When three independently conducted studies from the world's most influential strategy firms reach the same conclusion, the signal is worth listening to. In 2025, McKinsey, Bain, and BCG each published major AI research — different methodologies, different sample populations, different analytical frameworks. They arrived at the same finding: the primary determinant of enterprise AI value is not the model, the vendor, or the technology stack. It is whether the organisation redesigns its workflows around AI capabilities or simply layers AI onto existing processes.
This is not a nuanced difference. It is the difference between capturing transformative value and wondering why AI investments are not producing returns.
McKinsey: 2.8x the value from workflow redesign
McKinsey's 2025 State of AI survey provides the most granular data on this question. Across 1,993 respondents in 105 countries, the survey identifies a small cohort of "high performers" — the 6% of organisations that attribute more than 5% of EBIT impact to AI. When McKinsey examines what separates this cohort from the rest, one variable dominates: workflow redesign.
High performers are 2.8 times more likely to have fundamentally redesigned workflows around AI. The numbers are stark: 55% of high performers report significant workflow redesign versus 20% of the remaining organisations. Nearly 80% of all respondents layer AI onto existing processes without rethinking how the work flows through the organisation.
Only 21% of generative AI users have redesigned at least some workflows. The remaining 79% have adopted AI tools — chatbots, copilots, summarisation engines, code assistants — and deployed them within process architectures that were designed for humans working without AI. The tools are faster. The processes are unchanged. The value capture is a fraction of what is available.
This pattern is not surprising. It mirrors every previous technology wave. When spreadsheets arrived, most companies replicated paper ledgers in electronic form. The companies that redesigned financial workflows — building scenario models, automating reconciliation, creating real-time reporting — captured orders of magnitude more value from the same technology. When email arrived, most companies replicated paper memos in electronic form. The companies that redesigned communication workflows — eliminating approval chains, enabling asynchronous coordination, collapsing organisational distance — built fundamentally faster organisations.
AI is following the same adoption curve. The technology is adopted broadly. The workflow redesign that unlocks its value is adopted narrowly. The scaling gap is a workflow gap.
Bain: the biggest determinant is not the model
Bain's 2025 Technology Report states the finding with unusual directness: the biggest determinant of AI transformation success is not the sophistication of the models but the quality of workflow redesign and data cleanup. This is Bain, not an academic paper. When a strategy firm tells its clients that model selection is secondary to process architecture, the implication for enterprise AI strategy is profound.
Bain identifies five critical actions for AI transformation. Action number three — positioned at the centre of their framework — is to redesign entire workflows rather than deploy point solutions. The framing is deliberate: not "optimise processes" or "enhance workflows," but "redesign entire workflows." The distinction between optimisation and redesign is the distinction between faster horses and automobiles.
The economic evidence supports the distinction. Bain reports that tech-forward enterprises moving from single-task AI deployments to workflow-level agent systems are achieving 10 to 25% EBITDA gains. Single-task deployments — a chatbot here, a document classifier there, a code assistant for the engineering team — produce measurable but incremental gains. Workflow-level deployment — where AI handles an end-to-end process with human oversight at defined decision points — produces structural improvements in operating economics.
Bain also makes a governance argument that directly supports redesign. Their second critical action is to charge general managers, not CIOs, with AI transformation. The logic is that workflow redesign is a business decision, not a technology decision. A CIO can deploy AI tools. Only a general manager can redesign how a business function operates. When AI transformation is framed as an IT project, it produces tool procurement. When it is framed as an operating model change, it produces workflow redesign. The organisational ownership structure determines the outcome.
BCG: value concentration in core functions
BCG's analysis approaches the same conclusion from a different angle. Rather than comparing high performers to low performers, BCG examines where AI value concentrates across the enterprise. Their finding: 70% of AI's potential value is concentrated in core business functions — R&D, sales and marketing, supply chain, and pricing. These are not peripheral support functions where AI automates administrative tasks. They are the functions where the work directly produces revenue, margin, and competitive advantage.
BCG's concept of the "future-built company" is fundamentally about workflow redesign. Future-built companies do not simply adopt AI tools across functions. They systematically rebuild capabilities within functions, redesigning how R&D generates insights, how sales qualifies and closes deals, how supply chains sense and respond to demand signals, and how pricing adapts to market conditions. The distinction is between an organisation that uses AI and an organisation that is built around AI.
The 70% concentration finding explains why tool-level deployment underwhelms. If most AI value sits in core business functions, and those functions are defined by complex, multi-step workflows involving cross-functional coordination, then deploying AI at the individual task level — summarise this document, draft this email, classify this ticket — addresses the periphery while leaving the core untouched. A sales team using AI to write prospecting emails captures a small slice of the value available. A sales function redesigned around AI-driven lead scoring, automated qualification, dynamic pricing, and intelligent pipeline management captures the majority.
The anti-pattern all three firms describe
Across their analyses, McKinsey, Bain, and BCG describe the same anti-pattern with remarkable consistency. The pattern goes like this: an enterprise procures AI tools, deploys them into existing processes, measures adoption (how many people are using the tools), and reports progress. Meanwhile, the workflows those tools are embedded in remain unchanged. The same people do the same work in the same sequence with the same decision points — just with an AI assistant available at certain steps.
This is what McKinsey's 80% number represents. It is what Bain calls "point solutions" as opposed to "workflow-level deployment." It is what BCG implicitly critiques when they distinguish between companies that use AI and companies that are built around AI.
The anti-pattern persists because it is organisationally comfortable. Workflow redesign requires changing how people work, which triggers resistance. It requires redefining roles, which triggers anxiety. It requires new governance structures, which requires executive attention. It requires cross-functional coordination, which requires breaking silos. Procuring an AI tool and adding it to an existing process requires none of these things. The tool approach is easier, faster, and produces visible activity. It just does not produce material value.
This is the automation versus augmentation question at enterprise scale. Automation replaces a human step with an AI step within an existing process. Augmentation redesigns the process to leverage what AI does well and what humans do well as complementary capabilities. The firms are unanimous: augmentation through redesign produces multiples of the value that automation through insertion produces.
What redesign actually looks like
Workflow redesign is not abstract. It follows a concrete pattern. First, map the current workflow — not the idealised version in the process documentation, but the actual workflow as it operates, with all its workarounds, exceptions, and informal practices. Process mining is the discipline that produces this map.
Second, identify where AI capabilities change the economics of the workflow. Which steps are high-volume and pattern-dense — suitable for AI automation? Which require judgment but would benefit from AI-generated recommendations? Which are genuinely novel and should remain human-driven? The six dimensions of the AI Operating System provide the framework for this analysis.
Third, redesign the workflow around the new capability distribution. This is the step most organisations skip. It means changing the sequence of steps, the decision points, the escalation paths, the roles involved, and the metrics used to measure performance. A redesigned claims workflow does not look like the old claims workflow with an AI step added. It looks like a fundamentally different process — one where AI handles classification, routing, and initial assessment as a single automated flow, while human adjusters focus on complex cases, customer communication, and quality oversight.
Fourth, implement governance for the redesigned workflow. Who monitors the AI's decisions? What thresholds trigger human review? How are exceptions handled? What happens when the model's accuracy degrades? The decision architecture framework addresses these questions systematically.
The methodology alignment
The AI Operating System methodology is built on the same principle that McKinsey, Bain, and BCG reached independently. The methodology does not start with technology selection. It starts with workflow analysis. The six dimensions — strategy, data, technology, organisation, governance, and operations — are the dimensions of workflow redesign. The Three Levels of AI integration define how deep the redesign goes: Level 1 redesigns individual workflows, Level 2 redesigns entire functions, Level 3 redesigns the enterprise operating model.
This is not retrospective alignment. The methodology was developed from 25+ DACH enterprise engagements where the same pattern appeared repeatedly: the clients who redesigned workflows captured multiples of the value compared to the clients who deployed tools. What the Big 3 have now validated across thousands of survey respondents and hundreds of case studies is what operational experience in the Mittelstand already demonstrated — that the workflow is the unit of AI value, not the model.
If your organisation is using AI tools but has not redesigned the workflows those tools operate in, the evidence from McKinsey, Bain, and BCG is consistent: you are capturing a fraction of the available value. A Fit Call starts with your current workflows and identifies where redesign creates the largest impact — before any technology decision is made.
References: McKinsey & Company, "The State of AI: How Organizations Are Rewiring to Capture Value," Global Survey, November 2025; Bain & Company, "Technology Report 2025," 2025; BCG, "AI Radar 2025: From Potential to Profit," 2025; BCG, "How to Build a Future-Built Company," 2025; McKinsey Global Institute, "The Economic Potential of Generative AI," updated estimates 2024–2025.