The Manager's New Brief

2.1

Ch. 2 — The Player-Coach: Managing Consultants and the Three Modes

Part I — The mandate

💡 Core idea

A consulting manager is a player-coach: you run the project, you develop people, and you remain an expert who still does individual work. The job is not to escape contribution. It is to choose, for each task, one of three modes — Own it, Scope it, or Release it — and to move people from the second toward the third over time.

The tension is structural, not a personal failing. You are accountable for the engagement, for the people on it, and for a few topics where the client and the firm still want you. Individual contribution is expected. The failure is not that you still do work. It is doing the wrong work, in the wrong mode, for too long. This chapter applies Chapters 1, 5, and 6 to that reality. It also prepares Part III: moving someone from “I scope, you draft” to “you own it” is coaching, not just assignment.

The triple role — and its tensions

Three roles share one calendar.

Project. Scope, quality, timeline, client confidence, the next steering committee.

People. Staffing, coaching, feedback, and whether the team can run the next similar piece without you (Chapters 9 and 10).

Expertise. A short list of topics where you still have a genuine edge: a method, a domain, a type of client conversation. That edge is a firm asset. It is also a trap if every related task becomes yours by habit.

Delivery pulls you into Own it (mode 1) because you are faster; development needs Scope it and Release it (modes 2 and 3), which look slower this week. Expertise needs protected time (Chapter 3), or firefighting eats it. If you only play, the bench stays thin. If you only coach and never contribute, you lose the judgment the client is buying, and you cannot set a standard you no longer practice. The aim is a deliberate mix, reviewed weekly — not a personality type.

The three modes

Mode 1 — Own it. You own the task end to end. Use it when risk is high, client exposure is immediate, the deadline leaves no teaching time, or your expertise edge is the work. Own it is legitimate. It is a problem when it is the default.

Mode 2 — Scope it. You start: framing, constraints, definition of done — then someone else produces the deliverable, with a named review. This is the main development mode, and the one most often skipped. Wes Kao names the usual excuse: it would take too long to explain. Preparing the handoff is what makes delegation pay off. A thin “can you take this?” is not Scope it. It is Release it without the conditions for Release it.

Mode 3 — Release it. The other person owns the task, including the framing. You remain accountable for the project outcome, not for doing their thinking. You may still review at agreed gates, especially where the client will see the work. Bjorg: both sides must feel trusted, even when they will not do it exactly your way. Jason Cohen is the test: you are building a team that can do this work without you, not emptying your plate for a week.

An AI assistant is not a fourth mode. It is a target you can assign work to, or an amplifier of Own it (you produce more yourself). Who owns the framing, the quality bar, and the client-facing call is still Own it, Scope it, or Release it, with a human accountable (Chapters 12 and 14). The risk of getting that wrong is not theoretical. In a preregistered experiment with 758 BCG consultants — people in the same player-coach world this chapter is written for — Dell’Acqua and colleagues found that GPT-4 helped on tasks inside the AI model’s frontier (12.2% more tasks, 25.1% faster) and hurt on a task outside it: consultants using the AI were 19% less likely to produce correct solutions. Treat the AI assistant as a junior you still have to scope and review, not as a fourth mode that owns the call.

How to pick a mode, per task

Decide in writing, not in the corridor. Use these six criteria together:

When criteria conflict, name the conflict. “Own it for Friday’s steering committee; the next one like it is Scope it, with you” is a decision. “I’ll just do it” is not.

These criteria are not only local taste. Yukl and Fu (1999) found that managers delegated more to competent subordinates who shared their goals, had worked longer for them, were themselves supervisors, and had a better working relationship (LMX). Managers said they valued developing people, but were reluctant to give up control over important or sensitive decisions, or to assign major tasks to inexperienced people. That is the tension this chapter is built on: the fade from Scope it to Release it has to be earned, and hoarding important work is a documented habit, not a private vice.

Wrong-mode signals

Hoarding in Own it (mode 1). Your week is production, 1:1s slip, and you are still the only person who can talk to the client on your topics. Faster this week; more indispensable next month. Aviv Ben-Yosef applies: you may have taught the firm to expect that you will absorb the work.

Abdicating under “Release it.” You toss a vague request, skip the package and the review, rewrite the draft at midnight, and conclude that delegation does not work. That was not Release it. That was an absentee Own it.

Scope it that never ends. You still write the outline and keep the judgment, year after year, with the same person: a comfortable pairing, not a development path.

Own it disguised as review. You “review” by rewriting. The other person has not learned the standard. You have done the task twice. The 37signals Manager Playbook puts the same test in one question: “If you go back and look at your report’s projects, how much red pen is there? … how much of you or another teammate is represented in the final work product?”

What a good Scope it package contains

Write it down. A conversation that exists only in your head will be completed, badly, by the other person’s guesses. A short note plus a working session is enough.

Include at least: the question and who it is for; outcome and definition of done (format, quality bar, a thin example if the format is new); constraints (time, access, confidentiality, what must not be promised); inputs already in hand; decisions already made and open questions they may or may not close alone; out of scope; review gates; escalation (what must come back immediately).

Then stay out of production until the agreed gate, unless they escalate. Hovering is Own it with a spectator.

Moving people from Scope it to Release it

This is a planned fade, not a sudden vote of confidence. Over similar tasks, shrink what you own: first you write the question and the outline; later they propose the outline and you correct it; later they own the framing and you review only the client-facing risk. Keep a review until quality is stable, then widen the types of work, not only the size.

Say the move in the 1:1 (Chapter 9): “On this type of work, you are in Release it. I will review the client section, not the structure.” If the standard slips, step back to Scope it on the next piece — without turning the slip into a story about character (Chapter 1).

The point of the fade is capacity, not a quieter inbox. The 37signals Manager Playbook states the cost of always directing: “If you’re telling people what to do instead of teaching them how to think and problem solve, you’re forced to act in that authority capacity every time your report needs help.” The same playbook’s Multiplier Managers chapter adds the operational test: “If every decision flows up, you become a bottleneck.” Scope it and Release it are how you avoid that.

That fade is also the start of a harder career pivot. After twenty years in consulting, Maurizio (@themgmtconsult) calls “transitioning from a doer to a “leader of leaders”” the hardest move: you are “responsible for more than ever, yet you have the least amount of direct visibility into the work.” Mode 2 and mode 3 are how a player-coach begins that pivot without pretending the visibility problem has vanished. Chapter 17 returns to the craft of it.

Some teams are trying a tighter shape around the same problem. Chintan Turakhia reports that Coinbase began running “pirate and architect” pods in May 2026: “1 pirate, 1 architect, and a part-time designer.” The pirate ships and stays close to the customer; the architect “takes the pirate’s prototype, and drives its scale-out through spec for agents.” This is one practitioner’s account, not a general law of team size. It still needs an explicit mode, an owner, and a human who remains accountable. Chapter 12 returns to the workflow.

Protect time for your expertise

If the firm needs your expert contribution, it must appear on the calendar as Own it work you chose, not as leftover nights. Block it. Put it in the weekly priorities memo (Chapter 3). The grid should show a small, deliberate share of Own it where your edge is real — and a shrinking share on everything else. Do not spend that time rewriting work a clearer scoping package would have made reviewable.

Instructing someone through a serious performance lag is part of the job. It is not a reason to abandon your own contribution. The 37signals Manager Playbook is explicit: managers must instruct and correct, but “They are not, however, responsible for doing so at the expense of their individual contributor work.” If coaching underperformance is crowding out the Own it work you owe the firm, escalate — to People Ops, HR, or your own manager — rather than dropping the standard or disappearing into the rescue. You remain accountable for the call; you do not have to carry the whole process alone.

Mike Fisher borrows aviation “currency” for the same problem: “Proficiency means you can do the job yourself, and do it well. Currency means you stay close enough to the job to judge it.” You do not need to remain the best doer. AI can lower the cost of that closeness — he treats it as a cheap simulator — but it also tempts what Shopify, as Fisher reports it, calls “slop grenades”: generated work passed along without being read. A leader’s weekend prototype dropped on a team with rank attached is the same failure. “Use the simulator to stay qualified, then hand the aircraft back to the crew.” Currency is not a license to fly the mission, and it is not a hiding place from leading. Chapter 17 returns to the craft of it.

Tools

🤖 AI-era tip: An AI assistant can amplify Own it, or help you scope — it cannot own the mode

Use an approved AI assistant to help draft a scoping package, pressure-test the definition of done, or list open questions (Chapter 5). Use it, carefully, to check a deliverable against the package: what is missing, vague, or unsupported. Do not use it to do all the production yourself faster and call that leverage — that is Own it at higher volume, and it makes hoarding easier. Do not treat an AI assistant as Release it: Camille Fournier still holds — a system that cannot be held responsible cannot own the call. Do not send client work you have not judged: passing through generated text as if it were your analysis is the “meat proxy” failure (Niklas Gruhn; Chapter 14). A human owns the framing, the review, and what the client sees. Tobi Lütke, in “Learning on the Shop floor,” names the learning risk in that handoff: “The risk is not that AI does the work. The risk is that AI does the work and we never learn from it.” If the work happens only in a private window, “the only person who learns anything is the person at the keyboard.” Chapter 13 returns to this.

Takeaways

Evidence

Related chapters. Ch. 1 — What Management Is For (and What It Isn’t) · Ch. 3 — Your First Operating System: Time, Attention, and Trade-offs · Ch. 5 — Clarity Before Velocity: Goals, Priorities, and Scope · Ch. 6 — Decision-Making When Information Is Cheap and Judgment Is Scarce · Ch. 9 — Feedback, Coaching, and the 1:1 · Ch. 10 — Performance, Growth, and Difficult Conversations · Ch. 12 — Leading Hybrid Human + AI Workflows · Ch. 15 — Measuring What Matters When Productivity Is Easy to Fake