AI Transformation Is Workforce Transformation: The 70% Most Organisations Are Missing

Only 5% of organisations have achieved substantial financial gains from AI.

This is not because AI doesn't work. It is not because the technology is immature. It is because 95% of organisations are treating AI as a technology procurement when it is actually a workforce transformation.

BCG's 2025 Global Study of enterprise AI — surveying thousands of organisations across industries — found that the 5% capturing AI value show total shareholder returns roughly four times higher than AI laggards. The difference is not better algorithms. The difference is what they do with people.

This article explores BCG's 10/20/70 framework — the evidence for why workforce transformation is the determinant of AI value — and what the organisations that are actually capturing AI returns do differently.

The 10/20/70 Framework

Where does AI value actually come from? BCG's analysis offers a striking answer.

The framework is counterintuitive. The component that receives the most attention — algorithms, model selection, prompt engineering — accounts for the least value. The component that receives the least attention — people, skills, roles, operating models — accounts for the most.

This is not a commentary on the importance of good technology. It is an observation about what determines whether good technology produces good outcomes. A better model with no workforce transformation delivers no better result. A simpler model with deep workforce transformation can outperform.

The implication is uncomfortable for how most AI programmes are structured. Technology-led AI initiatives — driven by IT, digital, or data science teams — address the 30%. Workforce-led AI initiatives — driven by HR, operations, and leadership — address the 70%. Most organisations are investing heavily in the 30% while treating the 70% as a support function.

What the 5% Do Differently

The organisations capturing AI value — BCG calls them "future-built" companies — exhibit five behaviours that separate them from the 95%.

The gap is not subtle. Future-built companies plan to upskill more than 50% of their workforce on AI capabilities. Laggards plan to upskill 20%. Future-built companies are four times more likely to have structured learning programmes with protected time — not training that competes with delivery deadlines, but training that is built into workload.

Most significantly, 84% of organisations have not redesigned jobs or workflows around AI capabilities, according to Deloitte's 2026 State of AI in the Enterprise. They have deployed AI tools into existing roles. Same responsibilities, same success metrics, new tools. This is not transformation — it is layering.

The consequence of layering is adoption decay. Employees try the tool, encounter friction with how their work is actually structured, and quietly revert to old habits. The AI that worked in demos stops being used in production. The initiative is declared a partial success, and the organisation moves on.

Transformation requires redesigning the work — not just adding tools to it.

The 84% Problem

Deloitte's finding deserves emphasis: 84% of organisations have not redesigned jobs around AI.

This is the root cause of the gap between AI investment and AI returns. Organisations are spending on algorithms, infrastructure, and deployment — the 30% — while leaving the 70% unchanged. They are training employees on tools without redesigning how those employees work. They are deploying AI agents without clarifying who owns what when the agent handles part of a task but not all of it.

The result is a productivity paradox. Workday's 2026 research found that 85% of employees report saving 1–7 hours weekly using AI tools. That sounds like success. But nearly 40% of those time savings are lost to rework — fixing errors, rewriting content, verifying outputs. Only 14% of employees consistently achieve clear, positive net outcomes from AI.

The problem is not the AI. The problem is that work has not been redesigned to accommodate AI's strengths and limitations. Employees are using AI to produce first drafts — and then spending the saved time correcting what the AI got wrong. The workflow assumes AI is accurate. The reality is that AI requires curation, validation, and judgment. Without redesigning the workflow to include those steps, the efficiency gains evaporate in rework.

Skills Evolution, Not Replacement

The conversation about AI and workforce often defaults to replacement — which jobs will AI eliminate? The research tells a more nuanced story.

McKinsey's 2025 analysis found that more than 70% of skills employers seek today are used in both automatable and non-automatable work. Skills are not disappearing. They are evolving — applied differently in a world where AI handles certain tasks and humans handle others.

The shift is consistent across roles. Workers move from producing to curating — from creating first drafts to framing questions and validating outputs. They move from executing to exercising judgment — from doing the task to deciding if the AI's output is right. They move from individual contribution to orchestration — from solo execution to coordinating how tasks flow between humans and AI.

This shift requires intentional skill development. Demand for AI fluency — the ability to use, manage, and work alongside AI tools — has grown sevenfold in just two years, faster than any other skill category. But fluency alone is not enough. Organisations also need to develop curation skills (knowing what good output looks like), judgment skills (knowing when AI recommendations don't apply), and orchestration skills (designing workflows that flow between humans and AI).

The World Economic Forum's Future of Jobs Report 2025 found that 63% of employers cite skills gaps as the largest barrier to transformation. The gap is not that employees lack AI skills — it is that employees lack the evolved skills that AI-augmented work requires. Organisations that invest in training without investing in skill redefinition are training for yesterday's work.

The Investment Gap

How much should organisations invest in workforce transformation? The data suggests the answer depends on what kind of return you expect.

The baseline — $1,800 per employee, consistent with industry benchmarks from SHRM — delivers approximately 9% productivity improvement. This is the ad-hoc approach: self-directed training, no protected learning time, tools deployed without role redesign.

At $3,600 per employee — structured programmes with some protected time — organisations see approximately 15% improvement. At $5,400 or more — protected learning time, role redesign, operating model updates — organisations see 25% or greater improvement.

The difference is not just more money. It is a fundamentally different approach. The high-investment tier integrates training with workflow redesign and operating model change. It recognises that you cannot train someone to use AI effectively if their job is still designed around manual work. The low-investment tier treats training as an add-on — something employees do on top of their existing work, without changing what that work is.

The ROI gap compounds over time. Organisations that invest more see larger productivity gains, which fund further investment in AI capabilities, which enable more ambitious transformation. Organisations that invest less capture smaller gains, see weaker business cases for further investment, and fall further behind. The gap between leaders and laggards is not closing. It is widening.

What Workforce Transformation Actually Requires

The research converges on five requirements for workforce transformation that captures AI value:

1. Leadership ownership at CEO level. AI transformation is not a technology initiative that can be delegated to IT or digital teams. BCG's research found that organisations with CEO-level ownership of AI transformation — with 3–4 central priorities and strategic alignment across the organisation — are four times more likely to achieve substantial returns.

2. Upskilling at scale, not at the margins. The 5% plan to upskill more than 50% of their workforce. The 95% treat upskilling as something for early adopters or technical teams. Workforce transformation requires reaching the majority, not the minority.

3. Protected learning time. Training that competes with delivery deadlines loses. Future-built organisations build learning into workload — dedicated time that is not discretionary. This is operationally expensive and strategically necessary.

4. Role and workflow redesign. Tools without redesign is layering, not transformation. Every AI deployment should be accompanied by explicit redesign of the roles and workflows it affects. Who is responsible for curation? Where does human judgment apply? How do tasks flow between humans and AI?

5. Integrated operating model. Human-AI collaboration requires an operating model designed for it — clear task ownership, escalation paths, quality gates, and accountability. Bolting AI onto an operating model designed for human-only work produces friction, workarounds, and adoption decay.

The Economic Case

The economic argument for workforce transformation is increasingly clear. MIT Sloan research found that companies adopting AI extensively show 6% higher employment growth and 9.5% higher sales growth over five years. The gains come not from replacing workers but from using workers more effectively alongside AI.

McKinsey projects that $2.9 trillion in economic value could be unlocked through AI-powered agents and automation in the United States by 2030 — contingent on organisations redesigning workflows rather than automating tasks in isolation. The value is available. The constraint is organisational, not technological.

The capital markets are beginning to price this gap. BCG's research shows four times higher total shareholder returns for future-built companies. Organisations that invest in workforce transformation are building competitive advantages that compound over time. Organisations that treat AI as a technology procurement are buying tools without building capability.

The Decision

AI transformation is workforce transformation. The technology is ready. The constraint is organisational — whether leadership treats AI as something IT deploys or something the entire organisation transforms around.

The 10/20/70 framework offers a diagnostic. Where is your investment going? If it is concentrated in the 30% — algorithms and infrastructure — you are optimising for the minority of value. If it reaches the 70% — upskilling, role redesign, operating model evolution — you are optimising for where value actually lives.

The 5% are not deploying better AI. They are transforming how people work. That transformation is available to any organisation willing to make the investment — in time, in protected learning, in role redesign, in operating model change.

The question is not whether AI is ready. The question is whether your workforce strategy is.

---

Sources

• BCG (2026): AI Transformation Is a Workforce Transformation
• Deloitte (2026): The State of AI in the Enterprise
• McKinsey (2025): Agents, Robots, and Us: Skill Partnerships in the Age of AI
• World Economic Forum (2025): Future of Jobs Report 2025
• MIT Sloan (2025): How Artificial Intelligence Impacts the US Labor Market
• Workday (2026): Companies Are Leaving AI Gains on the Table
• Rework / SHRM (2025): Corporate AI Reskilling Budget Benchmarks