According to Gallup's 2026 State of the Global Workplace report, the average number of direct reports per manager rose to 12.1 in 2025, up from 10.9 in 2024—an 11% increase in a single year. The jump was driven largely by a two-percentage-point spike in teams of 25 or more direct reports, reflecting systematic middle management elimination across industries. Average spans have grown nearly 50% since 2013, when Gallup first began tracking the metric. In the same period, manager engagement fell nine percentage points—to 22%, its lowest level since 2014.
Here's what's actually happening:
When companies eliminate management layers, the cost savings are immediate and visible on the P&L. The costs are delayed and distributed across dozens of smaller performance signals that don't aggregate neatly into a single line item. Each removed manager's workload gets redistributed upward to surviving managers who are already stretched—more reports, more administrative tasks, no reduction in accountability for outcomes. Research from HR analytics firm Knoetic found that for every five additional direct reports per manager, eNPS scores decline by 2%. At spans above 20, average goal completion rates drop to 60%, versus 79% at lower spans. That 19-point gap is sitting invisibly across your project pipeline right now if any of your managers are in the overloaded tier.
Why it matters for you:
- Wide spans are a performance drag with a specific dollar value: Companies where managers oversee fewer than 6 direct reports have 7% higher eNPS scores than companies with 15 or more. When spans exceed 20, goal completion falls to 60% versus 79% at healthy spans. On a team of 10 people with $80k average salaries, a 19% productivity gap equals roughly $152,000 in lost annual output—about the loaded cost of hiring one manager. The de-layering savings often invert within 18 months once you model the productivity drag against the headcount savings.
- Coaching disappears above 12 reports: One-on-ones get cancelled first, then shortened, then perfunctory. Career development conversations disappear entirely. Gallup estimates that 70% of the variance in team engagement traces to manager behavior—and at wide spans, the behaviors that drive engagement most reliably (regular feedback, development conversations, visibility into individual work) become physically impossible to sustain. You aren't funding a manager at 15 reports. You're funding an administrator who processes status updates and approves PTO.
- The math almost always favors adding the layer back: A mid-level manager costs $120,000–$160,000 in fully loaded compensation. If that manager oversees 10 people at $80,000 average salary, even a 5% productivity improvement from better management pays back the hire within 12 months. Model it with your actual numbers. Most organizations that run the calculation find the de-layering economics inverted against them over a year ago—they just haven't updated the model.
Source: Gallup, State of the Global Workplace 2026 (April 2026); Knoetic, Span of Control Benchmarks
Watch this:
The data shows a two-point spike in teams of 25 or more direct reports—the most extreme tier. That's not normal stretch; that's structural failure dressed up as organizational efficiency. If you have managers at 20-plus reports, you're running a system that will generate attrition from both the manager and the team within the next 6–12 months. Watch 90-day voluntary turnover rates within those oversized teams as your earliest leading indicator. The exits will cluster there first.
The contrarian play:
Most leadership teams respond to span-of-control data with an org design initiative that takes six months and produces a deck no one acts on. The faster move is to empower your most stretched managers to explicitly triage. A manager with 14 reports can't give everyone meaningful attention—but she can identify the 4 people on critical work or at risk of leaving and give them something real, while routing the rest into group formats and async check-ins. Deliberate triage is much better than evenly distributed neglect. It won't fix the structural problem, but it buys time to fix the structure.