Leading through leaders

Leading through leaders is the job that begins when the people who report to you lead people of their own: selecting who leads, developing their judgment, and holding them accountable for their teams' outcomes and health without taking the teams back. It is not a larger version of leading engineers. The instincts that made someone a strong team leader — being close to every person and every problem — are nearly the opposite of what the job now requires, and the previous edition of this standard, which treated the work as the "Professional band" of delegation, feedback and performance, understated how different it is. This is the competency where a manager's leverage becomes largest and their information becomes worst, at the same moment. It matters because an organisation's capacity to execute is set by the quality of its managers, and that quality is produced or squandered here. The examinations assess it from EMP-II upward; there is no team-scope form, because leading leaders requires leaders to lead.

1. Select managers on evidence of judgment about people, not on seniority, tenure or retention. The sound reasons to move someone into management are a genuine desire to lead, demonstrated judgment about people in adjacent conditions, and the team's trust. Management as a reward creates managers who do not want the job.
2. Managers develop through progressively harder real situations with coaching. Courses and shadowing supplement; they do not substitute. The instrument is the judgment debrief: walking through the hard call they face or just made, probing the reasoning, coaching the thinking rather than issuing the ruling.
3. Hold them to outcomes and health, and do not take the team back. A manager is accountable for what their team delivers and for how the team is; when either slips, the response is coaching and explicit expectations, not silently absorbing their job.
4. Your information is filtered by design; build for it. A manager's own account is the most filtered source available. Skip-levels for signal, direct observation, peer input and team-health data are how a leader of leaders knows what is true, and the skip-level is never a decision bypass.
5. Span of control is a function of context, not a number. How many managers or teams one person can lead depends on the managers' maturity, the stability of the work and the coordination each report requires. A span that works in steady state fails during change.
6. The measure is what happens without you. The system is working when decisions that used to need you are made well without you, and the managers you developed are developing managers of their own.

The literature on managing managers is thinner and more recent than the literature on managing engineers. One large internal study is graded as research below; the rest is practitioner work with clear provenance.

What good managers do research

Google's multi-year internal study of its own managers, reported in David A. Garvin, How Google Sold Its Engineers on Management (2013), which set out to test whether managers mattered and found, against the company's engineering culture, that they did, and identified a short list of behaviours that distinguished the best: coaching, empowering rather than micromanaging, expressing interest in people's wellbeing, being results-oriented, communicating, supporting career development, having a clear vision, and having enough technical skill to help. It is a single-company study, and its list is unsurprising; its value is as evidence that the behaviours are measurable and that the difference they make is large enough to show in retention and satisfaction data. It is the closest thing to research evidence this competency has.

Managing managers practice

The chapter of Camille Fournier, The Manager's Path: A Guide for Tech Leaders Navigating Growth and Change (2017) on the transition to leading managers, and its characteristic failures: skip-level meetings used badly, the temptation to keep doing the old job, and the new manager who needs coaching rather than replacement. Fournier's practical point, which this unit adopts, is that the people who report to you now need you to hold them accountable for their teams and to develop their judgment, and that both are harder than managing engineers because the feedback loop is longer and the evidence arrives filtered.

Sizing teams and spans practice

The systems view in Will Larson, An Elegant Puzzle: Systems of Engineering Management (2019) of how many people a manager can lead and how many managers a leader of managers can support, and the argument that organisational design should follow from those constraints rather than from the org chart's tidiness. The specific numbers are context-dependent and the book says so; the useful transfer is the habit of treating an overloaded manager as a design problem before treating it as a performance problem.

Task-relevant maturity, applied to management practice

Grove's concept from Andrew S. Grove, High Output Management (1983), applied to the case it fits best: a brilliant engineer newly promoted into management has low task-relevant maturity at management itself, however senior they are technically, and needs close support on exactly the things they have never done. The leader of leaders calibrates support to the manager's experience with this kind of situation — their first performance conversation, their first hire, their first reorganisation — not to their title.

Coaching managers through change practice

The practical treatment in Lara Hogan, Resilient Management (2019) of how a leader supports managers through the transitions their teams go through — growth, reorganisation, loss — using coaching questions rather than answers, and sponsorship rather than protection. Its value here is the distinction between coaching a manager and rescuing them, which is the tension the whole competency turns on.

The manager one-on-one as judgment debrief

The weekly meeting with a manager is not a project review. Its core is the hard call they are facing or just made: what they saw, what they considered, what they chose and why. The leader of leaders asks questions that expose the reasoning and offers their own only after the manager has exhausted theirs. Status travels in writing.

The first six months of a new manager

A new manager is onboarded into a new profession: an explicit conversation about what changes, a named peer they can ask the embarrassing questions, protected room to learn, and a leader who coaches them to manage outcomes rather than tasks. The new manager who is reviewing every pull request and redoing junior work is not stripped of duties at the first stumble; they are coached to define what to delegate and what good enough looks like, and protected while they adjust.

Skip-levels for signal

Regular one-on-ones with the people who report to your managers, whose purpose is unfiltered signal about team health and the issues a manager might miss, and visible accessibility. Never a channel for decisions that belong to the manager, and never relayed so identifiably that people stop talking. What is heard is verified before it is acted on, and acted on through the manager.

Reading a manager

Once a quarter, for each manager, the leader writes three lines: the team's health (attrition, engagement, what skip-levels show), the team's delivery, and how the results were produced. Read over several quarters and in context. A manager whose personal heroics mask a team that cannot operate without them is a risk, not a success.

The overloaded manager

When a manager is still doing significant individual work on top of managing and is visibly struggling, the leader helps them shed it before they burn out, and reads the situation first as a design question — is the span right, is the team the right shape — and only then as a performance one.

The inherited leadership team

A leader who inherits managers of mixed quality assesses each individually against the bar over the first months, on evidence, and acts deliberately: neither a loyalty purge nor indefinite drift.

There is no team-scope form of this competency; an Associate leads engineers, and the EMA-I outline lists PL-6 as not assessed. At organisational scope the object is the managers who report to you — their selection, development and accountability — and the characteristic failure is reaching around them to their teams. At executive scope the object becomes the leadership team as a system: complementary strengths, clear scopes, shared accountability rather than a set of individually strong rivals for your attention; the leaders of leaders beneath you, whose readiness for larger scope you must see before they ask; and the decision to replace a senior leader when coaching has not worked. Delegating a team is PL-1 Delegation, ownership, and decision rights; designing the structure the managers sit in is SV-6 Organisational design and leading change; the pattern is in How Judgment Scales.

• Promotes people into management for the right reasons and supports the transition deliberately. org
• Selects a lead for a new team on demonstrated judgment in adjacent conditions, not on who is loudest, most senior or most available. org
• Understands what a strong engineer who does not want to manage actually wants, and develops or finds another leader. org
• Runs manager one-on-ones as judgment debriefs and stretches managers with real situations. org
• Coaches the new manager who micromanages toward outcomes rather than stripping their duties. org
• Helps an overloaded manager shed individual work before they burn out, and asks whether the span is right. org
• Uses skip-levels for signal, verifies what is heard, protects the sources, and acts through the manager. org
• Reads a manager's effectiveness from sustained team health, delivery and methods, over time and in context. org
• Names a manager's avoidance — the hard conversation not had, the one-on-ones that quietly stopped — as a leadership gap, and coaches it. org
• Channels a manager's ambition, including ambition for your own role, into growth and succession-relevant scope, and names corrosive politicking directly as behaviour. org
• Failure mode — using management as a seniority prize or a retention bribe. org
• Failure mode — drafting a reluctant engineer into leadership. org
• Failure mode — running manager one-on-ones as project reviews. org
• Failure mode — developing managers by course catalogue. org
• Failure mode — reaching around a manager to their engineers because you could fix it faster. org
• Failure mode — treating every escalation as yours to solve, so that the manager's job quietly becomes yours. org
• Failure mode — taking a team back when it struggles instead of developing its manager. org
• Failure mode — tolerating a manager whose personal heroics mask a team that cannot operate without them. org
• Failure mode — treating an ambitious manager as a rival, or ignoring corrosive politicking because the manager is talented. org
• Builds a leadership team as a system — complementary strengths, clear scopes, shared accountability — rather than a set of strong individuals competing for attention. exec
• Develops leaders of leaders, and sees the signals that a leader is ready for larger scope before they ask. exec
• Holds a senior leader accountable for an organisation's outcomes and health while acting through them, and knows when coaching has not worked and replacement is the call. exec
• Gives leaders political context rather than political shelter: enough to operate, minus the noise. exec
• Assesses an inherited leadership team individually and acts deliberately; neither a purge nor drift. exec
• Failure mode — a leadership team that behaves as rivals for the executive's attention. exec
• Failure mode — "protecting" leaders into naivety. exec
• Failure mode — a senior leader who refuses to develop successors, tolerated because they are excellent. exec
• Failure mode — leadership development as a programme owned by a function rather than an executive responsibility. exec

Coaching versus rescuing. A manager who reports to you is drowning and it would be faster to do their job. Every time you do, they learn that the job is not theirs and their team learns the same. The only version of help that leaves the manager stronger is the slower one.

Accountability versus protection. A new manager needs room to learn and their team needs a manager who is meeting the bar now. The resolution is explicit expectations with a time horizon: protected room is not indefinite, and the manager knows when the protection ends.

Signal versus bypass. Skip-levels exist because you cannot see a team through its manager alone, and every skip-level is an opportunity for the team to route around the manager. The line is decisions: signal flows up, decisions flow back through the manager, and the leader who breaks that line once has weakened the manager for a year.

Consistency versus autonomy. Managers who lead their teams differently produce teams that work differently, and some of that difference is the point. The judgment is which practices must be consistent across teams — the performance bar, the promotion mechanics, the values — and which are each manager's own.

Loyalty versus the bar. The manager who was there first, who was loyal to you, who got the team through a hard year, is not meeting what the organisation now needs. Respect and candour are owed; permanent scope is not, and the failure of letting loyalty freeze the structure is paid for by everyone under it.

A senior manager with four teams has one manager, promoted eight months ago from the team they now lead, who is visibly struggling. They are still doing significant individual work — they took the hardest tickets in the last two sprints — their one-on-ones with their own team have become irregular, and two of their engineers have said in skip-levels that they do not know what is expected of them. Delivery has not yet slipped, because the manager is covering the gap personally, and the manager's own weekly account of the team is "busy but fine". The senior manager used to lead this team and knows exactly how they would run it.

The pull is to step in: attend the team's planning, have the one-on-ones the manager is missing, or quietly take back the two projects most at risk. Each would fix the quarter and end the manager. The team would learn that its manager is not really its manager, the manager would learn that struggling means losing the job by degrees, and the senior manager would have added a fifth team to their four.

The judgment is to diagnose before responding, and then to coach with explicit expectations rather than absorb. The diagnosis is that this is a new manager with low task-relevant maturity at management itself, who has retreated to the work they are good at because the work they are new to is frightening, and who has not been told what the job now requires. That is a leadership gap in the senior manager first: eight months in, the expectations of the role were never made explicit.

In the next one-on-one the senior manager names what they have observed — the tickets, the irregular one-on-ones, the two engineers who do not know what is expected — as situation, behaviour and impact, not as verdict. They state the expectation plainly: the manager's output is the team's output, and the team needs a manager more than it needs another engineer. Then they coach the thinking: what would the manager have to stop doing to run one-on-ones every week; what is the definition of good enough that would let them hand back the hard tickets; which two things does the team most need to hear about expectations. They agree three concrete changes for the next month and a date to look at the evidence together, and they say what they will do: nothing in the team directly, and a weekly debrief of the hardest call the manager faced.

The senior manager also asks the design question. Four teams for one senior manager, with one new manager among them, may be the wrong span for this year; if the new manager needs more coaching than the senior manager can give, that is a structural fact to raise upward, not a reason to reach around. What is not done is the thing the senior manager could do best: run the team.

• PL-1 Delegation, ownership, and decision rights — delegating a whole team to a manager; reaching around as the organisational form of taking work back.
• PL-2 Feedback and difficult conversations — feedback about how someone leads, gathered at one remove.
• PL-4 Performance management — reading a manager's performance, and quality-controlling how they manage performance.
• PL-5 Career development and growth paths — the leadership pipeline, readiness and succession as a system.
• CC-2 Coaching versus directing — coaching the coaches: developing judgment in someone who develops it in others.
• SV-6 Organisational design and leading change — the structure managers sit in; span of control as a design question.

1. What are sound reasons to make someone a manager, and what are the unsound ones that are more common? org
   Answer

   A genuine desire to lead, demonstrated judgment about people in adjacent conditions, and the team's trust. Seniority, tenure and retention are the common unsound ones, and management as a reward produces managers who do not want the job.
2. A manager who reports to you is drowning. What distinguishes coaching them from quietly doing their job? org
   Answer

   Who is deciding at the end of the week. Coaching leaves the decisions and the team with the manager and develops their judgment through questions and explicit expectations; doing their job takes the decisions back while leaving them the title.
3. What is a skip-level for, and what is it never for? org
   Answer

   For unfiltered signal about team health and issues a manager might miss, and for visible accessibility. Never for decisions that belong to the manager, and never relayed so identifiably that people stop talking.
4. How do managers actually get better, and what is the role of training? org
   Answer

   Through progressively harder real situations with coaching — the judgment debrief on the hard call they face or just made. Courses and shadowing supplement; they do not substitute.
5. A newly promoted manager is reviewing every pull request and redoing junior work while their team's growth stalls. What is the intervention, and what is not? org
   Answer

   Coach them to manage outcomes rather than tasks, help them define what to delegate and what good enough looks like, and protect them while they adjust. Not stripping their duties or sending them back to individual work at the first stumble.
6. What determines a healthy span of control, and why is there no magic number? org
   Answer

   The managers' maturity, the stability of the work, and how much coordination each report requires. A span that works in steady state fails during change, so the right number is a function of context, not a constant.
7. What is the difference between a leadership team and a set of strong leaders, and what does the executive do about it? exec
   Answer

   A leadership team has complementary strengths, clear scopes and shared accountability; a set of strong leaders competes for the executive's attention. The executive designs the scopes, makes the shared accountability explicit, and refuses to be the arbiter of rivalries.
8. What is the strongest signal your leadership development is working? exec
   Answer

   Decisions that used to need you are made well without you, and the leaders you developed are developing leaders of their own.

• David A. Garvin, How Google Sold Its Engineers on Management (2013) — the internal study of what distinguishes effective managers.
• Camille Fournier, The Manager's Path: A Guide for Tech Leaders Navigating Growth and Change (2017) — the transition to managing managers and its failures.
• Will Larson, An Elegant Puzzle: Systems of Engineering Management (2019) — sizing teams and spans as a design constraint.
• Andrew S. Grove, High Output Management (1983) — task-relevant maturity, applied to the new manager.
• Lara Hogan, Resilient Management (2019) — coaching managers through change; sponsorship over protection.