The Great Flattening: Preparing Your Organization for Middle Management Reduction
Organizational Shift: Companies leveraging AI target middle management cuts, changing management to contributor ratios.
Management Challenges: Middle managers handle complex tasks AI can't automate easily, risking knowledge transfer loss.
Career Progression: Flattened hierarchies risk breaking traditional career paths, challenging employee advancement.
Support Initiatives: Investment in alternative structures and knowledge systems essential to mitigate risks of flattening.
For years, the narrative around AI and job displacement focused on factory floors and fulfillment centers. Robots would replace warehouse workers, automated systems would handle logistics, and blue-collar jobs would vanish first. That prediction is proving spectacularly wrong.
When Amazon announced in late October that it would cut roughly 14,000 corporate positions while leaving its warehouse workforce largely untouched, it signaled something executives have been quietly preparing for: AI's first major casualty isn't the factory worker. It's the middle manager.
"We're convicted that we need to be organized more leanly," CEO Andy Jassy told employees in announcing the cuts. The company aims to increase the ratio of individual contributors to managers by at least 15% by the end of the first quarter.
Keep Reading—and Keep Leading Smarter
Create a free account to finish this piece and join a community of forward-thinking leaders unlocking tools, playbooks, and insights for thriving in the age of AI.
Have an account? Log In
Translation: fewer people managing, more people doing. Amazon isn't alone in this calculation. According to Gartner, 20% of organizations will use AI to flatten their organizational structure through 2026, eliminating more than half of current middle management positions.
The speed of this shift has caught many by surprise. In 2024, middle managers made up 29% of all layoffs.
Target's new CEO Michael Fiddelke admitted the company created "too many layers and overlapping work" that slowed decisions. Walmart committed to freezing overall employment at 2.1 million for three years while incorporating AI into nearly all roles, with white-collar jobs targeted first.
This wave of cuts reveals something fundamental about how AI actually changes work. The technology doesn't just automate tasks, it eliminates the need for certain types of coordination.
When an employee needs software approval, a manager reviews risk, checks policy, and grants access. AI can run that decision tree instantly. When someone requests time off, a manager weighs team coverage, project deadlines, and company policy. AI handles the same analysis in seconds. The work middle managers do, translating strategic direction into tactical execution and coordinating resources across teams, increasingly happens through automated workflows.
The Translation Layer
Chris Williams, former VP of HR at Microsoft and leadership advisor, describes middle management's core function in terms most executives overlook.
A huge portion of what middle management is, is translating requirements from the vague to the specific. And deciding what is noise and what is not and what to bother your employees with. There’s this huge filtration process.
That filtration happens constantly. A CEO announces a strategic shift toward customer experience. The VP translates that into departmental priorities. The director breaks those into quarterly objectives. The manager turns those into weekly tasks that individual contributors can actually execute.
At each level, someone is filtering complexity, adding context, and making judgment calls about what matters and what doesn't.
Williams points to another dimension of this work that makes it hard to replace.
When you move to be that person's manager, not only are you one level removed from it, but you cannot advise. You can't skip the person who reports to you and tell the person two levels down how to do their job or judge the quality of the work. You have to work with the person who works for you to judge the quality of their work.
This matters because organizations that flatten assume senior leaders can provide direct oversight to frontline teams. They can't. The skills required shift.
"All of a sudden their domain expertise is nowhere near as valuable as many of the other skills like getting people to work together, communicating objectives across things, being clear and concise in your communication," Williams explains. "There's a whole bunch of things that are more important as a second level manager than you were as a first level manager."
Recently, I was chatting with Kate Barney, Chief People Officer at Smartly. She recalled a Harvard professor challenging the notion that entry-level positions would become obsolete. His argument was straightforward: development doesn't happen in leaps.
"You have to be one to be five," Barney explains, referencing his point that expertise requires progression through each stage. No one becomes skilled overnight. They need a pathway to get there, building capability at each level along the way.
That observation cuts to the heart of what organizations are about to lose. Middle managers don't just coordinate work. They develop people. They translate abstract strategic goals into concrete actions that entry-level employees can execute. They provide the feedback that helps someone progress from competent task completion to strategic thinking. They're the rung on the ladder between doing the work and leading the organization.
If you have a bunch of experts who are running around doing stuff, are they going to have the time and patience and the right attitude to be mentoring entry level college grads?" Barney asks. "What is that going to look like in the org design?
Organizations that eliminate middle management without answering that question are creating a succession crisis they won't recognize for years. The problem compounds because the effects are delayed. Today's entry-level employee won't be ready for senior leadership for another decade. By the time organizations realize they've gutted their leadership pipeline, the damage is done.
The Support Infrastructure That's Already Missing
Williams identifies a problem that exists even before companies start cutting these roles, which is that organizations already don't know how to develop second-level managers.
It's particularly easy to teach someone how to be a manager, sort of a 101 approach. There's a lot of literature out there. There's a lot of courses you can take," he says. "But the kinds of things that you need to be good at as a second and above level manager are things for which there's not a lot of literature out there and there's also not a lot of experience out there.
This creates a vicious cycle. Companies struggle to support these managers, so performance suffers, which makes them look expendable, which leads to cuts, which creates even less institutional knowledge about how to do the job well.
Williams describes the isolation these managers face.
"If you're a second level manager, going to your VP and saying, jeez, I don't know how to deal with this guy who works for me, it makes you look like an idiot. You can't go to your peers because they're often in competition with you. You can't go to your employees and ask them an opinion because they're the problem."
When you remove that entire layer, you end up with some important questions to answer.
- Where does the organizational problem-solving happen?
- Who filters the signal from the noise?
- Who translates strategy into execution?
The assumption is that AI will handle coordination and senior leaders will provide strategic direction. But there's a vast middle ground of judgment calls, people management, and contextual decision-making that doesn't fit neatly into either category.
The data on knowledge transfer makes this clear. When experienced employees leave, they take institutional knowledge with them. Middle managers, who typically have both technical expertise and organizational context, serve as the bridge.
They understand not just what needs to be done, but why decisions were made, which approaches failed in the past, and which relationships matter across departments. That information rarely lives in documentation. It lives in people.
The Mentorship Machine Stops Running
Traditional succession planning depends on structured progression. Someone moves from individual contributor to team lead to manager to director to VP. Each transition builds different skills. The team lead learns to delegate. The manager learns to develop others. The director learns to think strategically. Remove the middle rungs and that progression breaks.
The Korn Ferry 2025 Workforce Survey found that 41% of employees say their company has reduced management layers, and 37% say that's left them feeling directionless. That feeling of lacking direction comes from losing the people who provide context, coaching, and career guidance.
Senior leaders don't have time to mentor everyone. They're managing strategy, board relationships, and executive decisions. Junior employees need someone to translate that high-level thinking into actionable work they can learn from.
Companies that have successfully flattened recognize this. When organizations eliminate management layers and inflate the number of direct reports per remaining manager, those managers get overwhelmed. Span of control matters.
Someone who once coached five people while managing two projects can't effectively develop twelve people while juggling four strategic initiatives. The math doesn't work.
The current wave of flattening is also highly disruptive to organizational support networks that help create direct human contact between employees and leaders.
Some companies are attempting to solve this through technology. AI-powered coaching tools, automated feedback systems, and digital learning platforms promise to fill the gap. But those solutions miss what makes mentorship valuable.
A good mentor doesn't just provide information. They provide judgment. They know when to push someone outside their comfort zone and when to provide support. They recognize potential that isn't obvious in performance metrics. They advocate for someone when opportunities arise.
Algorithmic recommendations can't replace that human judgment, at least not yet. Organizations betting that AI mentorship will substitute for human relationships are running an experiment with their future leadership pipeline as the test subject.
What Gets Automated, What Gets Amplified
The organizational flattening trend rests on a specific bet that AI can handle the coordination and information-routing functions middle managers perform, freeing senior leaders to focus on strategy and empowering individual contributors to work more autonomously. In some contexts, that bet pays off. In others, it creates chaos.
Williams pushes back on the assumption that middle management work is easily systematized.
The kinds of situations that these second and third level managers deal with tend to be very tactical and situationally dependent situations," he explains. "They are deeply dependent upon what their business is and how it works and who the personalities involved are and often the technology that's involved. And so it's a really very custom problem.
That customization is precisely what makes the work difficult to automate. An AI can route a time-off request through an approval workflow. It can't decide whether to override policy because keeping a critical team member happy during a crunch period matters more than following standard procedure.
It can't read the room in a cross-functional meeting and realize that the real blocker isn't the stated technical issue but a territorial conflict between two directors. It can't mentor a promising junior employee through a career crisis that doesn't fit any training manual.
Efficiency narratives of the last year and the number of layoffs attributed to AI may lead you to be believe that the dam is breaking so to speak and middle management is already something we can successfully diminish in headcount terms.
But efficiency gains come with hidden costs. Organizations adopting AI report up to 25% reductions in middle management layers. That sounds like pure upside until you examine what happens to organizational knowledge.
When managers leave, their networks dissolve. The informal relationships that helped navigate bureaucracy, resolve conflicts, and move projects forward disappear. New employees don't know who to ask for help. Experienced employees don't have the bandwidth to mentor. Projects slow down even as the organizational chart suggests greater efficiency.
Beth Galetti, Amazon's senior vice president of people experience, described the company's cuts at the time as necessary to "reduce bureaucracy" and "remove organizational layers." The goal is to "operate like the world's largest startup."
But startups don't have Amazon's complexity. They don't need to coordinate across 1.5 million employees in hundreds of locations. Removing layers works when communication paths are short and everyone understands the mission. It breaks when coordination costs rise and tribal knowledge matters.
The National Bureau of Economic Research found that flatter organizations show different pay structures, with compensation more closely reflecting partnership models. Division managers in flat hierarchies earn less in salary and bonus than people in similar positions in companies with traditional hierarchies.
That shift creates its own problems. If management roles become less attractive financially, high performers simply don't pursue them. The people who remain in leadership positions might not be the ones best suited for the work.
Redesigning Before Being Forced
The question facing COOs isn't whether to flatten management structures. Market pressure and AI capabilities are making some flattening inevitable. The question is whether to preserve what middle managers do well while automating what they don't need to do, or to cut first and figure out the consequences later.
Williams argues that companies need to invest more in supporting these managers, not less.
One of the things I think companies should be really open and willing to do is to support these second and above level managers with coaches, advisors, consultants, whatever it is, with some resources to provide them the ability to ask questions and get honest feedback.
Companies that proactively redesign their management structures before implementing cuts create better outcomes. They:
- Identify which coordination functions AI can genuinely handle and which require human judgment
- Build systems to preserve organizational knowledge before people leave.
- Create alternative career progression paths that don't depend on traditional management hierarchies.
BCG's recent research on AI-driven organizational change emphasizes that transformation doesn't live in IT departments. It lives in HR, where work, roles, and culture are being redesigned. CHROs who lead successful transformations run what BCG calls a "two-speed agenda": stabilizing core HR functions while reimagining roles, teams, and operating models for AI-first work.
Organizations that skip this intentional redesign discover problems too late:
- Mentorship breaks down
- Knowledge transfer stops happening
- Entry-level employees don't develop strategic thinking because they never see it modeled
- High-potential talent leaves because there's no clear path to advancement.
Navigating this successfully requires recognizing that removing management layers requires adding new mechanisms to replace what they previously did, not simply devaluing the task.
Some organizations are experimenting with "digital middle management," where AI agents handle scheduling, coordination, and routine oversight while human managers focus on coaching and development. Early results suggest this can work, but only when it's designed thoughtfully.
Managers need training on how to work with AI tools. Employees need clarity on when they should escalate to humans versus rely on automated systems. Organizations need governance structures that prevent AI from making decisions it's not equipped to handle.
The Knowledge Transfer Crisis
Knowledge transfer doesn't happen automatically. It requires structured programs, documentation systems, and time for experienced employees to share what they know with less experienced colleagues.
Middle managers typically serve as knowledge brokers. They've worked across multiple teams, seen multiple strategies, and know which approaches work in practice versus theory. When they leave quickly, that knowledge leaves with them.
Companies can mitigate this through several mechanisms:
- Cross-functional rotations that expose employees to different parts of the organization build broader understanding. Someone who's worked in operations, customer service, and product development understands how decisions ripple across departments.
- Structured documentation programs capture institutional knowledge before people leave. This includes process documentation, but also decision logs that explain why certain approaches were chosen and what alternatives were considered.
- Communities of practice that connect people doing similar work across different teams create knowledge-sharing networks that don't depend on hierarchical reporting. When someone encounters a problem, they can tap into collective expertise rather than relying solely on their manager.
- Reverse mentoring programs where junior employees share technical skills with senior leaders create bidirectional knowledge flow. This becomes particularly important as organizations adopt AI tools that younger employees often understand better than their managers.
But these mechanisms require investment. Organizations that eliminate middle management to cut costs often don't fund knowledge management programs. They're betting that institutional knowledge is less valuable than quarterly savings. That bet might pay off in the short term. Long term, it creates fragility.
The Career Ladder Breaks
The traditional corporate career progression assumed a ladder with clear rungs. You started in an entry-level role, moved to a senior individual contributor position, then became a team lead, then a manager, then a senior manager, then an executive. Each step built on the previous one.
Flattening eliminates some of those rungs. The path from individual contributor to executive becomes a leap rather than a climb. Some people can make that leap. Most can't.
Inkson and Coe's research on management careers found that between 1983 and 1988, most manager job changes were upward moves. By the early 1990s, the proportion of upward moves declined markedly while sideways and downward moves increased. Job changes driven by organizational restructuring climbed from under 10% in the 1980s to 25% in 1992.
That trend has only accelerated over the last 30 years. Employees now face a labor market where promotion opportunities have shrunk, lateral moves are common, and restructuring drives frequent job changes.
Organizations that flatten without creating alternative career paths lose their best people. High performers want to grow. If there's no path to advancement internally, they'll find it externally.
Some companies address this through dual career tracks, where employees can advance as individual contributors without moving into management. Senior engineers, principal designers, and staff product managers command compensation and influence comparable to directors and VPs. This works in some functions. In others, it creates ambiguity about decision-making authority.
Other organizations focus on skill-based progression rather than role-based hierarchy. Employees advance by mastering new capabilities rather than managing more people. This requires robust skills frameworks, transparent assessment mechanisms, and compensation systems that reward capability development.
These alternative approaches can work, but they require intentional design. Organizations that eliminate management layers without creating new progression models are hoping employees will accept limited advancement opportunities. That hope rarely survives contact with a competitive labor market.
What COOs Should Do Now
The organizational flattening trend is accelerating, but COOs who act now can shape how it affects their companies rather than simply reacting to market pressure.
Audit management structure for coordination vs judgment work
Map what middle managers actually do. Which tasks involve routine coordination that AI could handle? Which require human judgment about people, culture, or strategic tradeoffs? The first category is fair game for automation. The second requires either preserving those roles or finding alternative mechanisms for that judgment.
Build knowledge transfer systems before people leave
Don't wait until announcing cuts to think about institutional knowledge. Create documentation programs, establish communities of practice, and implement shadowing systems that distribute knowledge across multiple people.
Design alternative career progression paths
If you're reducing management layers, employees need other ways to advance. Develop dual-track career systems, create skill-based progression frameworks, and ensure compensation reflects value beyond headcount managed.
Invest in remaining managers
If you're increasing span of control, managers need support. That includes coaching training, access to AI tools that handle administrative work, and realistic expectations about what they can accomplish. As Williams notes, these managers need someone they can turn to for advice without looking incompetent to their boss or vulnerable to their peers.
Create feedback mechanisms that replace middle management's signaling function
Middle managers surface problems that don't rise to executive attention but need organizational response. When you remove that layer, you need new systems to capture those signals. This might include regular employee surveys, AI-powered sentiment analysis, or structured feedback channels.
Test redesigns before implementing broadly
Run pilots in specific teams or divisions before restructuring the entire organization. Learn what works, what breaks, and what unexpected problems emerge. Flattening that works in engineering might fail in customer service.
Removing management layers can increase agility and reduce bureaucracy, but it can also eliminate the mechanisms that develop talent, preserve knowledge, and keep organizations aligned.