People management

The practical craft of managing a small team: what a manager's output is, one-on-ones, delegation, feedback, coaching, goals, performance problems, hiring, meetings, decision rights, managing up and workload, with templates and scripts for the hard conversations. Team-level dysfunction is in five dysfunctions, intent-based delegation in mission command, owning outcomes in extreme ownership, the theories behind all of this in leadership models, and decision frameworks in decision-making models.

The manager's job

Andy Grove's High Output Management (1983) is still the best single definition. Paraphrased:

Your own tickets, code and deals count only as far as they raise the team's output. A founder-engineer who ships the most code while the team stalls has low managerial output.

Managerial leverage

Grove treats every managerial activity as having a leverage: the output it produces per hour spent. Total output is the sum of activities times their leverage, so the job is to pick high-leverage activities.

Task-relevant maturity

Grove argues the right management style depends not on the person in general but on their task-relevant maturity (TRM): their experience and skill with this particular task. The idea closely parallels situational leadership (below). A senior engineer starting in a new domain has low TRM there.

The failure modes: micromanaging a high-TRM person (they leave) and abandoning a low-TRM person with "you've got this" (they flounder quietly). Monitoring is not the same as meddling: even high-TRM work needs a check-in cadence.

One-on-ones

A recurring private meeting between a manager and each direct report. Grove's view: it's the report's meeting, for their agenda, and it should be long enough to reach real issues. His point about leverage is that a short, regular investment of your time can raise the quality of their work for weeks.

ONE-ON-ONE AGENDA (shared doc, newest at top)
Date: ____
 
Their items (first, and most of the time)
- ...
 
My items
- context they need: decisions, changes, what's coming
- feedback (one piece, specific)
 
Standing questions (rotate; don't ask all every week)
- What's taking more energy than it should?
- Anything I should know before I hear it elsewhere?
 
Career / growth (monthly)
- Progress on the growth goal we agreed
 
Actions
- [ ] Owner - action - by when

Question bank

Delegation

Delegation is not binary. Management 3.0 (Jurgen Appelo, Management 3.0, 2011) defines seven levels of delegation, used in the "delegation poker" exercise and a delegation board (a grid of decision areas × levels, visible to the team).

DELEGATION BOARD (example: 5-person product team)
Decision area              | Level | Notes
Choice of library          |   7   | follow the tech radar
Sprint scope               |   4   | agree in planning
Hiring a new engineer      |   3   | team interviews, I decide
Production database schema |   5   | ask the staff engineer
Pricing changes            |   2   | founder decides, explains
Customer refunds under 500 |   6   | support lead, weekly log

How to hand over a task well:

• Define the outcome and why it matters, the constraints, and the deadline, not the method (see mission command for commander's intent).
• Name the delegation level explicitly: "You decide; tell me afterward."
• Agree the check-in points and what "done" means before work starts.
• Delegate whole problems where possible, including the interesting parts, not just the leftovers.
• When it's done differently from how you'd have done it, ask whether it's worse or just different.

Situational leadership

Paul Hersey and Ken Blanchard proposed the "life cycle theory of leadership" in 1969, renamed Situational Leadership in the 1970s. Style should match the follower's readiness for the specific task. Blanchard later produced his own version, SLII (1985, in Leadership and the One Minute Manager), with different labels, and the two versions are now sold by separate companies.

Note the two versions order readiness differently: in SLII the beginner starts enthusiastic (D1) and dips in commitment at D2. Evidence quality: widely taught, weakly supported. Empirical tests, for example Thompson and Vecchio (2009, Leadership Quarterly), found little support for its specific prescriptions. Use it as a prompt ("am I over- or under-supervising this person on this task?"), not as a science.

Feedback

SBI and SBII

The Center for Creative Leadership's Situation–Behavior–Impact model keeps feedback specific and about behavior, not character. The SBII variant adds Intent: ask what they were trying to do.

Bad: "You're rude in meetings." Good: the table above. For praise, the same structure makes it believable and repeatable: "In yesterday's incident, you posted updates every 15 minutes; support could answer customers without pinging us."

Radical Candor

Kim Scott's Radical Candor (2017) crosses two axes: care personally and challenge directly.

• Most new managers fail into ruinous empathy: they like their reports and avoid the discomfort, so the person finds out about the problem at review time or when they're let go.
• Scott's framing is about behavior in a given moment, not a personality label ("you did something obnoxiously aggressive", not "you are an obnoxious aggressor").
• Ask for criticism before you give it, and show you can take it.

Why not the feedback sandwich

Praise, criticism, praise. It's discouraged because:

• the criticism gets lost or softened, so the person remembers the praise and misses the point;
• people learn the pattern and start discounting praise as a warm-up for bad news;
• it feels manipulative to experienced recipients, which damages trust.

Instead: give praise and criticism separately and often, each specific. The evidence base for any particular feedback format is thin; what's consistently supported is that feedback should be timely, specific, about behavior and the task rather than the person.

Coaching

GROW was first published by John Whitmore in Coaching for Performance (1992); the name is credited to Max Landsberg and Graham Alexander, and the model was in use before it was named. Coaching means helping someone find their own answer; use it when they have the ability and need to build judgment. When someone lacks the knowledge, teach instead.

Coaching habits that matter more than the model:

• Ask one question at a time, then stay quiet.
• Resist giving the answer; if you must, offer it as one option among several.
• "What else?" is the most useful follow-up.
• End with a specific commitment and a date.

Goals: OKRs and SMART

OKRs

Objectives and key results came from Andy Grove at Intel in the 1970s (described in High Output Management). John Doerr learned them at Intel in 1975 and introduced them to Google in 1999; Measure What Matters (2018) is his book on them.

• Objective: qualitative, inspiring, time-bound: what you want.
• Key results: 3–5 measurable outcomes that show the objective was met: how you'll know.
• Google distinguishes committed OKRs (expected to be fully met, scored 1.0) from aspirational or stretch OKRs (expected to average around 0.7).

OBJECTIVE: New customers get value on day one
  KR1 Median time-to-first-report falls from 3 days to 4 hours
  KR2 Onboarding completion rises from 42% to 70%
  KR3 Day-7 retention of new accounts rises from 55% to 65%
  (not KRs: "build new onboarding flow", "run 5 interviews";
   those are tasks that serve the KRs)
Type: aspirational. Score at quarter end 0.0-1.0 per KR.

SMART

From George T. Doran, "There's a S.M.A.R.T. way to write management's goals and objectives" (Management Review, November 1981). His original letters: Specific, Measurable, Assignable, Realistic, Time-related. Later versions swap in Achievable, Relevant and others. SMART suits individual development goals and committed deliverables; it is a poor fit for exploratory work where the target isn't knowable yet, and Doran himself noted not every objective can be quantified.

Performance management

Most performance problems are expectation problems. Before concluding someone is underperforming, check:

PIPs done honestly

A performance improvement plan is a written, time-boxed set of expectations. Be honest about which kind you are running:

• A real improvement plan: you believe they can succeed; goals are achievable; you'll invest coaching time. Say so.
• A documented exit: you've already decided. Then a PIP is cruel theatre. A direct conversation and a fair severance are usually kinder and cleaner.

PERFORMANCE PLAN (real improvement version)
Role expectation: ______________________________
Gap (specific, with examples and dates): ________
What success looks like by [date] (measurable):
  1. ...
  2. ...
Support I will provide: weekly 30-min check-in;
  pairing with ____; training on ____
Check-ins: [dates]
If the goals are met: plan closes, noted as resolved.
If not met: next step is ____ (be explicit).
Signed/acknowledged: ____  Date: ____

Firing well

• No surprises: the person should already know from earlier conversations that their job was at risk.
• Decide first, then have a short conversation; don't negotiate or relitigate in the meeting.
• Do it early in the week, in private, with HR or a witness if needed, and with logistics ready (pay, equipment, access, what the team will be told).
• Be generous where you can: severance, a reference for what they did well, time to say goodbye if appropriate.
• Tell the team promptly and briefly, without disparaging the person.
• Employment law varies widely by country and state; get advice before any dismissal.

Hiring

What predicts job performance

Schmidt and Hunter's 1998 meta-analysis (Psychological Bulletin) put structured interviews, work samples and general mental ability (GMA) tests at the top. Sackett, Zhang, Berry and Lievens (2022, Journal of Applied Psychology) showed that earlier corrections for range restriction were overdone: most validity estimates fall by about .10–.20, and structured interviews became the top-ranked procedure.

• Values are corrected correlations with job performance. Even .42 leaves most variance unexplained: the 80% credibility interval for structured interviews runs from about .18 to .66.
• Structured means: the same job-related questions for every candidate, anchored rating scales, scoring each answer independently before discussing, and multiple interviewers.
• Unstructured "chat and gut feel" interviews are half as predictive and let bias in.

Scorecards and the A method

Geoff Smart and Randy Street's Who: The A Method for Hiring (2008) has four steps:

1. Scorecard: the role's mission, 3–8 measurable outcomes, and the competencies required.
2. Source: build a pipeline, mostly from referrals, before you need it.
3. Select: a screening interview, a chronological career interview (their "Who interview"), focused interviews on scorecard outcomes, and reference interviews.
4. Sell: persuade the candidate you want.

ROLE SCORECARD
Mission (1 sentence): why this role exists
Outcomes (measurable, 12 months):
  1. Ship X so that Y improves from A to B by [date]
  2. ...
Competencies (5-8, each defined):
  - Ownership: fixes problems outside their lane
  - ...
Interview plan: which interviewer probes which outcome
  or competency, with which questions.
Rating: 1-4 per competency, anchored examples for 1 and 4,
  scored before the debrief.

Meetings

Grove splits meetings into process-oriented (regularly scheduled: one-on-ones, staff meetings, operations reviews) and mission-oriented (called to solve a specific problem or make a decision, often once).

Written pre-reads. Amazon does no slide presentations: teams write narratively structured six-page memos, read silently at the start of the meeting in a kind of "study hall" (Bezos, 2017 shareholder letter). The memo forces complete thinking; the silent read ensures everyone has actually read it. See persuasive writing for how to write one.

DECISION MEMO (1-2 pages, sent 24h before)
Decision needed: one sentence. Decider: name. By: date.
Context: what's true now, with numbers.
Options:
  A. ... Pros / cons / cost / risk
  B. ...
  C. Do nothing: what happens
Recommendation and why.
What would change our mind.
Reversibility: one-way door or two-way door?

Meeting hygiene: no agenda, no meeting; start with the decision needed; end with decisions, owners and dates; shortest default length the calendar allows; anyone may leave a meeting they're not contributing to.

Decision rights: RACI and DACI

Most slow decisions have unclear decision rights, not hard trade-offs.

RACI suits ongoing responsibilities; DACI suits one-off decisions. The rule that matters in both: one accountable person or approver. Two approvers means nobody is.

Managing up

John Gabarro and John Kotter's "Managing Your Boss" (Harvard Business Review, 1980) argues the boss relationship is one of mutual dependence that you should manage deliberately. Their advice, paraphrased: understand your boss's goals, pressures, strengths, weaknesses and preferred working style; understand your own; and build a relationship that suits both, keeping them informed, being dependable and honest, and using their time selectively.

WEEKLY UPDATE TO YOUR MANAGER (5 minutes to read)
Headline: on track / at risk / off track, one line why
Done this week: 3 bullets, outcomes not activity
Next week: 3 bullets
Risks: what might slip, what I'm doing about it
Asks: decisions or help I need from you, by when
People: anything about the team you should know

Maker's schedule vs manager's schedule

Paul Graham's essay "Maker's Schedule, Manager's Schedule" (July 2009): managers work in one-hour slots; makers (programmers, writers) need half-day blocks at least. "A meeting commonly blows at least half a day, by breaking up a morning or afternoon." His fix was office hours clustered at the end of the day.

• As a founder-engineer you're on both schedules. Protect maker blocks deliberately (for example, mornings) and batch meetings into the rest of the day.
• Give your team the same protection: meeting-free days or blocks, async standups, and your own one-on-ones next to their existing meetings rather than in the middle of a clear afternoon.
• The more reports you have, the less maker time you'll get. A common rule of thumb: past about 5–8 reports, stop owning critical-path work and keep coding on things nobody is blocked by.

Burnout and workload

The WHO's ICD-11 describes burn-out as an occupational phenomenon (not a medical condition) resulting from chronic workplace stress that has not been successfully managed, with three dimensions: energy depletion or exhaustion; increased mental distance from, or cynicism about, one's job; and reduced professional efficacy.

Christina Maslach and Michael Leiter's work links burnout to mismatches in six areas of work life: workload, control, reward, community, fairness and values. Only the first is about hours; the others are about how the work is managed.

What a manager can do:

• Ask directly in one-on-ones: "How sustainable is your load, 1 to 10?" Act on anything under 6.
• Cut scope, not quality or sleep. Say out loud what the team will stop doing.
• Treat heroics as a process failure to fix, not a culture to celebrate.
• Model it: take your holiday, don't send messages at midnight (schedule them).
• Rotate on-call and unglamorous work fairly.
• Know the support available (employee assistance program, health cover) and point people to it; you're not a therapist.

Scripts for hard conversations

Keep them short, specific and kind. Say the headline in the first sentence.

UNDERPERFORMANCE (first direct conversation)
"I want to talk about the last two sprints. Three of your
five tickets slipped, and the payment bug went to
production without tests. That's below what the role
needs, and I'm worried about it. What's going on from
your side?"
[listen]
"Here's what I need to see over the next four weeks: ...
What support would help? Let's check in every Friday."

BEHAVIOR THAT AFFECTS THE TEAM
"In Tuesday's design review you called Sam's proposal
'junior-level thinking'. Sam didn't speak for the rest of
the meeting, and two people mentioned it to me. I need
disagreement to be about the idea, not the person. Can you
tell me how you saw it?"

SAYING NO TO A PROMOTION (for now)
"I'm not putting you forward this cycle. You're strong on
X and Y. The gap for senior is Z: leading a project across
teams end to end. Here's a project that would show it.
If you deliver it, I expect to put you forward next cycle."

LAYING SOMEONE OFF (role eliminated)
"I have difficult news. We're reducing the team and your
role is one of those affected. Today is your last working
day. This is a business decision, not a judgment of your
work. [HR name] will take you through pay, notice and
benefits. I'll gladly be a reference for you."

RECEIVING CRITICISM FROM A REPORT
"Thank you for telling me. That can't have been easy.
Can you give me an example so I understand it properly?"
[listen, don't defend]
"I'm going to think about this and try ___. Tell me
if you don't see a difference in a couple of weeks."

First 90 days as a new manager

Michael Watkins' The First 90 Days (2003) is the standard book on leadership transitions. A checklist for a first-time manager of a small team:

DAYS 1-30: LISTEN AND LEARN
[ ] One-on-one with every report: goals, frustrations,
    how they like feedback, what they'd change
[ ] Meet your manager: their expectations, how they'll
    judge success at 90 days and at a year
[ ] Meet key peers and stakeholders: what they need from
    your team, what's not working
[ ] Read the last quarter's docs, metrics, incidents
[ ] Set up weekly one-on-ones and a team meeting
[ ] Don't change processes yet, unless something's on fire
 
DAYS 31-60: CLARIFY AND DECIDE
[ ] Write a one-page team charter: purpose, priorities,
    how we work, what we won't do
[ ] Agree team goals or OKRs with your manager
[ ] Pick one or two early wins that matter to the team
[ ] Make the delegation levels explicit (delegation board)
[ ] Give your first specific feedback, positive and
    corrective
 
DAYS 61-90: IMPROVE AND BUILD
[ ] Fix the biggest process pain the team named
[ ] Address any performance issue you've confirmed
[ ] Agree a growth goal with each report
[ ] Ask the team and your manager for feedback on you
[ ] Plan the next quarter

Common new-manager mistakes

Linda Hill's research on new managers (Becoming a Manager, 1992; "Becoming the Boss", HBR 2007) found they expect their authority to come from the position and to control their reports, then discover that authority has to be earned through credibility and that the job is building the team, not managing individuals one by one.

References

• Andrew S. Grove, High Output Management (Random House, 1983; Vintage, 2015): managerial output, leverage, task-relevant maturity, one-on-ones, meetings, OKRs
• Jurgen Appelo, Management 3.0: Leading Agile Developers, Developing Agile Leaders (Addison-Wesley, 2011) and Management 3.0: Delegation Poker (opens in a new tab): the seven levels of delegation
• Paul Hersey and Kenneth H. Blanchard, "Life Cycle Theory of Leadership", Training and Development Journal 23(5), 1969: the original situational leadership model
• Ken Blanchard, Patricia Zigarmi and Drea Zigarmi, Leadership and the One Minute Manager (William Morrow, 1985): SLII
• Thompson and Vecchio (2009), Situational leadership theory: A test of three versions (opens in a new tab), The Leadership Quarterly 20(5): empirical test
• Center for Creative Leadership: Situation-Behavior-Impact and SBII (opens in a new tab): the SBI feedback model
• Kim Scott, Radical Candor (St. Martin's Press, 2017) and Radical Candor: our approach (opens in a new tab): the four quadrants
• John Whitmore, Coaching for Performance (Nicholas Brealey, 1992): the GROW model
• John Doerr, Measure What Matters (Portfolio, 2018): OKRs from Intel to Google
• George T. Doran, "There's a S.M.A.R.T. way to write management's goals and objectives", Management Review 70(11), 1981
• Schmidt and Hunter (1998), The validity and utility of selection methods in personnel psychology (opens in a new tab), Psychological Bulletin 124(2): the classic validity estimates
• Sackett, Zhang, Berry and Lievens (2022), Revisiting meta-analytic estimates of validity in personnel selection (opens in a new tab), Journal of Applied Psychology 107(11): revised, lower estimates; structured interviews top
• Sackett et al. (2023), Revisiting the design of selection systems in light of new findings (opens in a new tab), Industrial and Organizational Psychology 16(3): practical implications
• Geoff Smart and Randy Street, Who: The A Method for Hiring (Ballantine, 2008): scorecards and the four-step method
• Jeff Bezos, 2017 Letter to Shareholders (opens in a new tab): six-page memos and high standards
• Atlassian Team Playbook: DACI (opens in a new tab): decision roles
• John J. Gabarro and John P. Kotter, "Managing Your Boss", Harvard Business Review, January–February 1980
• Paul Graham, Maker's Schedule, Manager's Schedule (opens in a new tab) (2009)
• WHO: Burn-out an "occupational phenomenon" (opens in a new tab) (2019): the ICD-11 definition
• Christina Maslach and Michael P. Leiter, The Truth About Burnout (Jossey-Bass, 1997): six areas of work life
• Michael D. Watkins, The First 90 Days (Harvard Business School Press, 2003): leadership transitions
• Linda A. Hill, Becoming a Manager (Harvard Business School Press, 1992): research on first-time managers
• Camille Fournier, The Manager's Path (O'Reilly, 2017): engineering management from tech lead to CTO