What BCG's Own Research Proves: AI Success Is 70% People, Not Technology
BCG's 10-20-70 rule says AI success is 10% algorithms, 20% technology, and 70% people. The evidence across industries confirms it — and most organizations are still investing backward.
The Rule That Keeps Getting Ignored
Boston Consulting Group has been publishing what it calls the 10-20-70 rule across multiple AI research reports: successful AI transformation is 10% algorithms, 20% technology, and 70% people and process change. The formula is memorable, widely cited, and almost universally ignored in practice.
Most organizations invert the ratio. They spend the majority of their AI budget on tools, platforms, and model development — then allocate whatever remains to training, change management, and workflow redesign. The result is a predictable pattern: technically functional AI systems that fail to deliver value because the humans who interact with them are untrained, the processes around them are unchanged, and the organizational culture treats AI as an IT project rather than a business-wide shift.
BCG’s own data makes the case unambiguously. Companies that invest proportionally in people and process change alongside technology generate 2-3x higher returns from AI initiatives than those that lead with technology alone.
The Training Gap Is Getting Worse
Despite the rapid acceleration of AI deployment across industries, workforce readiness is not keeping pace. Salesforce’s 2026 Global AI Workforce Survey found that only 36% of employees feel adequately trained on the AI tools their organizations have deployed. That means nearly two-thirds of the workforce is interacting with AI systems they do not fully understand — a gap that manifests as underutilization, mistrust, and workaround behaviors that erode the value AI was supposed to create.
The training gap is not limited to technical skills. Employees need to understand when to rely on AI outputs and when to override them, how to construct effective prompts and instructions, how to evaluate AI-generated content for accuracy and bias, and how to escalate issues when automated systems produce unexpected results. These are judgment skills, not technical skills — and they require sustained investment in education, not a one-time onboarding session.
Organizations that treat AI training as a checkbox rather than an ongoing capability-building program consistently report lower adoption rates, higher error rates, and faster employee burnout in AI-adjacent roles.
The Case Studies That Prove the Point
Two widely reported AI deployments from 2025 illustrate the 70% principle in practice — one through success, one through the uncomfortable truths behind it.
Klarna deployed AI customer service agents that reduced average resolution time from 11 minutes to under 2 minutes, with a projected $40 million annual cost reduction. The technical achievement was impressive, but the organizational transformation was the real story. Klarna restructured its entire customer service operation around the AI system: redefining human agent roles, redesigning escalation workflows, retraining staff as AI supervisors rather than frontline responders, and implementing quality assurance processes for AI-generated interactions. The technology worked because the organization changed around it.
Allstate deployed generative AI tools for internal knowledge management and saw 3x improvement in employee productivity within the first four weeks of deployment. But the speed of the result was not accidental — Allstate invested heavily in structured onboarding, embedded AI champions within business units, and created feedback loops that allowed employees to surface problems and suggest improvements in real time. The 70% investment in people and process was what produced the 3x result, not the model.
Why 3-4 Use Cases Beat “AI Everywhere”
One of the most counterintuitive findings in BCG’s research — and corroborated by McKinsey, Deloitte, and Gartner independently — is that organizations focusing on three to four high-impact use cases consistently outperform those pursuing broad, organization-wide AI deployment.
The reason is resource concentration. AI adoption requires change management bandwidth that is inherently limited: executive attention, training capacity, process redesign expertise, and organizational patience. Spreading these finite resources across dozens of simultaneous initiatives dilutes impact. Concentrating them on a few well-chosen use cases creates visible success stories, builds internal capability, and generates momentum that makes subsequent deployments faster and less expensive.
BCG itself followed this pattern internally, deploying over 18,000 GPT-powered agents across its consulting practice — but doing so incrementally, starting with a small number of high-frequency use cases (research synthesis, document drafting, data analysis) and expanding only after those initial deployments demonstrated measurable value and organizational readiness.
The 70% That Actually Matters
The practical implications of the 10-20-70 rule are straightforward but demanding:
Allocate training budget proportional to tool budget. If an organization spends $500,000 on AI platforms, it should be spending at least as much on training, change management, and process redesign. Most spend a fraction of that and wonder why adoption stalls.
Redesign workflows, not just add AI to existing ones. Layering AI on top of unchanged processes captures, at best, incremental efficiency. Redesigning the process around AI’s capabilities — automation, parallelism, real-time analysis — is where the real gains are.
Measure adoption, not just deployment — and make AI literacy a basic requirement. A deployed AI system that employees don’t use is a sunk cost. Adoption metrics (active usage, task completion, employee satisfaction) should carry equal weight to technical performance metrics. And just as digital literacy became a baseline expectation over the past decade, AI literacy is becoming one too. Organizations that build this skill across all levels — not just in technical teams — will hire better, retain longer, and get more out of every AI dollar.
The technology is the easy part. The 70% is where AI initiatives succeed or fail.
Organizations looking to build AI-ready teams and processes can explore Toblero’s consulting services.