Microsoft surveyed twenty thousand AI users across ten markets this spring, then layered telemetry from trillions of Copilot signals on top. They ran random forest analysis across twenty-nine factors. The model explained roughly two-thirds of the variance in AI outcomes. That is a serious instrument. The numbers it produced should change how we run organizations.
The headline finding is the one most people skim past. Sixty-seven percent of AI impact comes from organizational factors. Thirty-two percent comes from the individual. The top three drivers are all organizational. The strongest single org factor is roughly two and a half times stronger than the strongest individual factor.
Translate that into plain English. The person you hire matters. The organization they walk into matters more than twice as much. Most companies have it backwards.
The three numbers that name the gap.
Buried in the Microsoft report is a smaller table that tells a sharper story. When a manager actively models AI use, three things move:
Employees of those managers are 1.4 times more likely to be high-frequency AI users. Same training. Same tools. Same compensation. The difference is whether the boss touches the product.
I call these Frontier managers because Microsoft does. They sit inside the nineteen percent of firms Microsoft labels Frontier organizations. Both axes are high. Individual readiness is strong. Organizational readiness is strong. The two reinforce each other.
The other zones tell the loss story. Ten percent of firms sit in Blocked Agency, where individuals are ready but the org is not. Sixteen percent are Stalled. Fifty percent are Emergent, still forming on both axes. Five percent have unclaimed capacity, where the org is ready but the people are not yet. If you add it up, thirty-one percent of AI users are actively misaligned with their own organization. That is the Coordination Tax in survey form.
The Frontier manager behavior set.
Microsoft compared Frontier managers against everyone else. Four behaviors separated them. The gaps are not small.
- Openly uses AI: 85 percent versus 64 percent
- Sets quality standards for AI work: 83 percent versus 57 percent
- Creates space for experimentation: 84 percent versus 61 percent
- Encourages work redesign with AI: 87 percent versus 61 percent
One more behavior carries the most weight. Frontier managers are twice as likely to reward reinvention even when it has not yet produced a result. Twenty-six percent versus eleven percent. They protect the experiment that has not paid off yet. They give air cover to the person trying something new before the spreadsheet supports the choice.
This is the manager behavior set the rest of the organization watches. The Say/Do Ratio I have been writing about for years is the question that emerges from this data. What the manager says about AI matters less than what they reach for at 10 a.m. on a Tuesday.
Why this is structural, not personal.
The temptation is to read these numbers and think the answer is a manager training program. It is not. I have been running change programs at scale for the better part of two decades. Manager training without a system change produces a brief lift and a slow regression. You can teach the behaviors. You cannot install them through a course.
The behaviors are downstream of three structural conditions.
First, the manager has a defensible reason to use the tool. If the manager's calendar is fully booked with status meetings and approval cycles, the manager will not pause to learn the new instrument. The Kill List in Phase 1 of the Velocity Framework is what creates the reason. Removing eight hours of zombie work per manager per week is what makes the AI question viable for the manager in the first place.
Microsoft has its own name for this dynamic. They call it the Transformation Paradox. Only twenty-six percent of employees say their leadership is clearly and consistently aligned on AI. Sixty-five percent fear falling behind if they do not use it. Forty-five percent say it feels safer to focus on current goals than redesign the work. Only thirteen percent say reinvention is rewarded even without results.
"Employees are ready to reinvent how they work, but the system around them, the metrics, the incentives, the norms, continues to reinforce the old way." Microsoft Work Trend Index 2026.
That is a structural diagnosis, not a motivational one. The Frontier manager behavior set is what emerges when the structure stops punishing the behavior the company says it wants.
Second, the manager is rated on whether the team redesigned the work, not whether the team consumed the tool. Adoption metrics that count seats and prompts produce theater. Outcome metrics that count work that used to take eight hours and now takes thirty minutes produce Frontier behavior. The performance management instrument has to change. If the rating sheet still rewards effort and visibility, the manager will still model effort and visibility.
Third, the manager has explicit permission to fund a non-result. Twenty-six percent versus eleven percent on rewarding reinvention. That is not a personality difference. It is a policy difference. Somebody told that manager, in writing, that protecting an experiment that did not yet ship was a green-light behavior. The other managers have not been told that.
What Clarity Certification certifies.
Phase 3 of the Velocity Framework exists to lock these three structural conditions into the operating rhythm. Phase 1 cleared the calendar. Phase 2 ran the pilot with the chosen agents. Phase 3 is the certification that the Frontier manager pattern is now the default, not a personality of the manager who happens to be enthusiastic.
Three artifacts come out of the phase.
The Frontier Manager Rubric. A one-page instrument that names the four behaviors and the reinvention-rewarded fifth, and that is wired into the performance review template. Managers know which behaviors will be observed. Employees know which behaviors to expect from their manager. The instrument is shared, not secret.
The Say/Do Ratio dashboard. A weekly leadership artifact that measures the gap between what leadership communicates about AI and what leadership demonstrably touches. The sibling article in this set goes deeper. The short version is that teams watch behavior, not endorsement.
The IMPACT Scorecard. A weekly six-dimension RAG board that turns the Phase 2 pilot momentum into a running cadence. The third article in the set walks the scorecard cell by cell.
The honest read on the nineteen percent.
Microsoft's Frontier zone is nineteen percent of firms. That is not a small number, and it is not a large one. It tells me two things at the same time. The pattern is reachable. The pattern is also the exception today.
If you were trying to be in the upper quartile on operating margin in 1998, you optimized for working capital and supply chain. If you were trying to be in the upper quartile in 2010, you optimized for digital experience. If you are trying to be in the upper quartile in 2027, you are optimizing for the gap between Frontier and Stalled. That is a manager-behavior gap, and it is structural.
The composition of the rest of the field matters too. Sixteen percent of firms are Stalled. Ten percent sit in Blocked Agency. Fifty percent are Emergent. The bottom and the middle together are more than three-quarters of the market. The Frontier nineteen percent is competing against a field that is, on average, paying the Coordination Tax in full and getting agent-era output back in cash form, not in the form of better margins.
The Clarity Certification is the certificate that the gap closed on your watch. It is also the rhythm that prevents the gap from reopening the next time the org chart changes.
What this means for the operator reading this.
If you are a CEO, your job in Phase 3 is to put your own behavior on the wall. The Say/Do Ratio piece in the sibling article exists because your team is already measuring you in private. You can either measure yourself in public, alongside them, or let the private measurement run unchecked. The Frontier manager data is the upstream reason your behavior is the strongest single signal the rest of the organization has access to.
If you are a CHRO, your job in Phase 3 is to wire the four behaviors plus the rewarded-reinvention behavior into the performance review template. Not the values poster. The actual instrument that determines promotions, ratings, and discretionary comp. The Frontier pattern is observable. It is also coachable. It is not, on its own, self-sustaining. Without the instrument change, the behaviors regress within two review cycles.
If you are a CFO, your job in Phase 3 is to defend the budget that protects the experiments that have not paid off yet. Twenty-six percent versus eleven percent on rewarding reinvention. That delta is not free. It is a small protected pool of capital that the rest of the organization can see being deployed. The protection is the signal. The signal is the behavior change. The behavior change is the Frontier zone.
None of these moves require a new vendor. They require a redesign of three instruments you already own. The performance review template, the operating-budget protection rules, and the leader's calendar. That is the work of Phase 3.
Source · Microsoft 2026 Work Trend Index Annual Report, May 2026. Survey of 20,000 AI users + M365 telemetry across 10 markets, fielded by Edelman Data x Intelligence, Feb 18 to Apr 20, 2026. Random forest permutation importance, 29 factors, R² 0.69. Foreword by Dr. Karim Lakhani, Harvard Business School. Frontier-manager behavior comparison table, p. 13.