The average US manager now oversees 12.1 people, up from 10.9 the year before and nearly 50% higher than when Gallup first measured in 2013. The median manager still leads five to six โ exactly where it has sat all along (Gallup, 2026).
Both numbers are accurate. Only one of them describes the manager sitting in your next one-to-one.
That gap between mean and median is not a statistical footnote. It is the entire structure of the "Great Flattening," and it determines whether widening span of control at your company is a defensible operating decision or a headline you are copying without the underlying data.
The Average Moved Because a Thin Band of Teams Got Huge
Gallup's own decomposition is unusually blunt about what drove the jump. The rise in average team size over the past year was largely produced by a two-percentage-point increase in teams of 25 or more employees (Gallup, 2026). A small number of very large teams pulled the mean up. The rest of the distribution barely moved.
The distribution itself is the part worth memorizing:
- 37% of managers oversee fewer than five people
- 66% manage fewer than 10
- 22% have 10 to 24 direct reports
- 13% oversee 25 or more
So two-thirds of managers lead single-digit teams. The Bureau of Labor Statistics arrives at a similar national picture from the other direction โ roughly one manager for every 11.5 employees (Bureau of Labor Statistics, 2026).
The Great Flattening is real, but it is concentrated. Large employers cutting management layers โ the megamanager story that business press has been tracking since the layoff waves (Business Insider, 2026) โ is a genuine phenomenon inside the top of the distribution. It is not a broad structural shift that has already reset what a normal team looks like.
If you are running a 50โ500 FTE company and benchmarking your span of control against "12," you are benchmarking against a number generated by organizations with a management bench, a talent-assessment function, and an HR operating model you do not have.
Widening Span of Control Is a Bet on Two Variables You Probably Haven't Measured
The useful part of Gallup's research is not the headline number. It is the finding that span of control has no universal optimum โ its effect is conditional on the manager.
In 2020, Gallup ran a meta-analysis of more than 200,000 manager-led teams to test the question directly. The result: a manager's innate tendencies shape how team size affects engagement more than team size does on its own (Gallup, 2020).
Two measurable conditions carry most of that effect.
Condition one: how much individual-contributor work the manager is still doing
97% of US managers report individual-contributor responsibilities on top of leading people. The median manager spends 40% of their time on that IC work (Gallup, 2026).
Below that 40% line, managers hold engagement above the 37% average regardless of how many people report to them. Above it, engagement is lower โ and it degrades further as the team grows. The organization has effectively stacked two jobs on one person and then added headcount to the second one.
Note what this implies for the mid-market specifically. Player-coach managers are not an edge case in a 200-FTE company; they are the default staffing model. Gallup's data shows managers who exceed the IC threshold already cluster in small teams: nearly half manage fewer than five people, about three-quarters fewer than 10. That is not a coincidence. It is the distribution finding its own equilibrium.
Widening those spans without removing IC load is not a flattening. It is a compression.
Condition two: whether the manager has the underlying talent
Gallup has spent five decades identifying five traits that predict a manager's ability to drive engagement and performance: motivation, workstyle, initiation, collaboration, and thought process (Gallup, 2020).
Measured across 3,579 managers, the pattern is sharp. Managers scoring high on those five traits show 57% engagement โ 20 percentage points above the roughly one-third baseline โ and they hold that level whether they have four reports or more than 15. Managers with moderate levels start equally high at four or fewer reports, then decline as the team grows. Managers with low levels are disengaged at any span.
The 2026 follow-up extends it to the extremes: facing 25-plus direct reports and a heavy IC load above 40%, high-talent managers still show higher engagement. Medium- and low-talent managers do not.
Gallup's own summary line is the one to take into your next org-design conversation: manager talent trumps span of control.
There is one more variable worth flagging because it is invisible on an org chart. Work location interacts with span. Growing direct reports drags on engagement across the board, but the effect is most pronounced for remote and hybrid managers. For on-site managers, increasing team size has minimal impact (Gallup, 2026). If you are widening spans and running hybrid, you are pulling two levers at once and will not be able to attribute the outcome.
Where Large Teams Actually Break
Gallup's global span-of-control meta-analysis โ 92,252 teams across 104 organizations, 26 industries, and 46 countries โ segments teams as small (4โ9), medium (10โ19), and large (20 or more) (Gallup Span of Control Meta-Analysis, 2026).
For small and medium teams, the relationship is stable and familiar: higher engagement links to higher productivity, better wellbeing, and lower turnover.
For large teams, it fractures. Engagement still predicts wellbeing. But its relationship to productivity and turnover varies by industry โ reliable in services, transportation, and healthcare; highly variable elsewhere.
Read that carefully, because it is the most operationally important sentence in the research. Past roughly 20 direct reports, the standard engagement-to-performance mechanism stops being dependable. Something else starts governing the outcome โ the nature of the work, how much coworkers can support each other, how standardized the tasks are.
That is the real risk in copying a 12-report benchmark. You are not just adding load. You are moving into a region where the management model you have been using to predict performance no longer reliably predicts it.
The AI Assumption Underneath the Whole Argument
Most current flattening proposals share an unstated premise: AI absorbs the coordination overhead โ status collection, reporting, scheduling, first-draft synthesis โ so each manager can carry more people.
Gallup names the pressure directly, noting that uncertainty about AI's effects on team size and management is adding to the urgency around defining an "ideal" span (Gallup, 2026).
Here I am extending the evidence rather than reporting it, and I want that boundary explicit. Gallup did not test AI-augmented managers at scale. What their data supports is narrower and still decisive: the binding constraints on wide spans are manager talent and IC workload, not coordination volume.
Coordination is the part AI plausibly compresses. Talent is not compressible by tooling. And IC load only falls if someone deliberately removes it โ which is a staffing decision, not a software outcome. If anything, the current pattern runs the other way: teams adopt tools, someone has to review and verify the output, and that verification work lands on the same manager, as IC work, unbudgeted.
The plainest version: AI may reduce the cost of coordinating twelve people. It does not make a low-talent manager into a high-talent one, and it does not remove the second job you already gave them.
Where This Argument Could Be Wrong
Three limits, stated before you act on any of it.
The engagement-to-performance link is correlational within these datasets. Gallup's meta-analytic base is large and consistent across industries and countries, but a wide span that lowers engagement is not thereby proven to lower output in your specific function.
The 12.1 figure is a US working-population average. Sector composition varies enormously โ retail and hospitality spans look nothing like professional services โ and a national mean is a weak benchmark for any individual company.
And the talent measurement rests on Gallup's own five-trait instrument. It is well-validated within their body of work and consistent with decades of manager research, but it is a single vendor's construct. The directional finding โ manager quality moderates the span-performance relationship โ is far more robust than any specific percentage attached to it.
None of that changes the operational conclusion. Median span is six. Talent and IC load moderate the effect of widening it. Both are measurable before you decide.
What to Do Before You Widen a Single Span
This is a two-week audit, not a program. It needs no budget.
1. Pull your actual distribution, not your average. Count direct reports per manager and look at the median and the shape. Most mid-market ops leaders discover their "average of nine" is four managers with three reports each and one with twenty-two. The outlier is where your risk already lives โ and it is usually the span you should be narrowing, not the benchmark you should be chasing.
2. Measure IC load per supervisor. One question, asked consistently: what share of your working time goes to individual-contributor tasks rather than leading people? Gallup's 40% line gives you a threshold with evidence behind it. Any manager above 40% is a poor candidate for a wider span, regardless of tenure or performance rating.
3. Assess manager talent with an instrument, not a reputation. Gallup's five traits โ motivation, workstyle, initiation, collaboration, thought process โ are one validated framework; a structured behavioral assessment against the same underlying constructs is another. What matters is that the judgment is measured rather than inherited from a promotion decision made three years ago on the basis of individual output.
4. Then widen selectively. High-talent, low-IC-load, on-site or well-structured managers can carry more, and the evidence says they will stay engaged doing it. Everyone else should not be moved on a ratio.
The internal version of this problem
If your 2026 operating plan already contains a target span-of-control ratio, it was almost certainly set from a benchmark rather than from your own manager data. That is the cheapest thing on this list to fix, and the least likely to be on anyone's agenda this quarter.
One Decision This Quarter
Take the one manager whose span you are most tempted to widen. Answer two questions with numbers, not impressions: what percentage of their week is individual-contributor work, and what evidence do you have โ beyond their last performance review โ that they have the talent to lead a larger team?
If you cannot answer both, you are not flattening the organization. You are transferring structural cost onto a single person and calling it efficiency.
The median span of control is six. That number has not moved in years, across every wave of technology that was supposed to move it. Before you decide your company is the exception, make it prove it with your own data.
