Field Guide · Organisational Behaviour

Reading Leadership

One test, six failure modes, and the sentences that give each of them away — long before the damage shows up in attrition numbers.

Start Here — The Single Test

Are the people around them getting bigger, or smaller?

Everything below is instrumentation for that one question. Every signal — how they delegate, what they explain, who they favour, how they behave when it goes wrong — is a way of measuring it earlier than eighteen months of working for them.

GreatGrow people past what those people believed they were capable of.

GoodKeep good people, and keep the current system running well.

BadShrink people — invisibly, until someone leaves and visibly flourishes elsewhere.

Leadership assessment usually happens too late. But leadership is unusually verbal: it leaks through phrasing, through what gets explained versus asserted, through who is in the room and who is spoken for. Sections below are collapsed — open what's relevant.

The Two Axes

Nearly every meaningful difference reduces to two questions: does information flow through them or stop at them, and do their people get bigger or smaller. Plot those and the archetypes separate cleanly.

The conflict-avoider is the one people misclassify. Warm, transparent, well-liked — and still growing nobody, because growth requires feedback that stings.

Great vs. Good — the baseline comparisonSix dimensions where competence and multiplication diverge

Good leaders make the current system work. Great leaders change what the system is capable of.

Dimension	Good	Great
Managing people	Manages the person and the work — real 1:1s, fair, consistent	Manages the conditions: spends their own capital removing obstacles the team can't reach
Sharing information	Shares conclusions — "here's what was decided"	Shares reasoning, rejected options and open uncertainty, so the team can generate the next decision alone
Growing people	Supports growth when asked; approves the budget, writes the reference	Engineers it — stretch assignments before readiness, and pushes people toward roles off their own team
Bad news	Handles it calmly when it arrives	Makes early bad news visibly rewarded, so it arrives from three levels down, unprompted
Being wrong	Rare, treated as a cost	Said out loud with specifics — the only thing that makes it safe for anyone else
Their absence	Things slow down	Nobody notices — and that's the deliverable

A good leader's team can execute the current plan. A great leader's team can write the next one.

How information movesThree org shapes: hoards, cascades, circulates

Information architecture is the fastest read on an organisation, and it is set almost entirely by the leader.

Dashed lines are information whose release the leader controls. Only the third pattern survives the leader taking two weeks off.

Nice, Kind, Harsh, Cruel

Two words get used interchangeably and shouldn't. Nice optimises for how the conversation feels right now. Kind optimises for the other person's interests over a longer horizon — which frequently means saying the uncomfortable thing. The same distinction runs on the other side: demanding raises what someone can do; harsh just costs them something.

Harsh and kind can contain the identical factual claim. The difference is whether it arrives with a path forward, and whether it arrives in front of an audience.

Approachability is not the same as availabilityWhy "my door is always open" predicts almost nothing

Available means you can get time with them. Approachable means it's safe to bring them something they won't like. These come apart constantly: leaders who answer Slack at midnight and are still the last to hear that a programme is failing.

The measurable version of approachability is not how often people talk to them — it's what people bring them. If the leader only ever receives finished work and good news, they are available and unapproachable, and the first genuine signal they get will be a resignation.

Available but not approachable

My door is always open. (and nobody has walked through it with a problem in a year)
I'd rather hear it from you than find out later. (said once; the last person who did was visibly punished)
Don't bring me problems, bring me solutions. (the single most reliable way to stop hearing about problems)

Genuinely approachable

What's the thing you're worried about that you haven't told me yet?
Bring me the half-formed version. I'd rather be early than polished.
Thank you for raising that — say it again in the staff meeting so it's on the record.

Approachability is measured in what arrives, not in how open the calendar is.

How It Feels — Short Term vs. Long Term

This is the instrument everyone actually uses and almost nobody uses correctly. The feeling at the end of a conversation is real data, but it is inverted for the two most consequential cases: the leader who makes you feel best in the moment is frequently the one leaving you least prepared, and the one who makes you uncomfortable is often the only person telling you something you can act on.

Judge the feeling about yourself, not the feeling about them. Liking someone after a meeting is weak evidence. Walking out with a clearer view of your own situation is strong evidence.

Vertical axis is how you feel about your own capability, not how much you like them. The nice leader and the kind leader start at opposite ends and swap places somewhere around month six.

The felt-sense tableWalking out of the room · six months in · two years on

Type	Immediately after	After a year or two
Great / kind	Slightly exposed. You said more than you meant to. Something landed that you didn't enjoy — but you know exactly what to do tomorrow.	You are visibly better than you were, in ways you can name. You still quote them. You'd take their call at any hour.
Good	Clear and unbothered. Business was done. Nothing to process afterwards.	Fine. Solid reference, no hard feelings, no transformation. You've largely stopped thinking about them.
Nice / avoidant	Warm and reassured. Genuinely pleasant. You had a nice time.	Blindsided. Feedback arrives from a promotion committee that they never gave you, and you realise years of "you're doing great" cost you the thing you wanted.
Harsh	Small, defensive, replaying it on the drive home. Possibly you also learned something true.	Cautious and self-protective. You may be more skilled and you are definitely less willing to try anything visible.
Micromanager	Oddly relieved — someone else is holding it. Also faintly insulted.	Deskilled. You've stopped forming opinions before checking. The confidence loss is the real damage and it follows you to the next job.
Gatekeeper	Special. You were told something. You're on the inside.	Isolated and unmarketable. No network outside the team, no visibility, and the specialness turns out to have been the whole compensation package.
Credit sink	Appreciated in private. They said good things, warmly, just to you.	Invisible. Your work has a reputation; you don't. You find out how much when you start interviewing.
Seagull / absentee	Free, then abruptly hammered. Whiplash rather than a steady state.	Chronically braced. You optimise for defensibility over quality, because you can't predict what will be attacked.
Favourites-player	Depends entirely which side of the line you're on — elated or vaguely diminished, with nothing concrete to point at.	Inside: dependent, and unaware how much of your success was allocation. Outside: withdrawn, still competent, no longer volunteering.

The sharpest single question to ask yourself after any meeting with a leader: do I now have a clearer picture of reality than I did an hour ago — including the parts that don't flatter me? Nice leaders leave your picture unchanged and your mood improved. Harsh leaders leave you with a worse picture of yourself and no better a picture of the situation. Great leaders leave you with an accurate one.

The lagging indicator that settles it: how you feel about them once you no longer work for them and they have nothing left to give you. Gratitude that survives the end of the power relationship is close to unfakeable.

Preference, Favouritism, and Morale

This is the failure mode with the largest morale multiplier, and the one leaders are least aware of committing. Very few people set out to play favourites. Most drift into it: they trust the person who is easiest to work with, so they give them the visible work, so that person develops faster, which retroactively justifies the trust. The loop closes quietly and looks like meritocracy from the inside.

Favouritism rarely produces complaints. It produces a slow withdrawal of unrequested effort — which shows up as "the team lacks initiative."

How favouritism actually shows upSix mechanisms, none of which involve liking someone more

Allocation of visible work. The same two names get the board presentation, the new product, the exec-facing project. Everyone else gets maintenance. This alone determines promotion outcomes two years out.
Asymmetric benefit of the doubt. When the favourite misses, it was circumstances. When someone else misses, it was judgment. Same event, different attribution — and the team tracks this with total accuracy.
Access. Some people get the pre-meeting; others get the meeting. The pre-meeting is where decisions are made.
Feedback asymmetry. The favourite gets specific, developmental feedback. Others get vague reassurance — which reads as kindness and functions as abandonment.
Interpretation rights. The favourite's read of an ambiguous situation is treated as the default account. Others have to argue for theirs.
Air time. Interruptions, whose idea gets built on, whose gets restated by someone else and then credited. Count it for one meeting; it's rarely close.

Statements that predict it

I just know I can rely on them.
Let me check with [same name] first. (regardless of whose domain it is)
They've got a lot on, so let's not add to it. (protection extended to one person only)
That's just how they are. (behaviour excused for one person, penalised in others)
I don't play favourites. (usually said by someone who has never audited it)

Statements that suggest they're actively guarding against it

Who hasn't had a high-visibility project in the last two quarters?
I've noticed I go to you first for everything. That's my habit, not your job description — let's fix the distribution.
Before we decide, I want to hear from the people who haven't spoken.

The morale mechanism. Perceived unfairness in allocation does more damage than the allocation itself. People will accept working harder than a peer; they will not accept working harder and having it invisible. The response is almost never confrontation — it's withdrawal of discretionary effort, the unrequested work that constitutes most of a team's actual value. That withdrawal is silent, arrives about a quarter before the resignation, and is usually diagnosed by the leader as a motivation problem in the individual.

Situational Leadership

The single most common misdiagnosis in this whole area: a leader is called a micromanager when they are actually applying the right amount of direction to the wrong person, or applying yesterday's correct amount to someone who has since grown. Style should be a function of the person and the task — not a fixed personality trait.

Two variables set it: competence at this specific task and confidence or commitment right now. Both move. A senior person on an unfamiliar problem is temporarily a beginner; a strong performer six months into a slog is temporarily disengaged.

The bottom-left quadrant is where most people quit or get written off — competence has stalled and belief has dropped. It is the quadrant that requires the most leadership and typically receives the least.

The four errors of mismatched styleRight instinct, wrong quadrant — and how each is misread

Error	What it looks like	What the team calls it
Directing an expert	Prescribing method to someone who knows the domain better than you	Micromanagement (and it functionally is)
Delegating to a beginner	"You own it" to someone with no map, then disappointment at the result	Being set up to fail
Supporting when they need direction	Encouragement and availability where the person needed a decision	Vague / absent
Coaching when they need clearance	Developmental conversation about a problem that is structural, not personal	Not listening

The signal that separates a situational leader from a micromanager: they can articulate why this person, on this task, right now gets this level of involvement — and they name the exit condition. Micromanagers apply the same intensity to everyone and can't tell you what would make them stop.

Situational, said out loud

You've done this twice; I'm not reviewing the third one.
I'm going to be closer to this than usual because it's your first, and I'll back off after the pilot.
You don't need my input on the how. You need me to defend the decision upward — that I'll do.
This is the part of the curve that feels bad. It's not a signal you're failing.

The Failure Modes

1 — The MicromanagerTell: decision latency when they're travelling

Reviews work at a level of detail exceeding their actual knowledge of it. The driver is rarely malice — it's anxiety, plus an inability to distinguish "different from how I'd do it" from "wrong." The cost isn't their time; it's that nobody below them builds judgment, because judgment only develops when you own an outcome.

Structural tell: everything queues behind one person. Ask anyone on the team how long a routine decision takes when the leader is away. Days means micromanagement whatever it's called.

Statements that predict it

Just loop me in on everything, I'll tell you what matters.
Send it to me before it goes anywhere.
It's faster if I just do it myself.
Why did you decide that without checking with me?

Contrast: "You own this. Name the two decisions you want me in, and I'll stay out of the rest."

How it feels: short term, oddly safe — someone else is carrying the risk. Long term, deskilling: you stop forming a view before checking, and the confidence loss travels with you to the next job.

2 — The Gatekeeper (Siloing)Tell: their reports have no direct cross-org relationships

Treats information as leverage and routes all outside contact through themselves. Silos get described as a structural problem; far more often they're a leadership choice, because someone whose value rests on being the interface has a direct incentive to keep the interface narrow.

Structural tell: if every cross-functional conversation goes through the leader, and their reports have never met their counterparts in the next org, the silo is deliberate.

Statements that predict it

That's above your pay grade.
Leadership has decided. (no name, no rationale, no ownership)
Let's keep this between us for now. (routinely, about routine things)
Don't go to them directly — come to me and I'll handle it.

Contrast: "Go talk to them directly. Tell me after if you need something unblocked."

How it feels: short term, flattering — being told things makes you feel inside. Long term, isolating and career-limiting: no network beyond the team, and the specialness turns out to have been the entire compensation.

3 — The Credit Sink / Empire BuilderTell: nobody on their team is known outside their team

Attribution runs upward and stops. Presents the team's work in the first person, represents people in senior forums rather than putting them in the room, and resists internal transfers because headcount is the visible proxy for their importance. This is the failure mode most likely to be rewarded a level up, which is why it survives.

Structural tell: listen to how they narrate past wins. Do other people appear in the story, by name?

Statements that predict it

I presented your work to the exec team — it went great.
I can't afford to lose you right now. (third year running)
You're not ready. (with no definition of ready)
What I built here…

Contrast: "You're presenting this. I'll prep you, but it's your work and your name on it."

How it feels: short term, appreciated — the praise is warm and private. Long term, invisible: your work has a reputation and you don't, which you discover the month you start interviewing.

4 — The Seagull / AbsenteeTell: late-stage reversals after the point of no return

Absent from the work until a moment of high visibility, then intervening with maximum force and minimum context, then absent again. Or simply absent — delegating not the work but the accountability. Both produce the same thing: an inability to predict what will be approved, which converts reliably into people playing it safe.

Structural tell: count decisions overturned after the point of no return. That number measures absence better than any calendar.

Statements that predict it

I was never comfortable with that approach. (retroactively, after it failed)
Why is this the first I'm hearing of it? (it was in four status reports)
Who signed off on this? (as the first question, every time)
I'll leave it with you. (as an exit, not a delegation)

Contrast: "I'm not in the detail, so I won't override you. Walk me through the reasoning and I'll flag risks you can't see from where you sit."

How it feels: short term, free — right up until the reversal. Long term, permanently braced: you start optimising for defensibility over quality because you can't predict what will get attacked.

5 — The Conflict-AvoiderTell: a known weak performer, tolerated for years

Warm, transparent, well-liked, and net-harmful. Ducking a hard conversation doesn't remove its cost — it transfers it to everyone else on the team, who absorb the missed work and learn that standards are decorative. The tolerated underperformer is a message to every strong person about what actually gets enforced.

Structural tell: ask three people on the team to name the weakest performer. If they all name the same person and that person has been there two years, you've found it.

Statements that predict it

Everyone's doing their best.
I don't want to demotivate them.
Let's give it another quarter. (fourth consecutive quarter)
It's not really my place to say. (it is exactly their place)

How it feels: short term, lovely — genuinely the nicest person to work for. Long term, blindsided: the feedback you never got surfaces in a promotion round, years late and from a stranger.

6 — The Favourites-PlayerTell: the same two names on every visible project

Covered in full above. Worth stating separately because it's the only failure mode that damages people who are performing well — the capable person outside the circle who is doing the work and watching it land on someone else's slide.

Structural tell: list the last eight high-visibility assignments and the names attached. Then list team size. If the ratio is worse than two-thirds coverage, the pattern is real regardless of intent.

How it feels: short term, nothing you can name — just a low, unprovable sense of being slightly less real in the room. Long term, withdrawal: still competent, no longer volunteering, and unable to articulate why to anyone including yourself.

The Phrasebook

The same situations in three registers. The fastest classification instrument available, because these sentences are said in the first month.

Situation	What you hear
Delegating	"Send it to me before it goes anywhere."
Delegating	"Take a first pass and we'll review it together."
Delegating	"You own it. What are the two decisions you want me in?"
A decision	"Leadership has decided."
A decision	"Here's what was decided and what it means for us."
A decision	"Here's the call, the options we rejected and why, and the part I'm not confident about."
Disagreement	"Are you on board or not?"
Disagreement	"Good point — let me think about that."
Disagreement	"I'm about to decide. Who thinks I'm wrong, and what would change my mind?"
Bad news	"Who signed off on this?"
Bad news	"Let's work out what happened and stop it repeating."
Bad news	"Thank you for flagging it early. What do you need from me?"
Weak work	"You're doing great." (nice)
Weak work	"This is amateur." (harsh — and in the group channel)
Weak work	"This isn't landing, here's specifically why, and here's what good looks like." (kind)
Assigning	"I'll give it to [same name] — I know I can rely on them."
Assigning	"Who has capacity?"
Assigning	"Who hasn't had a visible project this year? Let's start there."
Someone growing	"I can't afford to lose you right now."
Someone growing	"If there's a course you want, send it to me."
Someone growing	"There's a role elsewhere better for your trajectory than anything I have. Want an intro?"

Eight Fast Diagnostics

Who talks in meetings. Bad: the leader and one favourite. Good: everyone answers when asked. Great: people argue with each other and the leader mostly listens.
Air time, counted. One meeting, tally whose idea gets built on and whose gets restated by someone else and then credited. It's rarely close.
How you learned about the last big change. Rumour → gatekeeper. All-hands → good. Direct, early, with reasoning → great.
Decision latency without them. Days means micromanagement.
The last eight visible assignments. Count distinct names against team size.
The known weak performer. Tolerated for years means standards are fiction, and every strong person already knows it.
Late-stage reversals. Count them.
Where the alumni went. A bad leader's former reports leave and go quiet. A great leader's get promoted elsewhere and still ask them for advice.

Back to the Test

Every diagnostic above is a proxy for one measurement: whether the people around this person are getting bigger or smaller.

The two miscalibrations to guard against are symmetrical. Nice is not kind — warmth that avoids the costly sentence leaves people unprepared and calls it support. Harsh is not great — demanding leaders who leave a trail of diminished people are difficult people with a good narrative, not visionaries with high standards. Difficulty is easy to imitate, which is precisely why it gets imitated.