, reads the result, and repeats until the task is done. The core pieces are a controller model, tools defined as JSON schemas, a memory system, an orchestration loop, and a termination condition. Tools are exposed through function-calling APIs — the model returns a structured call, your code runs it, and you feed the output back. The hard part isn't the loop; it's knowing when NOT to build an agent. Most production systems are workflows with predefined paths, not autonomous agents — and Gartner projects over 40% of agentic-AI projects will be canceled by end of 2027 on cost and unclear value. This guide walks the architecture, tool calling, memory, patterns, failure modes, and deployment.

17 September 2026 · 12 min read](/intel/how-to-build-an-ai-agent-2026) [AI at Work

### How to Evaluate AI Agents: Trajectories, Tool Use, and Task Success (2026 Guide)

Evaluating an AI agent means judging both what it produced and how it got there. Unlike a single LLM call, an agent plans, calls tools, and takes many steps — so a correct final answer can hide a broken path. Combine outcome evaluation (task success rate, final-answer correctness) with trajectory evaluation (tool selection, tool-argument accuracy, step efficiency, error recovery). Score deterministic things with code and reserve LLM-as-judge for subjective quality — while guarding against its position, verbosity, and self-preference biases. Track cost, latency, and step count in the same traces as quality. Run offline evals on a fixed dataset before shipping, then monitor online in production and convert every failure into a permanent regression test. This guide covers the taxonomy, the real agent benchmarks (τ-bench, WebArena, GAIA, SWE-bench, BrowseComp), the tooling, and the mistakes.

17 September 2026 · 12 min read](/intel/how-to-evaluate-ai-agents-2026)

Canonical: https://stg.dexity.com/intel/ai-strategy-for-leaders-2026