1.11 Engineering management

Overview and motivation

Engineering management is the practice of getting good outcomes through a team of engineers rather than through your own keyboard. It is a different job from being a senior engineer, not a promotion for being a good one. The best individual contributor on a team is often measured by what they personally ship. A manager is measured by what the whole team ships, and by whether those people grow, stay, and do their best work. That shift, from doing to multiplying, is the single hardest transition in an engineering career, and most new managers stumble on it because nobody told them the job actually changed.

This chapter is about the daily craft of managing engineers: running one-on-ones, giving feedback, coaching, delegating, handling performance, protecting team health, and managing up and sideways. It is deliberately separate from the career-ladder mechanics in chapter 1.3, which covers levels, competencies, and calibration as a system. Here the focus is on what a manager actually does with the people in front of them. Get this right and a mediocre team becomes a strong one. Get it wrong and strong people quietly leave, taking their knowledge and their goodwill with them.

For large teams, management quality compounds. A poor manager does not just underperform; they degrade five to eight careers at once, and their attrition and rework ripple outward for years. Enterprises live or die on whether they can develop new managers faster than they promote people into the role. Government organisations manage under civil-service rules that constrain hiring and firing, which makes the softer levers, coaching, clarity, and mission, matter even more. Wherever you sit, the manager is the unit of leverage. Treat the role as a skilled discipline you can learn, not an honour you were handed.

Key principles

A manager’s output is the team’s output, not their own. You are a force multiplier.
Management is a distinct job that must be learned, not a reward for strong individual work.
Regular one-on-ones are the backbone of the relationship; protect them.
Feedback should be frequent, specific, and kind, never saved up for review season.
Delegate the work and the ownership, then match your support to the person and the task.
Underperformance handled late is cruelty disguised as kindness; be clear and humane, early.
Psychological safety is the manager’s responsibility to create and defend (chapter 1.1).
Trust and context scale; control does not.

Recommendations

Make the transition from doing to multiplying explicit

The most common reason new managers fail is that they keep doing their old job and treat management as overhead squeezed into the gaps. They stay on the critical path of the code, review everything themselves, and jump in to fix problems personally because it feels productive and familiar. Meanwhile the actual work of management, the one-on-ones, the feedback, the unblocking, goes undone, and the team stalls waiting on them. Name this trap out loud with every new manager: their job is no longer to be the best engineer in the room, it is to make the room better. Help them list what they must stop doing personally, then hold them to it. If they are still the top committer three months in, that is a warning sign, not a badge.

This does not mean abandoning technical judgement. In engineering, credibility comes from understanding the work, and a manager who cannot follow a design discussion loses the team’s respect. Stay close enough to the technology to ask sharp questions and spot risk. Just stop being the person who has to type the answer. Your leverage now comes from decisions, context, and people, measured in team outcomes, which is exactly the productivity framing in chapter 1.10.

Run one-on-ones as the core loop

The recurring one-on-one is the heartbeat of management. Schedule a regular, protected slot with each direct report, weekly or biweekly, and hold it sacred. Cancel it only in genuine emergencies, and reschedule rather than skip, because cancelling sends the message that the person does not matter. This is their meeting, not your status update. Let them set much of the agenda: blockers, worries, career questions, feedback for you, half-formed ideas they are not ready to raise in public. Your job is mostly to listen, ask questions, and remove obstacles. A good rule is that you should be talking less than half the time.

Use one-on-ones to build the trust that makes everything else possible. When someone believes you have their interests at heart, they will tell you about the failing project before it is a disaster, admit they are burning out before they quit, and take your hard feedback as help rather than attack. That trust is built in the ordinary weeks, not in the crisis. Keep light notes on what matters to each person, their goals, their commitments, the feedback you gave, so you can follow through. Following through is what turns a nice chat into a relationship people rely on.

Give feedback and coach continuously

Feedback should be a steady flow, not an annual event. Praise specifically and in the moment, so people know exactly what to repeat. Deliver constructive feedback promptly and privately, focused on behaviour and impact rather than character: “In the design review you talked over Priya twice and we lost her point” lands better and changes more than “you need to be more collaborative.” Kim Scott’s framing of caring personally while challenging directly captures the balance: withholding hard feedback to be nice is not kindness, it is a failure to help someone grow. The manager who never gives difficult feedback is not protecting their people; they are abandoning them to blind spots.

Separate coaching from telling. Telling gives the answer; coaching helps the person find it, which builds capability that outlasts the specific problem. When someone brings you a question they could reason through, resist the urge to solve it for them. Ask what they have considered, what they would do if you were not there, what the risks are. This is slower today and far faster over a year, because you are growing an engineer who no longer needs you for that class of problem. Reserve direct instruction for emergencies and for things the person cannot yet know.

Delegate with situational leadership

Delegation is how a manager scales, and it fails in two opposite ways: dumping work without support, and hoarding work out of anxiety. The fix is to match your involvement to the person and the task, an idea captured by situational leadership. For someone new to a task, be more directive: set clear expectations, check in often, teach. As their competence and confidence grow on that task, shift toward coaching, then supporting, then fully delegating with light oversight. The same person may need close guidance on an unfamiliar system and total autonomy on one they know cold. Read the task, not just the title.

Delegate outcomes and ownership, not just chores. When you hand off a problem, hand off the authority and the context to solve it, then let the person choose the how, even if their approach differs from yours. Define what success looks like and the real constraints, then get out of the way. Expect a different path and some mistakes; that is the tuition for growing capable people. The opposite, micromanagement, signals distrust, crushes initiative, and guarantees nothing scales beyond your personal attention. If you cannot take a vacation without the team seizing up, you have delegated tasks but not ownership.

Manage performance early, clearly, and humanely

Performance management is about the strong as much as the struggling. Stretching your best performers so they stay engaged and grow is just as much the job as fixing what is broken. For high performers, the risks are boredom and being taken for granted. Give them harder problems, more scope, visibility, and sponsorship, and connect their growth to the career framework in chapter 1.3. Do not let a quiet, reliable star go a year without a meaningful conversation about where they are headed. The people who need the least day-to-day management often get the least attention, and then they leave.

When someone is underperforming, address it early and directly. The unkindest thing a manager can do is let poor performance slide for months, then blindside the person at review time. By then the frustration has curdled, the team has noticed and lost trust, and the person never got a fair chance to improve. Name the gap plainly as soon as you see it, get specific about what “good” looks like, and separate a skill gap (coachable) from a will gap (a different conversation) from a fit gap (maybe the wrong role). Offer real support and a clear, time-bound picture of the change needed. If it does not improve, act decisively, with dignity, honesty, and respect. Handled well, even an exit can leave the person’s self-respect intact and the team’s trust in your fairness stronger.

Protect psychological safety and team health

Your team’s ability to do great work rests on whether people feel safe to speak up, take risks, admit mistakes, and disagree without fear, the psychological safety that chapter 1.1 treats in depth. The manager sets this climate more than anyone. How you react the first time someone brings you bad news teaches everyone whether honesty is safe here. Respond to a mistake with blameless curiosity about the system that allowed it, and you get more truth and faster fixes. Respond with anger or blame, and you get silence, hidden problems, and eventual disaster.

Watch team health as deliberately as you watch delivery. Burnout, festering conflict, and quiet disengagement are leading indicators of attrition and quality problems, and they are your job to catch. Notice who has gone quiet, whose hours have crept up, where tension is building. Normalise taking real time off by taking it yourself and not rewarding martyrdom. A sustainable pace is not a perk; it is how you keep a team performing over years rather than burning it out in a heroic quarter. Team health is a first-class result you own, right alongside the roadmap.

Manage up, sideways, and across distance

Managing your own team is only part of the job. You also represent your team upward and outward. Managing up means keeping your own manager informed with no surprises, escalating the right things at the right time, and translating leadership’s priorities into context your team can act on. Done well, this shields your engineers from organizational churn and wins them resources and air cover. Managing sideways means building real relationships with peer managers and partner teams, so cross-team work runs on trust rather than tickets. Much of a senior manager’s impact happens in these lateral relationships, and much of it feeds the decision-making and governance practices in chapter 1.5.

Distance raises the stakes on all of it. Managing remote and distributed teams (chapter 1.9) demands more intentional communication, because you lose the hallway cues that tell you someone is struggling. Write things down, default to transparent and asynchronous updates, and create deliberate moments for connection rather than assuming they will happen. As you become a manager of managers, your leverage shifts again: you are now developing other leaders, and skip-level conversations, meeting with your reports’ reports, become your window into what is really happening beneath the summaries you are handed.

Trade-offs: pros and cons

Decision	Pros	Cons
Promote strong ICs into management	Technical credibility; internal growth path	Losing a top engineer; new managers may struggle without training
Manager stays hands-on technically	Credibility, better judgement, spots risk early	Risk of doing instead of multiplying; bottleneck
High autonomy and delegation	Scales, grows people, builds ownership	Short-term mistakes; slower for the individual task
Wide span of control	Forces delegation; fewer management layers	Thin attention per report; coaching suffers
Fast, decisive action on underperformance	Protects team trust and morale; fair to the person	Emotionally hard; risk of acting too soon or unfairly

The central tension is between control and leverage. Everything that feels productive in the moment, reviewing the code yourself, making the call, jumping on the fire, trades long-term leverage for short-term control. Everything that builds leverage, delegating ownership, coaching instead of telling, letting people learn by making recoverable mistakes, feels slower and riskier today. New managers over-index on control because it is familiar and immediately gratifying. The craft is learning to tolerate the discomfort of leverage: to let someone struggle productively, to ship a solution that is not exactly yours, to measure yourself by the team’s results rather than your own visible output.

Questions to discuss with your team

Are we selecting and developing managers deliberately, or just promoting our best engineers and hoping? The most common path into management is a reward for strong individual work, which quietly assumes the two jobs use the same skills. They do not, and the Peter principle names the result: people rise to a level where they are no longer effective. Bring evidence: how many of your managers wanted the role versus fell into it, how much training they got before they had reports, and how their teams’ attrition and engagement compare. Consider making a senior individual-contributor track a genuinely equal alternative (chapter 1.3), so people manage because they want to and are suited to it, not because it was the only way up. The answer should change how you promote, how you train, and how you let people step back out of management without shame.
How do we actually handle underperformance, and how long does it take us to act? Every team has a story about a low performer everyone tolerated for a year while resentment built and strong people quietly updated their resumes. The competing pressure is real: acting too fast is unfair, conflict is uncomfortable, and in some settings the process is genuinely hard. Look honestly at your own history: how long from “we know there is a problem” to a clear, documented conversation, and how often does the person say they were blindsided. In government and other rule-bound environments, firing may be slow or constrained, which makes early, honest, well-documented coaching even more essential rather than less. The goal is not to be quick to fire; it is to be quick to be clear, so people get a real chance and the team sees that standards are fair and enforced.
What is our span of control, and can our managers actually manage at that ratio? Span of control, the number of direct reports per manager, silently determines how much coaching each person gets. Too narrow and you grow expensive layers of hierarchy and tempt managers to micromanage for lack of enough to do. Too wide and one-on-ones get skipped, feedback dries up, and managers survive by managing only the loudest problems. Bring the real numbers: reports per manager across the organisation, which managers are stretched past ten or twelve, and whether their teams show the strain in engagement or attrition. The right ratio depends on team seniority, work complexity, and how much the manager also carries individually, so the discussion should end with a defensible target and a plan for the outliers.
What protected time do our managers actually have to manage, and what are we implicitly asking them to drop? Many organisations promote someone into management and then keep loading them with individual delivery, on-call, and project work, so the one-on-ones, feedback, and coaching happen only in the leftover minutes that never arrive. The competing pressure is that a manager who does no technical work loses credibility and a manager who does too much becomes a bottleneck, so the honest question is how the week is really spent, not the title on the org chart. Bring calendars: hours per week each manager spends in one-on-ones, in coaching, and in personal delivery, and how many one-on-ones were skipped last quarter. In a large enterprise or a government unit where headcount is fixed and process is heavy, be explicit that management time is real work with a real cost, and decide deliberately what gets cut to make room, because a manager with no slack manages only the loudest emergency in front of them.
How do we grow and retain our strongest engineers so they do not quietly leave, and who is accountable for that? The people who need the least day-to-day management often get the least attention, and a reliable senior engineer can go a year without a real conversation about where they are headed, right up until they resign with an offer in hand. The tension is that stretching a strong performer competes with the daily pull of the struggling ones and the roadmap, so growth conversations slip unless someone owns them. Bring the evidence: who your strongest people are, when each last had a career conversation, what harder scope or sponsorship they were offered, and how their compensation and level compare to the market and to the framework in chapter 1.3. For an enterprise this is a portfolio risk, since the departure of a few deep experts can stall whole systems, and in government, where pay is capped by civil-service scales, name the non-monetary levers, scope, mission, and mastery, that you can actually pull before someone with irreplaceable context walks out the door.
When something goes wrong, do people bring it to us early, and what does our first reaction teach them? The manager sets the climate for honesty more than anyone, and how you react the first time someone brings you bad news teaches the whole team whether truth is safe here, so the real measure of psychological safety is whether problems surface while they are still small. The competing consideration is that accountability still matters, and the discussion should separate blameless curiosity about a failed system from tolerating genuine negligence, because safety is not the absence of standards. Bring concrete signals: how incidents get discussed, whether post-incident reviews hunt for causes or culprits, how long bad news typically takes to reach you, and whether anyone has been punished for raising a risk. In a regulated or public-sector setting where failures can become headlines and formal inquiries, a culture where people hide problems until they explode is an operational and reputational hazard, so treat the manager’s response to bad news as a controllable process worth measuring, not a matter of personality.

Sector lens

Startup. Your first manager is usually a strong engineer promoted under pressure, with no training and a full delivery load, so the risk is that management becomes an afterthought squeezed between tickets. Keep it lightweight but real: protect a weekly one-on-one with each person, name out loud that the job is now the team’s output, and hand one meaningful system fully to someone else so the founder-engineer stops being the single point of failure. Do not build heavy process; build the two or three habits that keep people from leaving while you are too busy to notice.

Small business. You have no dedicated management layer or HR function, so managers wear the role part-time on top of hands-on work, and there is no budget for a formal manager-development programme. Lean on cheap, high-leverage practices rather than tooling: regular one-on-ones, feedback given in the moment, and clear expectations you can enforce without a bureaucracy. Decide deliberately whether to grow a manager internally or hire someone who has done it before, since a wrong hire in a small team is felt immediately across everyone.

Enterprise. Management quality varies wildly across many teams, and that variance, not any single bad manager, is the real problem to govern. Standardise the practices that scale: new-manager training before anyone gets reports, a defensible span-of-control target, skip-level meetings, and engagement and retention data used to coach rather than blame. Treat manager development as infrastructure with a tracked payback, study what your best managers actually do, and make a senior individual-contributor track a genuinely equal path so people manage by choice and aptitude, not for the only available raise.

Government. Civil-service rules make hiring slow and dismissal genuinely hard, so managers cannot lead with fast firing or large raises and must rely on the levers they do control: crisp expectations, frequent coaching, careful documentation, and mission. Underperformance demands early, honest, well-documented conversations both to give the person a fair chance and to satisfy procedural fairness and any subsequent review. Emphasise the public value of the work as a motivator that pay cannot match, and record management decisions transparently enough to withstand audit and accountability to the public.

Examples

Startup. A twenty-person startup promotes its strongest engineer to lead a team of five, and for two months delivery slows because she keeps taking the hardest tickets herself and reviewing every line. Her own manager names the trap directly: her job is now the team’s output, not her commits. They set up weekly one-on-ones, agree on three things she will stop doing personally, and hand one of her signature systems fully to a mid-level engineer to own. Velocity recovers within a month, the mid-level engineer levels up fast, and she discovers she is energised by growing people. The lightweight intervention, honest naming plus a few concrete habits, is enough at this size.

Enterprise. A large software company grows headcount faster than it grows managers, and quality of management becomes wildly uneven: some teams thrive while others churn through people. It responds with a real manager-development programme, new-manager training before anyone gets reports, a manager community of practice, and skip-level meetings so senior leaders hear directly from the front line. It also studies what its best managers actually do, echoing Google’s Project Oxygen findings, and builds those behaviours into expectations and feedback. Manager quality, measured through team engagement and retention surveys, becomes a tracked outcome, and the worst-scoring teams get coaching rather than blame.

Government. A government engineering unit manages under strict civil-service rules where hiring is slow and dismissal is genuinely hard. Its managers cannot lead with the usual private-sector levers of fast firing and large raises, so they lean on the ones they do control: crisp expectations, frequent coaching, careful documentation, and above all mission. Engineers are reminded that their work delivers benefits to millions of citizens, and that meaning becomes a powerful motivator that pay cannot replace. When performance problems arise, managers address them early and document thoroughly, both to give the person a fair chance and to satisfy the procedural fairness the system requires.

Business case: motivations, ROI, and TCO

The return on good management shows up first in retention. People do not leave companies so much as they leave managers, and a regretted departure costs a large multiple of salary once you count recruiting, onboarding, and the months of lost productivity while a replacement ramps, plus the institutional knowledge that walks out the door (chapter 1.8). A manager who runs real one-on-ones, gives honest feedback, and grows their people directly attacks the top avoidable causes of attrition: feeling stuck, feeling unseen, and working for someone they do not trust. Multiply that across every team and management quality becomes one of the highest-leverage investments an engineering organisation can make.

The second return is throughput. A manager who delegates ownership and coaches unlocks the whole team’s capacity, while one who hoards decisions caps the team at their personal bandwidth. The costs of poor management are just as concrete but easier to ignore because they are diffuse: silent attrition, burnout-driven quality problems, decisions that stall waiting on one overloaded person, and the compounding damage of a bad manager degrading several careers at once. The investment is comparatively cheap: training, coaching, and protected time for managers to actually manage. To make the case to leadership, tie it to numbers they already watch, regretted attrition, engagement scores, internal mobility, and delivery predictability, and frame manager development as infrastructure with a clear payback rather than a soft nicety.

Anti-patterns and pitfalls

The player-coach who stays on the critical path: does the work themselves and lets the team stall.
Skipping or perpetually rescheduling one-on-ones: signals people do not matter and kills trust.
Saving feedback for review season: robs people of the chance to improve while it still counts.
Micromanagement: reviewing every detail, which crushes ownership and caps the team at your bandwidth.
Avoiding hard conversations: tolerating underperformance until resentment poisons the team.
Managing everyone identically: same oversight for a novice and an expert, so both are mishandled.
Ignoring the strong performers: assuming the quiet stars are fine until they resign.
Promoting into management as a reward: creating reluctant managers who wanted a raise, not a team.
Blaming individuals for systemic failures: teaching everyone to hide mistakes.
Managing up only: polishing the report to leadership while the team quietly struggles.

Maturity model

Level 1, Initiate: Management is ad hoc and reactive. Managers are promoted for individual skill with no training, one-on-ones are irregular or absent, feedback comes only at review time, and underperformance festers. Team health is invisible until people quit, and nobody treats management as a job with its own craft.
Level 2, Develop: Basic practices exist but vary by person. Some managers hold regular one-on-ones and run review cycles, yet quality is uneven and personality-driven, and neighbouring teams do it differently. Feedback, delegation, and situational leadership depend on the individual manager, and performance issues are handled inconsistently across the organisation.
Level 3, Standardise: Management is a recognised discipline with documented expectations enforced org-wide. New managers are trained before they get reports, one-on-ones and continuous feedback are the norm, delegation and situational leadership are taught, and performance is managed early and humanely against a shared framework. Skip-levels, manager-of-manager practices, and a defensible span-of-control target apply consistently across teams.
Level 4, Manage: Management is measured and controlled with data against baselines. Engagement, regretted attrition, internal mobility, one-on-one cadence, and team-health signals are tracked per manager and per team, compared to targets, and reviewed on a fixed cadence. Weak-scoring teams trigger coaching rather than blame, span-of-control outliers are corrected on evidence, and the cost of poor management and the payback of manager development are quantified for leadership.
Level 5, Orchestrate: Management is continuously improved and integrated across the organisation. The behaviours of the best managers are studied and spread, managers develop other managers, people move between individual-contributor and management tracks without stigma, and the organisation develops leaders faster than it promotes into the role. Practices adapt as the business, the workforce, and the operating model shift, with management treated as strategic infrastructure.

Ideas for discussion

Should we measure and reward managers primarily on team outcomes rather than any personal output, and how would we do it fairly?
What is the right amount of technical hands-on work for a manager, and does it change with team size and seniority?
How do we tell the difference between productive struggle we should let a report work through and a person who is genuinely stuck and needs us to step in?
What is our honest track record on acting on underperformance, and what stops us from being clear sooner?
How do we keep managers accountable for psychological safety and team health, not just delivery?
Can someone step back from management to an individual-contributor role here without it being seen as failure?

Key takeaways

Management is a distinct job. Its output is the team’s output; you succeed by multiplying, not by doing.
The core loop is regular one-on-ones plus frequent, specific, kind feedback and real coaching. Protect it.
Delegate ownership and match your support to the person and the task; micromanagement caps the team at your bandwidth.
Handle performance early: stretch your strong people, and address underperformance clearly, humanely, and soon.
You own psychological safety and team health, and you manage up and sideways, not just down.
Develop managers deliberately; do not just promote strong engineers and hope.

References and further reading

Camille Fournier, The Manager’s Path
Andrew S. Grove, High Output Management
Michael Lopp, Managing Humans
Julie Zhuo, The Making of a Manager
Kim Scott, Radical Candour
Lara Hogan, Resilient Management
Ken Blanchard, Patricia Zigarmi, and Drea Zigarmi, Leadership and the One Minute Manager (situational leadership)
Marcus Buckingham and Curt Coffman, First, Break All the Rules
L. David Marquet, Turn the Ship Around!
Google re:Work and the Project Oxygen research on effective managers