Are middle managers necessary?

There have been books and think pieces about flat organizations. No bosses. No managers. Just autonomous people shipping great work. The idea sells because it promises speed without friction and trust without overhead. It also ignores how real work fails. Work fails in handoffs, in unclear ownership, in culture that sounds good in all hands but falls apart on Tuesday afternoon. When that happens, the missing layer is not a tool. It is a manager who behaves like a leader.

The trope of the useless middle manager survives because many managers were hired or promoted into a tracking function. They track tasks. They track hours. They track people. Tools already do that better. Jira and Asana do not coach under pressure. They do not set standards when the product brief is fuzzy. They do not absorb a team’s anxiety after a failed release and redirect it into better execution next sprint. That is a leadership job. The role needs a rebuild, not a eulogy.

The pandemic made that clear. Teams learned they could stay productive without someone standing over their shoulder. Leaders read that signal as evidence that managers are optional. What actually happened was a transfer of micromanagement risk from the manager to the tool. People kept producing because the mission was clear and the feedback loop was tight. When those conditions faded, performance drifted. The fall guy became the layer that was never equipped to lead in the first place.

So the question do we need middle managers is a distraction. The real question is whether the system that turns specialists into people leaders is credible. Often it is not. High performers are promoted for craft excellence and told to manage humans as a reward. They get no training, a tighter budget, a reshuffled team, and a dashboard full of vanity metrics. Burnout is not a surprise. Compliance leadership follows. Teams feel policed rather than led. The myth of the useless middle manager confirms itself.

The fix is not a memo about empowerment. The fix is a manager operating system that makes leadership visible in daily work. If a manager cannot point to their operating system in two minutes, they do not have one. They have preferences. Preferences do not scale. Systems do. Here is what that system needs to include if you want managers to be force multipliers rather than expense lines.

Start with accountability architecture, not task lists. The team needs three visible lines that never blur. The first line is outcomes, expressed as user value delivered and quality standards met. The second line is ownership, expressed as a single name for each outcome with peers who can challenge but not override. The third line is cadence, expressed as the meeting rhythm and feedback windows where decisions get made. When those three lines are clear, velocity increases without extra meetings. When they are not clear, velocity becomes theater.

Next, build the coaching loop into the calendar. A manager who begins their day by opening the project board is doing control, not leadership. A manager who begins by scanning people load, unblocking dependencies, and shaping scope is doing the job. A simple rule helps. Every two week cycle, give each direct report thirty minutes. Spend the first ten on the person’s work. Spend the second ten on the team as a system. Spend the last ten on your own performance as their manager. Ask what to start, stop, and continue. Document one change and show it in the next cycle. That small loop pays back faster than any quarterly engagement survey.

Then set real spans of control. Six to eight direct reports is a workable range for a manager who is expected to coach, set standards, and protect focus time. Anything past that and you are buying task tracking in disguise. If the budget or headcount model forces a wider span, remove scope. Take ceremonies off the manager’s plate. Make ops or a rotating IC host the standup. Use writing to reduce the meeting surface. The goal is to free the manager to do the two things only they can do. Decide what good looks like. Grow the people who can deliver it reliably.

Do not tie career progression only to managing people. The fastest way to break a team is to make management the only path to status and pay. Create master tracks for deep craft. Let a principal engineer or principal designer earn as much as a people manager. The organization then promotes managers who want the work, not just the title. The culture learns that influence can be technical and that leadership is not an accident of hierarchy.

Manager less stories make the rounds for a reason. A flat structure can work in small, senior heavy teams that share context, values, and stakes. It fails in scale, in onboarding, in cross functional tradeoffs, and in risk management. It also fails in representation. When priorities float toward whatever the majority loves building, anything unglamorous is starved. Security, platform reliability, documentation, and diversity work become after hours projects. A strong manager solves that by making important work visible and rewarded.

If you insist on testing a flatter model, treat it like an experiment with guardrails. Define the decision rights you are removing. Define the failure reports you will read if it backfires. Define the sunset clause. Leadership is not opposed to autonomy. Leadership designs for autonomy to survive pressure. Without that design, flat becomes fragile.

Metrics need a redesign too. Many teams celebrate sprint points burned or tickets closed. Those are movement metrics. Movement without value is noise. Replace them with three numbers. The first is repeat value created per user segment per cycle, which tells you if the team is delivering outcomes that users return for. The second is cycle predictability, the share of commitments delivered inside the agreed window, which tells you whether the team can plan with credibility. The third is manager leverage, the ratio between team output and the manager’s time spent in status meetings versus coaching. If the ratio tilts toward status, the manager is trapped in reporting. If it tilts toward coaching and standards, you will feel it in retention and in product quality.

Hiring rules matter. Never hire ahead of manager readiness. If a function plans to add two more ICs, first test whether the manager can enable current ICs to operate without constant escalation. Run a two week simulation where you remove the manager from tactical calls and watch whether the system holds. If it collapses, you do not have a headcount problem. You have a design problem. Fix the rituals and the ownership map. Then hire.

The daily rituals need explicit defaults. One useful default is written first, spoken second. Force critical decisions to start as short memos with clear options, risk notes, and a call. The meeting then becomes a decision forum, not a verbal backlog grooming session. Another default is office hours over drive by Slacks. Set a window where the manager is available for ad hoc unblocks. Outside that window, route through the team’s owner so that triage happens closer to the work.

Culture is what your people do when you are not in the room. If your culture depends on the manager being in every room, you do not have culture. You have a bottleneck. Teach managers how to model standards publicly and how to enforce them quietly. Praise in the open by naming the behavior and its impact. Correct in private by naming the gap and one change to try next cycle. Keep it simple. Repetition beats intensity.

What does this look like in a week. Monday begins with a written brief that defines outcomes and risks. The manager runs a ten minute acceptance gate on scope and a five minute owner check. Tuesday through Thursday, the manager spends ninety minutes a day on coaching and unblocking, not status review. Friday is a forty five minute retro that is psychological safety in practice. Two wins that map to outcomes, one experiment for the next cycle, one debt item that will be paid or intentionally deferred. The manager closes the week by sending a two paragraph note to stakeholders that references outcomes, not effort. Consistent, boring, and powerful.

Founders often ask when to add the first layer of managers. The answer is not a headcount number. The answer is when coordination tax starts erasing product learning. If engineers and designers spend more time aligning than shipping, you are late. If the roadmap turns into a graveyard of half built experiments, you are late. Add managers then, but make the job a system. Give them authority over standards and coaching. Keep you and the exec team out of day to day prioritization except when strategy truly shifts.

Flattening is attractive when you want to cut cost without admitting that you are also cutting capability. Upgrading managers is harder because it asks you to define leadership as a craft with its own standards. Do it anyway. Build the promotion bar around coaching impact and system clarity. Build compensation that rewards managers when their teams become less dependent on them. Make yourself able to disappear for two weeks without the system stalling. If you cannot, the risk is not your managers. The risk is your design.

So yes, we need managers, but not as human project trackers. We need managers who set standards, coach the system, and protect focus. We need managers who can say no to scope that violates the cadence and yes to experiments that sharpen the product. We need managers who turn team accountability into a habit rather than a meeting. The myth of the useless middle manager fades when the job is built correctly. The work becomes calmer. The delivery becomes more predictable. The culture becomes less dependent on heroics.

The next time someone argues that managers are obsolete, ask them to run without leaders through one hard quarter and show you the outcomes. Speed matters. Autonomy matters. Clarity matters more. Redesign the role. Hold it to a standard. Then hire and grow managers who lead like operators. That is how teams scale without breaking.