"We are exiting on a compressed timeline people where reskilling is not a viable path." That was Julie Sweet, Accenture's CEO, last week. Behind the quote: 11,000 roles cut, 70,000 staff trained in AI, 4,000 new hires specifically for AI capability. One quarter of activity. One of the largest professional services firms in the world just rewrote its workforce around AI.
You saw the headline in the weekly roundup. What the roundup didn't cover — and what makes Accenture the rare firm actually executing this — is the data that explains why so few firms are able to.
McKinsey published its State of Organizations 2026 report last month. Ten thousand senior leaders across fifteen countries. The headline number everyone will repeat: 88% of organizations are deploying AI. That's not the number that matters.
The number that matters is 14%. That's the share of organizations where leaders consistently champion AI with a clear strategy. Not "we bought licenses." Not "we sent the team through a course." Consistently championing, with a strategy. Fourteen percent.
Which means 86% of organizations have AI tools, AI budgets, maybe even AI training — and no one at the top personally driving the outcome. If you're wondering why your practice's AI experiments aren't turning into firm-wide capability, this is probably why. The tools aren't the bottleneck. You are.
The fifth wall is in the partner's office
I wrote recently about the four walls your team keeps hitting — the Better Google, the Platform Wall, the Output Fantasy, the Deployment Gap. Those are real failure modes. But they're team-level failure modes. McKinsey's data adds the floor above.
There's a fifth wall. When the person who controls budget, priorities, and firm culture doesn't personally use AI, doesn't protect the investment when fees tighten, and doesn't sponsor the first builds — the team hits the other four walls faster and with less support. The 14% don't have smarter teams. They have leadership that owns the outcome.
The ratio that explains your training problem
McKinsey's sharpest finding for practice owners: for every dollar spent on AI technology, organizations should invest five dollars in the people who'll use it. And I want to be clear about what that five dollars is — it isn't salaries. It's training. It's workflow redesign time. It's protected experimentation budget. It's the tooling around the tool. Everything that turns a subscription into a capability.
Firms that get this ratio right are four times more likely to maintain top-tier financial performance. Five to one. People over tools. Most CAS practices have it inverted. Subscriptions first, certifications maybe, and then surprise that nothing changed. The skills gap piece showed what happens downstream — teams collecting certificates instead of building artifacts. McKinsey just quantified the upstream cause. The investment was wrong from the start.
This connects to which problem you're solving. If AI is a cost play — do the same work faster — the people investment feels unnecessary. (I'm going deeper on the cost-versus-revenue distinction on Thursday.) If it's a transformation play — new services, deeper advisory, rebuilt delivery — five dollars on people for every dollar on tools makes obvious sense. The 86% are buying tools and hoping. The 14% are investing in the humans who make the tools work.
What the 14% actually do at CAS scale
Back to Accenture. They made the leadership move that McKinsey says 86% haven't. The same week, PwC restructured its entire consulting network around an AI-mediated delivery platform. Two of the biggest advisory firms in the world, both in the 14%.
You don't run Accenture. You run a much smaller practice. But the behaviors translate.
Use AI yourself. Daily. Not "know about it" — use it. Draft the scope letter with it. Prep for the advisory call with it. If your team sees you using AI in the room with a client, the permission structure changes overnight.
Protect the budget. When fee pressure hits, AI investment is the last thing cut, not the first. The training hours. The experimentation time. The subscriptions. The 86% cut these first and wonder why the team stopped building.
Sponsor the first builds personally. The Champion + Operator framework works — but only if the Champion has enough authority to make the build matter. In a CAS practice, that's the partner. Pick the first three workflows. Sit in the room while they're built. Decide what ships.
Make the uncomfortable call. The 14% includes firms like Accenture that exited people who couldn't reskill. You probably don't need to fire anyone. But you do need the same willingness to restructure roles, reallocate time, and stop protecting processes that existed before AI did.
The question you're avoiding
McKinsey surveyed 10,000 leaders. Seventy-five percent of roles need fundamental reshaping — not next year, now. The practices that figure this out build a compounding advantage. Every month your team operates in reshaped roles, they generate the structured data that makes the next round of AI better. Every month they don't, the gap widens.
Most partners want their team to change. They want someone on staff to become the AI champion and transform the practice. McKinsey's data says it doesn't work that way. The 14% don't have better teams. They have better leadership.
Which side of the line are you on — and what are you doing about it this week?
If you're ready to step into that leadership role yourself, and your firm is large enough that the structural change is the point, AI Practice Transformation is the cohort program built for exactly that — partners and champions committed to actively leading AI transformation inside their own firm. Three weeks, one day a week, five hours a session. theaiaccountant.ai/transformation.
If your firm is smaller — solo or under ten — and you want to start by equipping your team to explore and build, without the heavier structural overhaul, AI Essentials is the lower-commitment entry point. Platform, curated workflows, guided onboarding, and a monthly live implementation call. theaiaccountant.ai/essentials.
Either way: the 14% don't have smarter teams. They have leadership that went first. The question is whether you will.