Becoming an Engineering Manager Is a One-Way Door Most People Don't See

The engineering manager move is sold as a promotion. It's actually a different job with a different skill set and a different market. The real cost isn't that managing is hard. It's that every month you spend out of the codebase, the market quietly marks down your value as an engineer. And after about five years, the door back to a senior IC role has swung most of the way shut.

Here's the part nobody says out loud at the offer conversation. You're not deciding whether you want to manage people. You're deciding whether you can afford to stop shipping code for the next two to five years. Those are very different questions, and most people only answer the first one.

Why is becoming an engineering manager a one-way door?

Because the skill you traded away to take the role is the same skill the market uses to price your return. When you stop writing code, your technical recency starts decaying. Not your intelligence. Your recency. And the job market for senior engineers runs on recency, not potential.

Charity Majors, who has done the full engineer-to-manager-to-engineer loop more than once, lays out the timeline plainly: come back inside two years and it's still pretty easy. Wait three to five years and it gets progressively harder. Past five, it gets significantly harder. Google's own engineering research arrives at the same structural claim from a different direction: when you move to one side of the pendulum, your skills on the other begin to deteriorate. Two independent sources, same conclusion. The door isn't a metaphor. It has a hinge and a clock.

What are you actually trading away?

Two opposed modes of work. Deep engineering needs long, protected, uninterrupted focus. Management is interrupt-driven by design: one-on-ones, standups, escalations, hiring loops, roadmap fights. You cannot do both well at the same time, which is exactly what people discover the hard way.

Nikolas Rist at LeanIX became an engineering manager on January 1, 2023. By November 15 that same year, under twelve months in, he stepped back to Staff Engineer. His verdict: being an engineering manager while also making significant contributions as a developer is almost impossible. That's the honest version. Most new managers tell themselves they'll keep a hand in the code "on the side." The calendar doesn't allow it.

The damage here is subtle because it doesn't feel like damage. You're busy. You're senior. You're in the important rooms. Meanwhile the tools moved, the patterns moved, the muscle memory faded, and nobody sent you a depreciation notice.

What is the five-year clock, exactly?

It's the gap between two things that decay at different speeds: your ability to learn engineering back, and the market's willingness to bet on you doing it.

Your learning ability barely moves in five years. You could relearn the stack in a quarter. But Majors' sharper point is that the tipping point for hireability comes long before the tipping point for learning ability. Employers get reluctant to pay engineer-level salaries to someone who hasn't shipped code recently. The market stops waiting before your brain does.

Think of a surgeon who steps out of the operating room for five years to run a hospital wing. They still understand medicine. But no hospital grants them OR privileges on the strength of their administrative track record. The profession moved, the hands got slow, and the credential expired. Software is the same. We just don't say it out loud, because the industry pretends management is the next rung on the same ladder rather than a different credential entirely.

Weak way to think about it: "Going into management might hurt my technical skills, but I can always go back."

Strong way to think about it: "Every month in one-on-ones and sprint planning is a month the market updates its recency score on my engineering ability, and that score decays faster than my actual skill. The market charges me for the absence before my brain does."

Does the ladder really get narrower with every step up?

Yes, and this is the part that compounds against you. Senior ICs sit in a liquid market with, in Majors' words, fifteen bajillion job openings. Engineering manager roles are scarcer. Director scarcer still. VP is thin air. As Majors puts it, the higher you climb, there are fewer jobs, with more competition, mostly at bigger companies. And right now the structural ground is moving under management specifically.

Signal	What it shows	Source
Companies flattening management layers	41% of employees say their org has slashed management layers	Korn Ferry Workforce 2025
Engineering leaders' read on manager headcount	More report shrinking than growing manager headcount	LeadDev Leadership Report 2025
Amazon's stated direction	Raising the IC-to-manager ratio by at least 15% by end of Q1 2025	Andy Jassy, Amazon

Shawn Zeller climbed from IC through engineering management all the way to VP Engineering, then chose to return to IC. Not a layoff. A preference: the IC seat was more comforting to him than the managerial ones. But notice what the return cost at VP level. There are very few comparable VP roles to move sideways into, so the choice was effectively binary. Keep climbing a ladder he didn't want, or step all the way back down. The higher you go, the fewer the exits, and the longer the drop.

Doesn't the pay make it worth it?

At the top of big tech, the pay argument is weaker than people assume. Stack Overflow's 2025 data puts the global median engineering manager salary at $130,000 and the US median at $200,000. Solid numbers. But the senior IC track sits right next to them. On Levels.fyi, a Google L6 Staff Engineer in the US shows a median total comp around $598K, against roughly $591K for an L6 Engineering Manager - a difference of well under ten thousand dollars. At Amazon, an L7 Principal SDE shows a median total comp around $652K, squarely in the same band as a senior engineering manager at that level. Near parity.

So at FAANG levels the financial trade is honest and stark: you accept technical skill depreciation and a harder, slower return path in exchange for a roughly lateral pay move. That's the deal, and it belongs in the same uncomfortable category as the promotion that quietly costs you money. Name it that way and a lot of people would pause.

Outside that tier, the math flips, and you have to be honest about it. At most companies the Staff and Principal IC track doesn't exist, and management genuinely pays more and carries more scope. If your company has no real IC ladder past Senior, the EM move may be the only path to growth available to you. That's a real structural constraint. Just know which situation you're actually in before you decide.

So is management a mistake?

No. That's not the argument. The argument is against sleepwalking into it.

The pendulum works for people who treat it as a deliberate two-phase career and come back before the recency window closes. Majors is the proof case: she's run the full cycle repeatedly and credits it for making her better at both. Management also compounds in ways pure-track ICs never get: hiring judgment, org design, roadmap ownership, cross-functional trust, a network. An engineer who managed for a few years and returned brings pattern recognition the lifelong IC simply doesn't have. The market doesn't reliably price that, but it's real value.

And the macro trend cuts both ways. If management layers keep getting cut, more managers flow back to IC, which over time could normalize the return and soften the market's recency penalty. Those cuts tend to spread the way layoffs copy from one company to the next rather than following any clean logic, so the flattening is unlikely to stay isolated to a few names. The one-way door may be partly a cultural artifact that's already loosening on its hinge.

There's also a real cost to staying. The role is heavy. Jellyfish's 2024 survey found 65% of respondents experienced burnout in the past year, and flagged it as particularly acute for short-staffed engineers and the leaders running large organizations. Plenty of people leave management not because they failed at it but because it quietly ground them down. Wanting out is not a character flaw. It's information.

How do you actually decide whether to become an engineering manager?

Stop asking "do I want to manage people?" first. Answer these three, in this order.

1. Can you afford two to five years without shipping? Majors' floor is real: she suggests 7+ years of shipping code before managing, and committing at least two years once you do. If you take the role, treat it as a deliberate tour with a planned exit, not a one-way escalator. Decide your return date before you start.

2. Do you want this specific job, or just the signal it sends? The title, the bump, the status at the dinner table. That's a social reward, and accepting a social reward you don't actually want is how people end up trapped in a job that's depreciating their real asset. This is the same reason a sideways move often builds more career capital than a title does: the org chart rewards the signal, but the market pays for what you can actually do. Separate the work from the signal. If you'd hate the work but love the signal, you have your answer.

3. Does your company have a real IC ladder? If Staff and Principal exist and pay competitively, management is a choice with a strong alternative. If it doesn't, management may be your only growth path, and that changes everything. Find out before the conversation, not after.

The Praxy worldview here is simple: this is your decision, framed by your priorities, not the org chart's. Agency over fatalism. The door is one-way, but you're the one choosing whether to walk through it, with your eyes open and a return date in hand. The title, or the lateral pay move, is what you're paying for with your technical recency. If you've done the math and you're ready, go.

Trying to decide whether the EM move is right for you, or how to map a deliberate path back to IC before the clock runs out? Talk it through with Praxy on WhatsApp. I'll help you frame the actual trade-off, not the one the org chart wants you to see.