Why 15 Direct Reports May Break Your Team’s Productivity

When Coinbase announced its reorganization to support small, agile teams with managers overseeing up to 15 direct reports, the move seemed like a natural evolution in the age of AI. After all, smaller teams move faster, technology amplifies individual output, and flatter hierarchies promise clearer accountability. The logic is sound—until you account for the human element.

The challenge isn’t just about structuring teams; it’s about reconciling the fundamental disconnect between a manager’s schedule and a maker’s workflow. Paul Graham highlighted this tension in his 2009 essay, where he contrasted the two approaches to time management. For knowledge workers—those who build, create, or solve—long, uninterrupted blocks are essential. A single 30-minute meeting doesn’t just take 30 minutes; it disrupts the mental flow required for deep work, often costing half a day or more in lost productivity. For managers tasked with leading and contributing as player-coaches, this becomes an unsolvable math problem. The calendar can’t stretch to accommodate both roles without sacrificing either leadership or productivity.

The Unspoken Cost of Managerial Overload

At first glance, reducing 1:1 meetings from weekly to biweekly or monthly seems like a smart trade-off. The math appears to work: fewer meetings equal more time for focused work. But the real cost isn’t measured in hours; it’s measured in relationships. Even the most advanced AI tools can’t replicate the value of human connection, and no amount of automation can substitute for the trust and context built in regular, meaningful interactions.

Consider the implications of scaling 1:1s to 15 direct reports. A weekly 30-minute meeting with each direct report consumes 7.5 hours—an entire workday. When interruptions, rescheduling, and the overhead of switching between tasks are factored in, the actual time commitment balloons. Meanwhile, the quality of those interactions suffers. A manager relying on AI-generated briefs—summarizing an employee’s performance, noting a preference for direct feedback, or flagging a promotion conversation in Q3—gains efficiency at the expense of depth. Reading a dossier on an employee isn’t the same as knowing them. It’s the difference between skimming an actor’s filmography and understanding their craft.

Dunbar’s Number: The Cognitive Limit of Team Size

This is where Robin Dunbar’s research on human social relationships becomes critical. Dunbar’s studies revealed that humans maintain meaningful connections in layered social circles: about 5 people in the innermost circle, 15 in the next layer, and 50 beyond that. These aren’t arbitrary numbers; they’re cognitive load limits. The innermost circle—close friends, family, and closest colleagues—receives roughly 40% of our social and emotional energy. Relationships in the 15-person layer are real but thinner, requiring consistent investment to remain functional.

The overlap with Coinbase’s 15-direct-report model isn’t coincidental. It sits right at the edge of where humans can meaningfully maintain close working relationships. Push beyond this threshold, and managers aren’t just adding meetings—they’re asking their brains to care about more people than evolutionarily designed. The result? Superficial interactions that fail to build the trust, alignment, and psychological safety essential for high-performing teams.

Brooks’ Law and the Myth of Parallelizable Work

Frederick Brooks’ The Mythical Man-Month, published in 1975, offers another lens through which to view this challenge. Brooks’ central thesis—that adding more people to a late project can delay it further—applies beyond engineering teams. Management work, too, has sequential and non-parallelizable components. While AI can automate note-taking, summarize performance data, and draft updates, it can’t replace the relational work of leadership.

For example, a manager preparing for a 1:1 using AI-generated insights might review a summary like: "Taylor struggled in the last sprint but excels in direct communication and is targeting a promotion discussion in Q3." This information is useful, but it’s not the same as understanding Taylor’s motivations, recognizing subtle shifts in their confidence, or sensing when they’re holding back in a meeting. Relationships aren’t data points. They’re built through repeated, authentic interactions that technology can augment but not replace.

Rethinking Org Design for the Human Era

The push toward AI-driven efficiency often overlooks the fact that connection is a cost of doing business—one that can’t always be engineered away. As organizations scale, they must balance the benefits of flatter hierarchies and nimble teams with the cognitive and emotional limits of human relationships. Here’s how teams can adapt:

• Prioritize depth over breadth in management. Instead of maximizing the number of direct reports, focus on the quality of interactions. Smaller teams allow managers to invest meaningfully in each relationship.
• Leverage AI for administrative, not relational, work. Use tools to automate note-taking, draft communications, and surface patterns, but reserve 1:1s for human connection and strategic alignment.
• Acknowledge Dunbar’s limits. Recognize that teams naturally organize into layers of 5, 15, 50, and 150 individuals. Design org structures that respect these cognitive boundaries to maintain cohesion.
• Experiment with cadence, not just scale. Biweekly or monthly 1:1s may free up calendar space, but they risk eroding the foundation of trust and psychological safety that drives performance.

The future of work isn’t just about building faster, smarter teams. It’s about building human teams—where technology amplifies what’s already working and helps address what’s not. The most successful organizations won’t be those that chase the myth of perfect scalability, but those that remember the people at the heart of every line of code, every line of strategy, and every line of growth.