Expertise · Leadership
Where judgement becomes behaviour
A decision nobody lives is only a document. Leadership is where a decision proves itself: in what a leader says on Monday, how a team answers, and what happens when the plan meets the shop floor.
Illustration · where decisions leak
Where a decision gets lost on its way to action
Start from the problem, not from a style
We do not hand out leadership types or ask anyone to become a different person. We start from a concrete problem and look at behaviour in its context: does the leader give direction, can people tell what is expected, and does anyone notice what the change costs them? Two dimensions keep turning up in research on how people judge each other: agency (direction, competence, getting things done) and communion (care, warmth, being on the side of the people). They are common property of the field, not our invention. In everyday language: where are we going, and who is looking after the people who must get us there?
Either alone fails predictably. Direction without care produces compliance that evaporates under pressure. Care without direction produces comfortable teams that do not know what to do. What the right balance looks like depends on the situation, which is why we do not offer a universal recipe.
Capable people, a sound plan, and nobody follows it
The director of a fictional twelve-person analytics team introduces a standard review step for every analysis before it leaves the team. The plan is sensible. Within a month almost nobody uses it. The replies are polite. The best analysts, the ones everybody respects, are the least compliant.
The usual diagnosis is “resistance to change”. It explains nothing, and it puts the problem inside the analysts. Four mechanisms can produce exactly this picture, and each calls for a different move:
1 · Role clarity
Mechanism: nobody is sure whether the step applies to exploratory work, or who signs off. Cue: ask three analysts who approves an analysis and you get three answers. Move: a one-page statement of decision rights. Check after a week whether the questions stop.2 · Autonomy
Mechanism: the most skilled people object to being told how, not to being asked what. Cue: objections come mainly from senior analysts and are about method, not purpose; they would accept a checklist they wrote. Move: fix the requirement, hand over the design of the how, and see whether use rises within two weeks.3 · Distrust
Mechanism: people suspect the review data will be used against them. Cue: questions about who sees the results and whether they feed appraisals; a previous “quality initiative” preceded job cuts. Move: state in writing how the data will and will not be used, protect it from appraisals for six months, and keep the promise.4 · Misaligned incentives
Mechanism: the step adds time, and recognition goes to the number of analyses delivered. Cue: compliant people deliver less and look worse in the monthly count. Move: change what is counted or the target. No speech competes with a metric.They can coexist, and the cues will not be perfectly clean. A good first test is also the cheapest: ask every analyst, individually, “what would make you use this next week?” and sort the answers. That is what a Decision Review does at larger scale.
What psychological safety means here
It does not mean a comfortable atmosphere. It means that a challenge which costs something can be voiced and answered on its merits: an analyst can tell the director “this will make our results worse”, and nothing happens to her appraisal. A reliable cue is how the last person who disagreed was treated, and everybody in the team knows the answer.
Short, observable interventions
Pick one move. Say in advance what you would see on a Tuesday morning if it worked: who does what differently. Run it for two weeks. Look again, and write down whether the prediction held. If it did not, the mechanism you chose was probably wrong, and that is useful too.
The leader’s own pattern
One behaviour, different effects in different rooms
The two dimensions of agency and communion are a way of describing what a leader does, not of sorting leaders into kinds. The same behaviour can serve in one room and corrode in another. A firm instruction is right in an emergency and does damage in a design discussion. Warm encouragement builds a new hire and, offered instead of a decision, leaves a team waiting.
It also matters that behaviour draws a reply. Pressing tends to produce either pushing back or backing off, and warmth tends to produce warmth. So a leader’s pattern cannot be read off the leader alone; it has to be read together with how the team answers. Some leaders are surprised to find that the “passive team” they complain about is the reply to their own constant correcting.
What we look for is flexibility: whether the leader can choose the behaviour the situation needs, or whether one familiar way of acting is used everywhere because it is the only one on hand. We look at specific moments (a Monday briefing, a budget argument, a missed deadline), not at character.
Power and fairness
Why people follow, and when they stop
People follow for different reasons, and the reasons matter for what works. We hold the following as working ideas, not as laws.
- Legitimacy. People are far more willing to carry out a decision when they see the person who took it as entitled to decide.
- Expertise. Deference follows demonstrated competence in the matter at hand, and ends where the competence does.
- Fairness of process. People tolerate an unwelcome decision better when they were heard, the criteria were applied the same way for everyone, and reasons were given.
- Reciprocity. Effort is returned where effort has been seen. Where it has been ignored, the return is quietly withdrawn.
The practical consequence is that a leader’s authority is partly lent by the team, and the loan is renewed daily by how decisions are taken. A leader who spends it on shortcuts finds that the next real emergency has less to draw on.
When behaviour is harmful
Observable conduct, not labels
Some leaders misuse power. We pay attention to what can be seen: intimidation, manipulation, suppressing feedback, undisclosed conflicts of interest, taking credit and passing on blame. We look at conduct in context and at the arrangements that make it easy or hard: whether there is an independent channel for concerns, whether decisions leave a record, whether the people who raise problems are protected.
We do not attach psychiatric or popular-psychology labels to named people, and we do not offer to diagnose anyone. A label ends the inquiry, is often wrong, and cannot be checked from outside. Conduct can be described, evidenced and addressed, and doing so protects the people involved, including the person accused. Where the facts amount to a legal or safety issue, the work stops and goes to the proper route.
The work
What a leadership engagement looks like
- One concrete problemNot “develop our leaders” but, for instance, “the review step is not used” or “the team stops talking when the director enters”. You get: a problem statement you can check against behaviour.
- Behaviour in its contextIndividual conversations, observation of a few real moments, and the data that already exist. You get: observations marked as seen or inferred.
- Rival mechanisms and the cue that separates themAs in the case above: at least three explanations, and for each what we would see if it were true. You get: a ranked list with the evidence.
- One small, reversible moveWith a prediction written down first. If coaching, mentoring or work on communication is the right move, it follows the diagnosis, as one of the focused ways of working. You get: a plan for two to four weeks.
- Look againCompare with the prediction and with the baseline. You get: a short note on what held and what did not.
What would count as it working
Agreed before the move is made
- The Tuesday-morning testWho does what differently, specifically, within a stated time. If nobody can say, the move is not yet a move.
- The thresholdHow much change counts, and over how long, so that a small uplift is not called success afterwards.
- Side effects to watchWhether speed, trust or the willingness to disagree suffers as a result.
- If the prediction failsWe drop that mechanism and say so. A failed prediction tells us something about the situation.
How certain are we?
What research says about leadership
- EstablishedPerceptions of people divide along agency and communion, and leader behaviour is well described along these two dimensions.
This is a long-standing finding in social and interpersonal psychology. It describes how behaviour is seen; it does not by itself say which behaviour works.
- EstablishedPerceived fairness of process, and the experience of having a voice, go with greater acceptance of decisions.
Well replicated as association. The size of the effect varies with the stakes and the setting.
- Plausible applicationLeader behaviour draws a patterned reply from the team, so a pattern has to be read from both sides.
Rests on interpersonal research, mostly outside workplaces. We check it by asking the team and observing.
- Our hypothesisMost “resistance to a sound plan” is explained by one of four things (role clarity, autonomy, distrust, incentives) that can be told apart in about a week.
We will report cases where none of the four fits.
No single leadership style for everyone
- A leadership type, score or profile. We do not offer a diagnostic instrument, and nothing on this site should be read as claiming that one has been validated.
- A standard leadership programme. Whatever we do follows from the problem.
- An assessment of anyone’s personality or mental health.
- A guarantee that a team will follow its leader.
Discuss a leadership question
Leadership is where the other areas meet. Depending on what the diagnosis shows, the next page is usually one of these.