There’s a number that’s been floating around the business press for two decades now. Different versions, different sources, different transformations — but the headline never really changes.
Roughly 70% of major transformations fail.
Cloud adoption. Digital transformation. ERP. Platform-based architectures. Agile. Now AI. The technology changes. The failure rate doesn’t.
That should be the most uncomfortable statistic in modern business, because it’s telling us something the industry has been studiously refusing to hear. The failures aren’t about the technology. They’ve never been about the technology. They’re about everything that has to happen around the technology for it to actually work.
And the consulting firms whose entire business is studying this — McKinsey, BCG, Deloitte, the Stanford HAI Index — have been remarkably consistent on the math. Let me walk through what the most recent research actually says, because the picture in 2026 is sharper than the old “70% fail” cliché.
What the 2025-2026 data actually shows
McKinsey’s 2025 State of AI survey tested 25 attributes across nearly 2,000 organizations and found that workflow redesign had the single strongest correlation with EBIT impact from AI. Not model selection. Not data infrastructure. Not vendor choice. Workflow redesign.
High-performing organizations — the roughly 6% that reported meaningful financial returns — were nearly three times more likely to have fundamentally redesigned their workflows around AI rather than layering AI onto existing processes.
BCG’s 2025 global study of 1,250 companies tells the same story in different words. Only about 5% create substantial AI value at scale, while 60% generate no material value from their AI investments despite meaningful spending.
Sixty percent. Spending real money. Getting nothing back. Not because the technology doesn’t work — the technology works fine — but because the differentiator is not the technology. It’s whether the organization treats AI as a tool to add or a reason to redesign how work gets done.
This is the punchline buried under thousands of pages of consulting research: the technology is the easy part.
The 10-20-70 principle
BCG has been promoting a framework that captures this better than any other framing I’ve seen. They call it the 10-20-70 principle.
Companies should devote 10% of their efforts to algorithms and 20% to technology and data; the remaining 70% of their efforts should focus on people and processes to make sure that the changes stick.
Read that again, because it’s almost the inverse of how most enterprise AI budgets are actually allocated.
Most leaders I work with are spending roughly the opposite. Seventy percent of the budget goes to platforms, tools, models, vendors, and infrastructure. Twenty percent goes to data and integration. Ten percent — if that — goes to the people and processes that determine whether any of it actually changes the way work gets done.
Then they’re surprised when nothing changes.
The 10-20-70 numbers are directional, not absolute. MIT Sloan found that 70% of AI’s value depends on complementary investments in people and process, not on the sophistication of the technology. Different study, different methodology, same conclusion.
The technology is increasingly commoditized. Any mid-market company can access the same frontier models from OpenAI, Anthropic, or Google. What’s not commoditized is your organization’s capacity to absorb, adopt, and operationalize that technology. That’s the moat now.
This pattern is not new
Here’s the part that should give every leader pause.
This isn’t a new finding. We’ve seen this same pattern in every major technology wave of the last quarter-century.
Cloud adoption (2010-2020). Most enterprises that moved workloads to AWS or Azure without changing their operating model got cloud bills instead of cloud value. The 70% that struggled weren’t fighting the technology — they were fighting their own org charts, procurement processes, and engineering cultures.
Digital transformation (2015-present). McKinsey’s research on digital transformation has reported failure rates of 70-80% for over a decade. Same root cause. Companies bought new tools and tried to run them on old operating models.
Platform-based architectures (2018-2024). Microservices, APIs, event-driven systems. The tech worked. The pattern of failure was the same — teams structured around the old monolith couldn’t suddenly behave like teams owning loosely coupled services, no matter what the architecture diagram said.
Agile adoption (2010-present). This one is almost a parody at this point. Every large enterprise has “adopted agile.” Very few have actually changed how decisions get made, how funding flows, or how work gets prioritized. The framework is on the wall. The behaviors haven’t moved.
In every case, the pattern is identical:
- The technology works.
- The leaders adopt it.
- The org doesn’t absorb it.
- Two years later, someone declares the transformation “stalled.”
- A new technology arrives.
- The cycle restarts.
AI is going to follow this script unless leaders deliberately choose otherwise.
What “deliberately choosing otherwise” actually looks like
The good news is that the small minority of organizations getting this right are extremely well-studied. We know what they do differently.
High performers are 3.6x more likely to pursue transformational change and 55% fundamentally rework workflows when deploying AI. They’re not adding AI on top. They’re rebuilding the work around AI.
They invest the 10-20-70 the way BCG describes it — most of the money goes to change capacity, not to compute.
They have executive sponsors who are publicly committed and operationally engaged, not just funding the program from a distance.
They build shared language across the organization, so the CEO, the CIO, the frontline manager, and the analyst all mean the same thing when they say “AI.” Without shared language, every conversation restarts from zero.
They redesign workflows end-to-end rather than automating existing ones. If you take a broken, manual, approval-heavy workflow and add AI on top of it, you get a slightly faster broken workflow. That’s not transformation. That’s expensive friction reduction.
They measure value rigorously. Not “did we deploy the tool” but “did the work change, did the cycle time drop, did the cost-to-serve improve, did the customer outcome improve.”
And — this is the one most leaders skip — they invest in workforce capability ahead of deployment. Companies that are realizing the most value from AI also have the most ambitious upskilling programs, and put the resources in place to support them.
The hard truth for Canadian enterprise leaders in 2026
If you’re running an AI program right now, you are statistically much more likely to end up in the 60% generating no material value than in the 5% capturing real returns.
That’s not pessimism. That’s what the data says.
The path to being in the 5% is not picking a better LLM. It’s not waiting for the next model release. It’s not hiring more AI engineers.
It’s investing the 70% of effort that almost everyone underspends — change capacity, workflow redesign, executive fluency, shared language, workforce capability, governance that doesn’t choke delivery.
This is exactly why the AI-Native curriculum is structured the way it is. Foundations builds shared language. Change Agent develops the people who can actually land the change. Leading the AI-Native Organization equips executives to make the release-rate, prioritization, and workforce decisions that determine whether the 70% gets the attention it deserves.
The data has been telling us this story for twenty years. Cloud. Digital. Platforms. Agile. Now AI.
The pattern doesn’t change because the technology changes.
The pattern changes because leaders choose to invest where the value actually lives.
The closing thought
There’s a comforting story leaders sometimes tell themselves: this time it will be different, because this technology is more powerful.
It’s not going to be different. It’s never been different.
The technology was powerful in 2005. It was powerful in 2015. It’s powerful in 2026. And in every era, the same 60-70% of organizations have failed to extract value from it for the same reason — because they treated transformation as a technology problem when it has always been a people-and-process problem.
The 5% who get this right aren’t smarter. They’re not better resourced. They’re not luckier.
They’re just willing to spend their effort where the value lives — not where the marketing budget is loudest.
Build the 70%. The other 30% will follow.
