A Week Nobody Planned
Haute Lumière · The Manager's Week · Chapter One
Chapter One
A manager's week is mostly written by other people, and the first gain is in reading it back.
A Week Nobody Planned
Haute Lumière · The Manager's Week · Chapter One
It is twenty to seven on a Friday, and the director of a regional operations division is the last person on her floor. She has opened her calendar to write the weekly note to her own manager, and she is scrolling backwards through the week to remember what happened in it. Monday has eleven entries. Tuesday has nine, two of them overlapping. Wednesday was meant to begin with a clear morning for the network plan, the one document her chief executive has asked her for twice, and the morning is now a supplier call, a recruitment panel and something titled quick sync that lasted an hour. The plan is open in another window. It is at page two, where it was last Friday.
There is a mug beside the keyboard with an inch of cold tea in it, and the cleaner's trolley is somewhere down the corridor. Her shoulders are up near her ears, and they have been there, she realises, since about Tuesday lunchtime. She is good at her job. Everyone in the week's forty-one meetings would say so. And she cannot, looking at the screen, find the hour in which she did the thing she is paid most to do.
She is a composite, drawn from the way these evenings are described in the research below and in the house's own practice, and no firm is meant. Anyone who has run a team of any size has had her Friday.
Ammanuel Santa Anna's canon has a sentence for what her shoulders are doing, and it comes from a book about the body's relationship with time rather than a management text, which is part of why it lands:
Linear time, untempered by somatic awareness, produces a particular kind of suffering that is so pervasive in modern life that we have almost stopped recognizing it as suffering. We simply call it "being busy."
This chapter makes one claim. A manager's week is mostly written by other people, and the gain in it is not found by working longer or by a better system for lists. It is found by reading the week back: seeing who asked for each hour, and how many of those hours held work that only she could have done. The calendar is the first place a leader's capacity is spent, and because it is spent before anyone has decided how, it is also the cheapest place to begin.
The modern study of what managers do begins with a doctoral student who decided to watch. In the late 1960s Henry Mintzberg followed five chief executives, one week each, and recorded everything: every letter and memo, every call, every meeting, who started it, how long it lasted and what it was about. He published the analysis in Management Science in 1971 and the book, The Nature of Managerial Work, in 1973.2
What he found contradicted the picture of the manager that the textbooks of the time carried, the planner at a clear desk who organises, coordinates and controls. The managers he watched worked, as his 1971 abstract puts it, "at an unrelenting pace," and their work was "characterized by variety, discontinuity and brevity." Activities were short and constantly interrupted; later summaries of the book give the median time spent on any one issue as about nine minutes. The managers preferred, he wrote, issues "that are current, specific, and ad hoc," presented in conversation rather than on paper. And the conclusion he drew from it was blunt for a journal devoted to the science of management: "there is virtually no science in managerial work."
He also proposed an account of why the work looked like that, and it is kinder to managers than the word fragmented suggests. The managers he watched, his abstract says, performed ten roles in three groups: interpersonal, as figurehead, liaison and leader; informational, as "the nerve center" of the organisation's information system; and decisional, at the heart of how resources were allocated and disturbances handled. A nerve centre is interrupted because that is what it is for. The brevity was not a failure of discipline. It was what a position looks like when every part of an organisation routes its questions through one person.
That reading matters, because it moves the problem from the manager's character to the shape of the position. A director whose week is broken into nine-minute pieces may be disorganised. She may also be the only point at which several streams of work meet, and the pieces are the streams arriving.
A word on the strength of this evidence. Five people, each for one week, is a small and a particular sample: chief executives of American organisations in the late 1960s, observed by one researcher. Mintzberg knew it, and leaned on the earlier diary studies for support, among them Sune Carlson's study of Swedish managing directors in 1951 and Rosemary Stewart's diaries of British managers in the 1960s. A review of the evidence by Colin Hales in 1986 found the broad pattern held across many later studies, while warning that much of it described what managers did rather than what made them effective.3 And in 2006 Stefan Tengblad repeated Mintzberg's method with a new set of top executives and found some of it had changed.4 The workload was much larger, more of the contact was with subordinates in groups, and, most interesting of all, "fragmentation of time" was "not as prevalent in the new study." The pattern Mintzberg saw is common. It is not a law of nature.
That last finding is the one the director at twenty to seven should hear first. If fragmentation were the essence of the job, nothing could be done about her Wednesday. Tengblad's executives suggest it is at least partly a habit of a place and a period, which means it can be read, and changed.
Santa Anna's canon has a line on why the habit is so hard to see from inside. The body, it says, has its own sense of pace, and "It knows the difference between productive urgency and anxious rushing." The difficulty is that a week of forty-one meetings produces both, and from the inside they feel the same.
A thousand diaries
Haute Lumière · The Manager's Week · Chapter One
For forty years after Mintzberg, studies of managerial time remained mostly small and observational, because watching an executive for a week is expensive and few will allow it. In 2020 four economists, Oriana Bandiera, Andrea Prat, Stephen Hansen and Raffaella Sadun, published a study that changed the scale.5 They built a method for collecting the week of a chief executive through daily calls with an assistant, hour by hour, recording each activity's length, who was in it and what it was for. They applied it to 1,114 chief executives in six countries.
The diaries sorted, with the help of a machine-learning algorithm, into two broad types. One type the authors called "leaders": chief executives whose weeks held more multifunction, high-level meetings, with people from several parts of the business in the room. The other they called "managers": chief executives whose weeks held more individual meetings with the core functions, one area at a time. The firms that hired the first type performed better. It took about three years after a new chief executive arrived for the difference to show. And the authors' structural estimates suggested something more careful than leaders are better. The productivity differences, they wrote, were due "to mismatches rather than to leaders being better for all firms." Some firms needed one kind of week and got the other.
It is worth being precise about what this shows. It is an association between how a chief executive spends her time and how her firm performs, in a large and carefully collected sample, with the authors' modelling to separate cause from selection. It does not show that any particular calendar will lift any particular firm. But it does establish, at scale, the thing Mintzberg could only suggest: that where a leader's hours go is not noise around the real work. It is a measurable choice, it differs between people in the same role, and it is tied to results.
Read the director's Wednesday against it. The network plan is the one piece of work in her week that needs the whole division in view at once, the very kind of cross-functional thinking the study's "leaders" did in their meetings. Her week is full of the other kind, one function at a time, one supplier, one vacancy, one quick sync. None of those meetings was a mistake. Most were somebody else's urgent item. The question the study raises is whether, added together, they are the week her division needs.
Santa Anna's canon describes one reason weeks fill this way, and it is not the one leaders expect. Writing about how organisations hold their resources, it describes gatekeeping, the habit of controlling access, as a source of comfort rather than strategy: "The person who controls the budget, the calendar, the information flow, the approval process, experiences each act of gatekeeping as a micro-dose of safety." A manager whose approval is needed for a great deal is, by that arrangement, a manager whose week belongs to everyone who needs it. The canon is clear that nobody chooses this out of vanity. "This is not malice. It is survival behavior wearing a business suit."
If the cost of a fragmented week fell only on the manager, it would be her problem and her diary's. It does not. In the late 1990s the organisational scholar Leslie Perlow spent nine months with a team of software engineers and published what she found under a title that has passed into ordinary use: The Time Famine.6 The engineers felt they had too much to do and not enough time to do it. They had difficulty getting their own work done, the abstract reports, "because they were constantly interrupted by others." A crisis mentality and a reward system built on individual heroics kept the interruptions going, since the engineer who dropped everything to rescue a colleague was the one who was praised.
Then Perlow and the team changed how the group used its time. The abstract records the result plainly: "Altering the way software engineers used their time at work, however, enhanced their collective productivity." The study is one team in one firm, observed closely, and its strength is in the depth of the observation rather than in any figure that could be carried to another firm. What it shows is the mechanism. Time at work is not owned by each person separately. It is a shared resource, and each interruption moves a cost from the person who interrupts to the person interrupted.
A manager sits at the centre of exactly that exchange, in both directions. Her own week is broken into pieces by the people who need her. And each of those pieces, seen from the other side, is somebody's work waiting for an answer. The recruitment panel was waiting for her availability. The supplier call had been moved twice. The quick sync existed because a question from above had reached two of her team, and they could not settle it without her. The Margin, another of the house's books, reads [the attention every team spends and nobody budgets](https://hautelumiere.life/editions/rsd-master-rsd-in-acedemia-final-manuscriptt/page/), and why fewer windows, kept well, serve a team better than many, kept badly. A manager's calendar is where that budget is set, usually without anyone noticing that a budget exists. The canon's advice to a creative team could have been written for Perlow's engineers, and it treats the whole group's time as one thing to be defended: "Protect collective air fiercely. Block unstructured time in the group calendar and defend it against the constant pressure to fill it with meetings."
There is a second cost, quieter than the first. The more of her week a manager spends answering, the less of it she spends on the decisions only she can take, and those decisions wait. A network plan at page two for a fortnight is not only her lost Wednesday. It is three teams who cannot plan their spring because the plan they are waiting for has not been written. How many people a single week of hers is asked to hold, and what happens when the number grows past what a week can carry, is the subject of Chapter 2, and it is a question of arithmetic before it is a question of character.
The director has a habit that most managers share. When she reviews a week, she asks what she did. It is a reasonable question, and it produces a list. A more useful question is who wrote each hour.
Mintzberg recorded exactly this for his five executives: for every contact, who initiated it. The pattern he found has become one of the best-known findings in the field, though the precise proportions vary from study to study and are best not repeated without the book open. The general shape is that a large share of a senior manager's contacts were started by other people. What that suggests, and it is this book's reading rather than a finding of the study, is that control over a manager's week is less a matter of having fewer interruptions than of having decided in advance which ones to take.
One author of her week deserves a sentence of his own, because managers rarely count him. The director's own manager wrote four of Wednesday's hours without meaning to: a request for figures by noon, which became an hour of her team's time and half an hour of hers; a forwarded email marked thoughts?; and the quick sync, which began as his question to two of her people. None of it was unreasonable. All of it was invisible to him, because from where he sat each request took a minute. A week is written from above as well as from below, and the hours written from above are the hardest to decline, which is why a diary that records them is a gift to both people.
That distinction changes how a week reads. An hour asked for by a customer whose contract is at risk, an hour asked for by a new hire who is stuck, and an hour asked for by a colleague who wants a decision she could take herself all look the same in a calendar. They are not the same. The first is the job. The second is the job too, and probably the best hour of the week. The third is a question that belongs somewhere else, and answering it teaches the colleague to bring the next one, which is the subject of Chapter 4.
The canon's book on creative work has a line that could be pinned above any manager's calendar, and it is not sentimental about it: "We live in a culture that has declared war on spaciousness." The same chapter is practical about what can be done without pretending a leader can withdraw from her organisation: "The practical challenge is real: most of us cannot simply clear our calendars. But even small pockets of temporal air make a difference." It is a modest claim, which is why it is credible. Nobody is asking the director to cancel the week. The question is whether two of its forty-one blocks can be moved to hold the work only she can do.
It helps to say what that work usually is, because managers often answer the question too narrowly. It is rarely a task. It is the thinking that needs the whole of her area in view, of the kind the study's "leaders" did in their meetings; the conversations that only her position can hold, with her own manager, with a peer whose team is the other half of a problem, with a person on her team who is ready for something larger; and the decisions that cannot be taken by anyone below her because they change what the people below her are asked to do. Most of that work does not appear in a calendar as an appointment, because nobody else asked for it. That is exactly why the calendar loses it.
The week read back
Haute Lumière · The Manager's Week · Chapter One
Picture the same director the following Monday, having decided to do one thing differently. She does not buy a system. She keeps a diary.
For one ordinary week she writes down, at the end of each block of the day, what the last hour or so actually held. Not what the calendar said; what happened. Beside each entry she makes three small marks. Who asked for it: herself, her team, her own manager, a peer, someone outside. Could anyone else have done it, and who. And did it move the network plan, or the two or three other things she has said in writing this quarter are hers. It takes a few minutes a day, done at the end of each block rather than reconstructed on Friday, because a week remembered on Friday is the week she wishes she had had.
On the following Friday she reads it back. The figures are her own and would not be anyone else's, but the shape is common. Most of the hours were asked for by other people. A good number could have been done by someone on her team, often someone who would have learned from it. Two standing meetings, held every week for a year, had not changed a decision since the summer. And the hours that moved the plan were few, and scattered, a twenty-minute gap here and there, never long enough to think in.
What she does next is small, and the order matters more than the size. She tells her team what she found, briefly, and that she is going to keep two mornings a week for the plan and for the work only her role can do, and when she will be reachable instead. She asks one of her managers to take the supplier call from now on, with authority to decide, because the diary showed that her presence added nothing the manager could not. She proposes ending one of the two standing meetings and changing the other to a fortnightly one with a written note between. And she leaves the recruitment panel exactly where it is, because the diary showed it was some of the most important time in the week: choosing who joins is work only she can sign.
Chapter 7 carries the diary page, the marks, the reading-back and the order of the first changes, and Chapter 6 places them at the start of a first ninety days. What can be said here is where the practice most often goes wrong. A manager who keeps the diary for a day instead of a week records her best day. A manager who protects her mornings without telling her team when she is reachable has simply moved the interruptions to the afternoon and added a grievance to them. And a manager who reads the week back looking for hours to cut, rather than for whose hours they were, will cut the conversations that were the job.
There is a gain here, and it is worth naming in the terms an organisation can count, because the chapters that follow return to it. The most expensive hours in most teams are its manager's, and some of them are spent on work that a colleague could have done and learned from. The decisions only the manager can take are delayed by the same fragmentation, and every day of delay is borne by the people waiting. Both can be seen in records a firm already keeps: in calendars, in the dates on which decisions are taken, in how long a team waits for an approval. How such a gain is measured honestly enough that both the manager and the organisation would sign the number is the question of Chapter 5, and it has an old and practical answer. The house has spent a whole book, Where the Flow Stops, on how a measure is agreed before anyone knows what it will say.
None of this guarantees that the plan is written. Some weeks are rightly lost to a crisis, and a manager who protects her mornings from a genuine emergency has mistaken the practice for a rule. The gain is not that the week becomes quiet. It is that the week becomes chosen, and that the person whose time is most expensive knows, for the first time in a while, where it went.
The article this chapter began as ended on a Friday a month later. The book can go further, into the finer points a manager meets when she tries this in her own week, and the first of them is choosing the week.
The week should be ordinary. Not the week of the board meeting, not the week after a holiday, not the week of a crisis. A manager who keeps the diary in an unusual week will spend the reading-back explaining why it was unusual, and will learn nothing she can change. If every week feels unusual, that is itself the finding, and the diary should be kept anyway, with a line at the top saying so.
The second point is the size of the block. Mintzberg recorded every contact, which a researcher can do and a manager cannot. A quarter of an hour is fine enough to show fragmentation and coarse enough to keep. The entry is written at the end of each block or each meeting, in a few words, and the three marks are added at the same moment: who asked for it, who else could have done it, and whether it moved one of the two or three things the manager has said in writing are hers this quarter. Entries made at the end of the day are already a story. Entries made on Friday are a wish.
The third is honesty about the work that does not appear in a calendar. Much of a manager's week is spent in conversations nobody booked: the question at the coffee machine, the message answered on the way to a meeting, the twenty minutes after a meeting spent with one person who stayed behind. Kotter's general managers were notable for exactly these short, unplanned exchanges, and they are often the most useful part of a manager's week.7 The diary records them as blocks like any other, because a week read back without them looks emptier and more controllable than it was.
The fourth is how to read it. The manager sorts the blocks, on Friday, in three ways. By who asked: herself, her team, her own manager, peers, outside. By whether someone else could have done it, and if so whether that person would have learned from it. And by whether it moved her own few priorities. Most managers find three things, in varying proportions. A large share of the week was asked for by other people, which is ordinary and often right. A noticeable share could have been done by someone else, and some of that was work the other person would have grown by doing. And the hours that moved her own priorities were few and scattered, never long enough to think in.
The fifth is the order of the first changes, and it matters more than any of them. The first change is to tell the team, in a sentence, what the diary showed and what she is going to do about it, so that what follows is not experienced as withdrawal. The second is to protect a small number of longer blocks for the work only she can do, two mornings or three, and to say when she is reachable instead. The third is to hand one recurring piece of work to a named person, with the authority to decide, because the diary showed that her presence added little. The fourth is to look hard at the standing meetings, which are the easiest blocks to change and the ones most often left alone. Chapter 7 carries the diary page and the reading sheet, and Chapter 6 places the week in the first ten days of a ninety-day plan, before anything else changes.
There are three mistakes worth naming. The first is to treat the diary as a time-management exercise and look for blocks to cut. The point is to see whose hours they were and what the position is being asked to carry, which is a question about the organisation as much as about the manager. The second is to protect mornings without the team's agreement, which moves the interruptions to the afternoon and adds resentment to them. The third, and the most common, is to keep the diary once. A week read back is a photograph. Kept again for one week each quarter, it becomes a record against which the manager, and later the organisation, can see whether anything has moved.
And there is one decision that changes the result more than any other. It is whether the manager shows the reading-back to her own manager. A diary kept privately changes one person's week. A diary shared upward, with the hours her manager wrote highlighted without complaint, changes two, and it is the beginning of the conversation about what the position is for.
It is twenty to seven on a Friday, and the director is not the last person on her floor; one of her managers is finishing the supplier summary he now writes himself. The calendar has thirty-two entries this week. Two of them are mornings with a single word in them, plan, and on one of the two she was interrupted, because a site lost power, and she went.
The network plan is open in the other window. It is at page nine, and it has a date on it for the chief executive. The mug beside the keyboard is empty, and she does not remember making the tea.
She writes the weekly note in ten minutes, because for once she knows what the week held, and closes the laptop.
Chapter Two
A span of control is a count of relationships, and past a certain size no leader can hold it.
Seventy-Four Names on One Rota
Haute Lumière · The Manager's Week · Chapter Two
It is a quarter past seven on a Wednesday, and the nurse manager of two surgical wards is standing at the desk between them with the week's rota in her hand. It is printed, because the screen version will not show both wards at once. There are seventy-four names on it. Each of them is, on the organisation chart, her direct report. She has a yellow highlighter, and she is marking the people she has not spoken to properly in a month, and she has stopped at nineteen because it was making her jaw ache.
At half past seven the night charge nurse will hand over. At eight there is a bed meeting. Between nine and eleven there are two return-to-work interviews, a complaint from a family to answer and an incident review. The newly qualified nurse who started on Monday has been told she will meet her manager this week. It is Wednesday.
She is a composite, drawn from the way these mornings are described in the research below and in the house's own practice, and no hospital is meant. Most managers will know some version of her rota, whether the names are nurses, drivers, call handlers or engineers, and whether there are seventy-four of them or fourteen.
Ammanuel Santa Anna's canon, writing about managers as the people on whom an organisation's development most depends, puts the stakes of her morning in three sentences that read quite differently when the list is seventy-four names long:
Every one-on-one conversation is a developmental moment. Every delegation decision is a growth opportunity or a growth suppression. Every team meeting either builds collective intelligence or diminishes it.
This chapter makes one claim. A manager's span of control is not the number of boxes beneath hers on a chart. It is the number of working relationships one person's week can actually hold, and when the chart asks for more than that, the manager's time stops being leadership and becomes triage. The cost of that does not appear on the chart. It appears on the ward.
The first person to put numbers to the question was a management consultant named Vytautas Andrius Graicunas, who in 1933 published a short paper on relationships in organisations, later reprinted in the collection that Luther Gulick and Lyndall Urwick edited in 1937, Papers on the Science of Administration.9 His point was simple and has been quoted ever since. When a manager adds a subordinate, she does not add one relationship. She adds a relationship with that person, relationships with every group that person now belongs to, and the relationships between that person and each of the others, which she must also hold in mind, because they shape the work.
Graicunas counted them. His formula gives the number of relationships for a manager with n direct reports as n multiplied by (2 to the power n, halved, plus n, minus one). With two subordinates there are six. With five there are a hundred. With six there are two hundred and twenty-two. The count roughly doubles with every person added, long before the number of people looks large.
It would be a mistake to take the formula literally, and Graicunas's critics did not. Nobody holds two hundred and twenty-two relationships consciously, and most of the possible groups never form. What the arithmetic captures is the shape of the problem rather than its size: the load on a manager grows much faster than the headcount, because people do not work alone, and every new person changes the relationships between the ones already there. A rota of seventy-four is not seventy-four conversations. It is a ward's worth of pairs, shifts and alliances, and a manager is expected to know which of them matters this week.
Santa Anna's canon describes the manager's place in exactly these terms, as the point where people and structure meet. The position, it says, is "close enough to individual contributors to influence their daily experience, and connected enough to organizational systems to shape the conditions in which work happens." Both halves of that sentence take time. The first half takes most of it.
Gulick's three questions
Haute Lumière · The Manager's Week · Chapter Two
Gulick, writing in the same volume, gave the principle its best-known form: that one person can directly supervise only a limited number of others, and that the right number depends on the work.9 He named the conditions that would move it: how different the functions being supervised are, how much time is available, and how far apart the people are in space. A supervisor of people doing one kind of work, in one room, with time to spare, can hold more than one who supervises five specialisms across three sites.
In 1946 Herbert Simon published a famous attack on this whole style of reasoning, The Proverbs of Administration.10 The principles of administration, he argued, came in pairs that contradicted each other, like proverbs: a narrow span of control was good because it allowed close supervision, and few levels of hierarchy were good because they kept decisions close to the work, and an organisation could not have both. The critique was so widely accepted that the subject was largely abandoned. Kenneth Meier and John Bohte, returning to it in 2000, put the irony plainly: "little systematic research existed on many of the principles of management at the time Simon leveled his critique."11
Meier and Bohte then did the research. Using data on 678 Texas school districts over four years, they tested how the span of control at different levels of a district related to how its students performed, and found that "span of control relationships among organizational personnel significantly shape student performance."11 In a second paper in 2003 they tested Gulick's three conditions directly, and found him broadly right that diversity of function, time and space shape spans, with the twist that these conditions "have different meanings that depend on the level of organizational hierarchy analyzed."12 The right span for a head teacher is not the right span for a superintendent.
Gulick's conditions had already been seen at work in industry by then. In the 1950s the British sociologist Joan Woodward surveyed manufacturing firms in south Essex and found, in the study she published as Industrial Organization in 1965, that the spans of first-line supervisors varied with the kind of production a firm ran: unit and small-batch work, mass production and continuous process each settled at a different typical span.13 The detail of her figures is best read in the book, which was not reopened for this book. The direction is enough. The number of people one supervisor holds well is set by what the work asks of the supervisor, not by a ratio chosen in head office.
That finding is worth holding beside the rota. A surgical ward that admits from theatre all day, with patients changing hour by hour and staff rotating across two units and three shifts, has every one of Gulick's conditions pulling the right span downward: varied work, little time, people spread across space and hours. It is close to the opposite of the conditions under which a wide span is easy.
These are studies of schools, and of the American public sector, and they measure association across districts rather than the effect of changing any one district's structure. Their value here is narrower and firmer: they show that the span is not a matter of taste. It is a design variable, it varies with the work, and it is connected to results that someone outside the organisation can count.
The canon, writing about how authority moves through organisations, makes a related point about what structure is for. Its aim, it says, is to redesign how power flows "so that authority moves toward intelligence rather than intelligence being forced to route through authority." A span too wide for a week to hold is one in which intelligence has to queue for authority. Seventy-four people, each with something the manager ought to know, are routing it through one person who has nineteen of them highlighted.
The most direct evidence on what a wide span does comes, as it happens, from wards like hers. Hospital restructuring in many countries during the 1990s and 2000s gave nurse managers responsibility for more units and more staff, and researchers in Canada set out to measure what followed.
In 2009 Ann McCutcheon, Diane Doran and colleagues reported a study across seven teaching and community hospitals: 51 units, 41 nurse managers, 717 nurses and 680 patients.14 They measured the managers' leadership styles, their spans of control, how satisfied the nurses were in their jobs and how satisfied the patients were with their care. Leadership style mattered, as other research had found: the nurses of managers rated as more transformational were more satisfied. But span mattered too. In the authors' words, "the wider the span, the lower the nurses' job satisfaction and patient satisfaction." And the finding that should interest anyone who designs an organisation came last: "as spans of control increase in size, no leadership style, even transformational, can overcome the negative effects."
Two years later a related team, with Raquel Meyer as first author, looked more closely at how a manager's time with her staff fitted in.15 Across four acute care hospitals, 31 front-line nurse managers and 558 nurses, they combined surveys with the managers' own work logs. They had expected that the time a manager spent in contact with staff might explain more than the raw count of reports. It did not, quite: span interacted with leadership and with the unit's hours of operation to explain how satisfied nurses were with their supervision. Their practical recommendation was to design the role around the manager's "daily span (average number of direct report staff working per weekday)," rather than the total on the rota, because a unit that runs around the clock spreads its seventy-four people across shifts the manager is rarely present for.
A word on the strength of this evidence. Both are cross-sectional studies, which find associations at a point in time and cannot say for certain which way the effect runs. Nursing is one kind of work, with shifts and a clear patient outcome. But the finding is consistent with Graicunas and with Gulick, it comes from people with no interest in proving a management theory, and it carries a sentence that is hard to forget: past a certain size, being a better leader does not help.
That sentence changes where the problem sits. It also explains something that managers in her position describe, and that the research does not set out to measure: the particular tiredness of being responsible for people one has no time to know. It is not the tiredness of hard work. It is the tiredness of a debt that grows every week, nineteen names long and rising, which no amount of effort in the remaining hours can repay. A nurse manager whose staff are unhappy and whose patients complain is usually sent on a course. If the span is the cause, the course will not help, and she will conclude that the failure is hers.
The nurse managers' spans widened because their organisations removed layers, and in this they were part of a larger movement. In 2006 the economists Raghuram Rajan and Julie Wulf published a study of the reporting structures of more than three hundred large American firms over the 1980s and 1990s.16 Firm hierarchies were becoming flatter. "The number of positions reporting directly to the CEO has gone up significantly over time while the number of levels between the division heads and the CEO has decreased." The chief executive's own span had widened, and so, beneath her, had many others.
The authors were careful not to call this simply centralisation or decentralisation, and the movement had good reasons behind it: fewer layers can mean shorter distances between a decision and the people it affects, less translation and lower cost. The house's own canon is not nostalgic for tall hierarchies. But it is equally clear about what the opposite enthusiasm produced: "This is not the elimination of hierarchy. That simplistic prescription — "flatten everything" — has produced its own failures". Removing a layer moves its work. On the two wards, the layer that went was a ward sister for each unit, and her work did not disappear with the post: the rota, the return-to-work interviews, the conversation with the family, the newly qualified nurse's first week. Some of it went up to the manager, who now held two wards. Some of it went sideways to the charge nurses, who were never told it was theirs. Some of it went nowhere, and that part shows up later, as the nurse who leaves in her first year without anyone having asked how she was. If nobody decides where the work goes, it goes to the people on either side of the gap, and they absorb it in their evenings.
The canon's sharper line on structure comes from a book about executive perception, and it is a question worth asking of any span: "If we replaced every person in this organization with an equally talented alternative, would the same problems recur?" Put to the two wards, the answer is plain. Another nurse manager, equally good, would stand at the same desk with the same rota and stop at the same nineteenth name. The problem is not in the person. It is in the number of people the chart has asked one week to hold, and what her week actually spends itself on, the subject of Chapter 1, follows from that number before she has made a single choice.
Counting the real span
Haute Lumière · The Manager's Week · Chapter Two
Picture the nurse manager a fortnight later, with a different sheet of paper and a decision to make before her director's review.
She has stopped counting names and started counting relationships, in three ways. The first is the daily span the Canadian study recommended: on an ordinary weekday, how many of the seventy-four are actually on duty while she is? About twenty-two. The second is who, in a week, brings her something that needs a decision or a conversation, which she has recorded for five days on the back of the rota: thirty-one different people, of whom nine account for most of the contact. The third is the people she is responsible for and has had no working contact with at all in a month, which is the column the highlighter was really drawing: nineteen, mostly on nights and weekends.
The count is the beginning, not the answer. What it shows her is that the chart and the week are describing two different organisations. The canon has a phrase for the person the second organisation depends on, and it describes the charge nurses exactly: "the middle manager who understands the team's real capacity". The charge nurses know who is tired, who is ready for more, who is quietly covering for whom. The chart does not show them as managers at all. On the chart she has seventy-four reports. In the week she has about nine relationships she can hold well, about twenty she holds in passing, and a large group she holds not at all, among them the new nurse who has been waiting since Monday. And it shows something the chart hides completely: three of her senior nurses, the charge nurses who lead the shifts she is absent from, are already doing much of the managing on those shifts, informally and without the time or the authority for it.
What she takes to her director is not a request for a third ward manager, though that may come. It is a proposal to make the existing structure true. Each charge nurse would hold a defined group of around eight to twelve people as their own, with protected time to do it and the authority to take the decisions they are already being asked to take. She would hold the charge nurses, the newly qualified, and the decisions that cross both wards. The counting is the evidence; the redesign is a decision her director can make, and the conversation is easier because the evidence is in the room.
Chapter 7 carries the count as a working page, with its columns, the daily span, the arithmetic and the questions to ask of each group, and the section below takes up its finer points. Its finer points are where it goes wrong. A manager who counts only the names repeats the chart's mistake. A manager who gives the charge nurses a group without the time to lead it has added work to the people already absorbing it. A manager who redesigns the span without asking the people inside it will draw groups that make sense on paper and split pairs who have worked every night shift together for five years. And the most common mistake is to treat the redesign as a promotion for some and a demotion for others, when what it changes is not anyone's rank but who is answerable to whom on a shift, which is the question the view from the middle of any organisation is built around, and the subject of Chapter 3.
The director's question, when the proposal reaches her, will be what it costs, and the honest answer is that most of it is already being paid. The charge nurses are already doing the work, unprotected, in time taken from patients and from their own breaks. The proposal makes the cost visible and gives it a shape. It may need some time bought back from the rota, and that is a decision for the director, who has the budget. It does not need a new post to begin.
There is a gain in this, and it can be counted in records a hospital already keeps: in the satisfaction surveys the Canadian researchers used, in turnover among new staff in their first year, in the time from an incident to its review, in sickness absence on the shifts the manager rarely sees. Which of those is the right measure, over what history and with what allowances, is a conversation for each organisation to have before anything changes, and Chapter 5 returns to it; the house has written at length, in Where the Flow Stops, about what it means to agree a measure before anyone knows what it will say. The principle does not depend on the details. A span that a week can hold is one in which leadership reaches the people it is meant for.
The article this chapter began as ended with a rota and a drawer. The book can go further, into the finer points a manager meets when she counts her own span, and the first of them is what to count.
The chart gives one number. A manager who wants to know what her week is being asked to hold needs four. The first is the formal span, the names on the chart. The second is the daily span, the number of those people who are at work while she is, on an ordinary day, which the Canadian researchers recommended designing around and which is very different from the first number in any organisation that works in shifts, across sites or across time zones. The third is the working span, the number of different people who, in an ordinary week, bring her something that needs a decision or a conversation. And the fourth is the unheld span, the people she is responsible for and has had no real working contact with in a month. The fourth is the one the chart cannot show, and the one the highlighter was drawing.
The second finer point is the people who are not on the chart at all. Most managers hold relationships the chart does not record: a contractor who works inside the team, a person in another department who reports to her on one project, a peer whose team is the other half of a process, an informal lead who brings her the problems of a group that officially reports elsewhere. They take time like anyone else. A count that leaves them out will show a span that looks manageable and a week that is not.
The third is the arithmetic, held lightly. Graicunas's formula is not a planning tool, and no manager should set her span by it. What it is good for is a single conversation: showing a director, who sees the span as a number of people, that the load grows with the relationships between them. A team of five that works closely together asks more of its manager than a team of eight who each work alone. Chapter 7 gives the formula with its first values worked, for use in exactly that conversation and no other.
The fourth is the work itself, read through Gulick's three questions. How different are the things the people in the span do? How much of the manager's time is actually available for them, once the week's other demands are met? And how far apart are they, in space and in hours? A manager whose span includes three specialisms across two sites and three shifts is holding a much larger span than the count suggests, and the same count in a single room doing a single kind of work may be well within what a week can hold.
The fifth is what to propose. The count usually shows one of three things, and each has a different answer. Sometimes the span is simply too wide for the work, and the answer is a structural one that belongs to the manager's director: another manager, or a split. Sometimes the span is reasonable on paper but the manager is holding relationships that belong elsewhere, and the answer is to move them, which she may be able to do herself. And often, as on the two wards, the organisation already contains informal leaders doing much of the work without the time or the authority, and the answer is to make the structure true: to name the groups they already hold, give them protected time to hold them, and give them the decisions they are already being asked to take. This last answer is the cheapest, because most of its cost is already being paid.
There are mistakes worth naming. Counting only the chart. Proposing a new structure without asking the people inside the old one, which splits working pairs and loses the knowledge of who works well with whom. Giving informal leaders a group without the time, which converts their goodwill into overtime. And treating the conversation as a request for headcount, which invites the answer no before the evidence has been read. The count is evidence about how the work is held. It is best presented that way, beside the records the organisation already keeps about the people in the span: their turnover, their sickness absence, how long new joiners take to settle.
And the decision that most changes the result is whether the manager asks, for each group in the new structure, who will hold the person who is newest. The newest person in any span is the one most likely to be unheld, and the one whose first months decide whether she stays. Chapter 6 places the count in the first weeks of a ninety-day plan, and Chapter 7 gives the four columns and the questions as a single page.
It is a quarter past seven on a Wednesday, and the nurse manager is at the same desk with a new rota. It still has seventy-four names on it. Beside each name, now, is a second name, the charge nurse who holds that person this month, and three of the charge nurses have their own protected hour on Thursdays, which is written in the rota as if it were a shift, because it is.
The newly qualified nurse who started on Monday met her charge nurse on Monday, and her manager on Tuesday. The highlighter is in the drawer.
At half past seven the night charge nurse hands over, and stays for five minutes afterwards, to tell her something about one of his eight.
Chapter Three
The middle manager is the organisation's translator, and the upward half of the translation is where the gain is.
The Translator's Thursday
Haute Lumière · The Manager's Week · Chapter Three
It is ten past four on a Thursday, and the manager of a claims team at a regional insurer is sitting in a small glass room with the door shut and three drafts of the same email open. She came back an hour ago from a two-day leadership offsite in a hotel by the motorway, and she carries a decision from it. The telephone line her team answers, the one customers ring after a flood or a break-in, will close in the spring, and claims will be made through an online form. The decision was explained well. The figures were persuasive. She agreed with most of it in the room.
Her team meets at half past four. Fourteen people, several of whom have answered that telephone line for longer than the director who announced its closing has worked for the firm. The first draft of the email is enthusiastic and sounds like the slides. The second is careful and sounds like a lawyer. The third begins I know some of you will have questions, and she deleted the rest of it because it sounded like an apology. Her throat is tight in the way it gets before she has to say something she half believes.
She is a composite, drawn from the way these afternoons are described in the research below and in the house's own practice, and no firm is meant. If you have managed people who are managed by someone else, you have sat in that glass room.
Ammanuel Santa Anna's canon has an exercise for senior leaders that asks them to imagine the middle of their own organisation, and the questions it asks are the ones in her throat:
Inhabit the perspective of middle management — the translators between strategic intention and operational reality. What pressures do you feel from above and below? What information is lost in translation? What impossible contradictions are you expected to resolve?
This chapter makes one claim. The middle manager is not a relay through which the organisation's signal degrades. She is its translator, and translation runs both ways. Most organisations use only the downward half, carrying decisions to the people who will live with them, and leave the upward half, carrying what those people know back to the people who decided, to chance. The upward half is where the gain is, and it costs almost nothing to collect.
The middle of the organisation has rarely had a good press. In the early 1990s, as firms reengineered their processes and removed layers of management, the middle manager became the standard example of cost without value: the person who passed messages up and down, and whom a good information system could replace. Steven Floyd and Bill Wooldridge, writing in 1994, put the prevailing mood in their title: Dinosaurs or Dynamos?18
Their answer was that something had been lost in the delayering that nobody had measured. In their research, the abstract says, middle managers played a role in each element of how an organisation learns: interpreting the world, uncovering new market opportunities, focusing existing resources and accumulating new ones. And they "uncovered a strong relationship between the roles and organizational performance." Their wider body of work describes four strategic roles for the middle, two looking upward and two looking downward: championing alternatives and synthesising information for the people above, facilitating adaptability and implementing deliberate strategy for the people below. The names are from their books and are given here from memory. The shape is the point: half of what they saw the middle doing was carrying knowledge up.
It is worth dwelling on what that means for the claims manager, because the delayering argument assumed the opposite. A message passed down a hierarchy does lose detail at each layer; that much of the relay picture is true, and a firm that uses its middle only as a relay is right to ask whether it needs so many of them. What Floyd and Wooldridge saw was that the same people, used differently, were the ones who noticed when the plan from above did not fit the market below, and who argued for the alternative before anyone senior had seen the need. A firm that removed them saved their salaries and lost that noticing, and the loss appeared nowhere, because nobody had counted what the noticing was worth.
Santa Anna's canon describes the position in the same two directions, and it is honest about why the upward half so often fails. Writing about how different people in an organisation see the same process, it says:
Middle managers occupy the most paradoxical position: they see enough to know that the view from the top does not match the reality at the bottom, but they often lack the authority to act on this knowledge and may be incentivized to smooth the discrepancy rather than surface it.
The last clause is the one that matters for the claims manager at ten past four. She knows, better than anyone at the offsite, what her team will say about the telephone line, and some of it is a forecast the offsite needs: which customers will not use a form, which claims cannot be described in one, what the team does on a flood weekend that no process map records. Every incentive in her position tells her to smooth it. The director wants the decision carried. Her team wants it softened. The email that sounds like the slides is the smoothest one, and it is the one most likely to be sent.
Three years inside a change
Haute Lumière · The Manager's Week · Chapter Three
The most careful look at what middle managers actually do during a large change was made by Quy Huy, who spent three years inside one large firm's attempt at radical change and published his account in 2002.19 He was looking at emotion, which the literature on strategy had mostly treated as noise, and he found that the middle managers who helped their groups adapt did two things at once that look as if they should conflict.
They committed themselves, personally and emotionally, to change projects they had chosen to champion. And they attended to the emotions of the people the change was happening to. Huy called the combination emotional balancing, and he described what happened when either half was missing. "Low emotional commitment to change led to organizational inertia, whereas high commitment to change with little attending to recipients' emotions led to chaos." Where both were present, the groups managed change and continuity at the same time: customer service held up while the new work was learned.
It is one firm, studied inductively, and its strength is in the depth of the observation rather than in any figure. But it describes the claims manager's dilemma with unusual accuracy. The enthusiastic email is commitment without attending. The careful one is attending without commitment. Neither is what Huy saw working. What he saw was a manager who meant the change and also meant to hear what it cost, and who did both in the same conversation.
It also helps to say what emotional balancing is not. It is not the manager absorbing her team's feelings so that the change can proceed without them. Huy's managers attended to their people's emotions as information about the change: where it would be resisted, what it would cost, what would have to be protected for the work to go on. The attending was part of the strategy, not a kindness offered beside it.
There is a reason the middle can do this and the top usually cannot. A director announcing a change to four hundred people can attend to their emotions only in general. A manager with fourteen people knows that one of them was answering that line the night her own house flooded, and will hear the news differently. The same knowledge sits lower down too, in the charge nurse of Chapter 2, who already leads a shift nobody has told her is hers, and the size of the group a middle manager holds decides whether she can know her people this well at all. The canon names the position as "the precise point where human development and organizational structure meet." Huy's study is a description of what happens at that point when it is used well.
If the middle carried only feelings, a firm might still decide it could do without them. The more surprising evidence is about results.
In 2012 Ethan Mollick published a study of a creative industry, the making of computer games, in which a firm's success is usually credited to its innovators, the designers and engineers whose ideas become products.20 Using a statistical model that could separate the contribution of the firm from the contribution of the individuals who worked in it, he asked how much of the variation in firms' performance could be traced to individuals, and to which individuals. The abstract states the answer without decoration: "variation among middle managers has a particularly large impact on firm performance, much larger than that of those individuals who are assigned innovative roles." In an industry that tells its history through its creators, the people who organised the work explained more of the difference between firms than the people who invented it.
It is one industry, and the middle managers in it are project managers of a particular kind. But the finding points the same way as Floyd and Wooldridge's, from a different method and two decades later. The middle is not where the signal degrades. It is, often, where the work is made to happen.
How people arrive in the middle helps explain why so many organisations underuse them. In 2019 Alan Benson, Danielle Li and Kelly Shue studied the promotion of sales workers to manager at 131 firms, and found evidence consistent with the Peter Principle: firms tended to promote the best salespeople, "at the expense of other observable characteristics that better predict managerial performance."21 The best seller was not, on average, the best manager, and the costs of the mismatch were high. The claims manager was promoted because she was the best claims handler on the floor. Nobody, at the time, asked whether she was good at carrying things up.
There is a cost to the middle that the people above it rarely see, and a theory paper from 2017 gives it a name. Eric Anicich and Jacob Hirsh, reviewing decades of research on power, noticed that it had mostly compared people with power to people without it, and had neglected the experience of having both.22 They proposed that people in positions of middle power switch frequently between two patterns of behaviour, one for the people above and one for the people below, which they called vertical code-switching, and that frequent switching, under certain conditions, produces role conflict and a tendency to hold back. It is a set of propositions, not a finding, and it should be read as one. But anyone who has sat in a glass room between an offsite and a team meeting will recognise the description.
The theory has a practical consequence that is easy to miss. If the strain of the middle comes partly from switching between two registers, then some of it can be relieved not by asking the manager to be braver, but by giving her a third register that both audiences recognise: a written form, used every time, in which she says the same thing upward that she has said downward. A habit carries less risk than a confession, and it is heard differently.
At the offsite the claims manager spoke in one register: she asked sensible questions about the timeline and agreed with the figures. At half past four she will speak in another. And between the two, in the glass room, she is doing the work that neither audience sees, deciding what she can say to each without betraying the other. The canon describes the structure that produces this in a sentence about one organisation's difficulties, and it is general enough to be true of many: "the power structure that concentrated decision-making authority in the C-suite while distributing accountability to middle management." She will be accountable for how her team takes the news. She had no part in the decision.
The Cairn, another of the house's books, reads [the leaders who are the best listeners in the building and the last to say how they are](https://hautelumiere.life/editions/rsd-rsd-and-self-confidence/page/), and the middle is where that pattern is most often learned. A manager who spends her week translating in both directions, and is heard in neither, learns to keep her own view to herself. When that happens across a whole layer of an organisation, the canon's book on how organisations move through change predicts what follows, in a phrase that should worry any director planning a change: the discovery that "middle management has been quietly undermining the transformation because no one asked them what they thought."
Carrying it up
Haute Lumière · The Manager's Week · Chapter Three
Picture the claims manager the following Thursday, having decided to do one thing differently. She sent none of the three drafts. She told her team in person, at half past four, in her own words: what was decided, why, what she thought of it, and what she did not yet know. The same exercise in the canon that asks a leader to imagine the middle asks her first to imagine the front line, and one of its questions is close to the one the claims manager had chosen: "What do you know that never makes it into reports?" Then she asked them a question she had written on a card, so that she would ask it the same way every time: What do you know about this that the people who decided it might not?
The answers took forty minutes and filled two pages. Some were grief, and she let them be grief. Some were preferences. And some were forecasts of exactly the kind the offsite needed: that a good share of their callers, in the team's own experience, were elderly policyholders who had never used the online account; that the form had no place for a claim involving two properties; that on a flood weekend the team triaged calls by the sound of the caller's voice, which no form can hear.
The next morning she did the thing most middle managers do not do. She wrote to her director, on one page, with three headings: what the team understood and accepted; what it would cost them, said plainly; and what they knew that the decision might not have accounted for, with the two pages attached. She sent it within a week of the offsite, while the design of the change was still open, and she asked for one thing only, a reply.
The order of what she did matters more than the note. She carried the decision down before she carried anything up, so the team heard it from her and not from the intranet. She separated the grief from the forecasts, and carried only the forecasts as evidence, so that her director read them as information and not as resistance. She did it fast, because a forecast that arrives after a design is settled is a complaint. And she made the note a habit rather than an event: one page upward for every significant decision carried down, so that the director learned to expect it and to read it.
Chapter 7 carries the note upward as a working page, with its three headings and the question card, and the section below takes up the points at which it goes wrong. Its commonest failures are predictable. A manager who carries only the grief up is heard as defending her team. A manager who carries nothing up, because the decision is made, teaches her team that their knowledge stops at her door, which is the meeting after the meeting moved one floor down, the pattern Where the Flow Stops found in the corridor outside every boardroom. And a director who receives the note and does not reply has taught the manager not to send the second one.
It is worth looking at the note from the director's side, because the practice depends on her as much as on the manager. The director who announced the closing of the line did so on the strength of figures that were real: call volumes falling, the cost of each call, the share of customers already using the online account for other things. What she did not have, and could not have had from where she sat, was the floor's knowledge of which calls the figures described badly. She had asked, at the offsite, whether anyone had concerns, and nobody had, because the people with the concerns were fourteen miles away answering the telephone. The note is how that knowledge reaches her while the design is still open. A director who reads three such notes in a quarter, from three managers, has a better map of her own change than any survey could give her, and it cost each manager an hour.
The note also changes what the middle manager is for. A manager who only carries decisions down is the relay the delayering decade thought it could remove. A manager who carries forecasts up is the only person in the organisation who can tell the people at the top what their decision will meet, in time to change it. The gain is the cost of the problems the forecasts prevented: the claims that would have failed in the form, the complaints that would have followed, the weekend of floods handled badly. It can be counted in records an insurer already keeps. And it depends on something that has nothing to do with the manager's skill, which is whether the questions that belong to her team are left with them, the subject of Chapter 4, so that she has the hours to listen.
The article this chapter began as ended with an email read aloud. The book can go further, into the finer points a middle manager meets when she makes the note upward a habit, and the first of them is the conversation that comes before it.
The note is only as good as what the team says, and what the team says depends on how the decision reaches them. The order is fixed. The decision is carried down first, in person, by the manager, in her own words: what was decided, why, what she thinks of it, and what she does not yet know. Saying what she thinks is not disloyalty to the people above her, provided she says it as her view and not as a campaign. Saying what she does not know is the most useful sentence of the four, because it tells the team that the decision is not finished and that what they know can still reach it.
The second point is the question. The claims manager wrote hers on a card so that she would ask it the same way each time, and the habit is worth keeping. What do you know about this that the people who decided it might not? is a question that assumes there is something to find, cannot be answered with silence, and asks for knowledge rather than opinion. It is different from any concerns?, which invites the room to reassure her, and from how do you feel about it?, which is a fair question for another moment and produces a different kind of answer.
The third is sorting what comes back. A team hearing a decision that changes its work will produce three kinds of sentence, and they are the same three that the meeting after the meeting produces in Where the Flow Stops: forecasts of what will happen, preferences for what the speaker would have done instead, and grievances, which are usually a forecast with a history attached. The manager hears all three, and honours all three in the room. She carries only the forecasts upward as evidence, together with a plain account of what the change will cost the team. Grief carried upward reads as resistance; a forecast carried upward reads as information, and the director can use it.
The fourth is the form. One page, three headings, the same each time: what the team understood and accepts; what it will cost them, said plainly; and what they know that the decision may not have accounted for, with enough detail that a designer could act on it. It goes within a week, while the design of the change is still open, and it asks for one thing only, a reply. A note that asks for a reversal will be read as an appeal. A note that asks for a reply will be read.
The fifth is the director's side, which the manager cannot control and should still prepare for. Some directors will not reply. The note should go anyway, for three weeks, because a habit is how it becomes ordinary. If there is still no reply, the manager has learned something about the organisation that is worth knowing, and the next conversation, with her director, is about how the people who decide want to hear what the people who carry out know. That conversation goes better with three unanswered notes in hand than with none.
There are mistakes worth naming. Carrying a decision down by email, which removes the moment in which the team's knowledge could be heard. Carrying nothing up because the decision is made, which teaches the team that its knowledge stops at the manager's door. Carrying everything up, unsorted, which buries the forecasts in the grief and makes the manager look as if she has taken her team's side against the organisation. And writing the note in the register of the slides, which says nothing the director does not already know.
And the decision that most changes the result is whether the manager names, in the note, the people on her team who saw what the decision missed. A forecast with a name attached can be followed up. It also tells the team, when the reply comes, whose knowledge reached the people who decide, which is the strongest reason they will ever have to say what they know the next time. Chapter 6 places the note upward in the middle of a ninety-day plan, once the manager's own week and span have been read, and Chapter 7 gives the question card, the sorting and the one-page form.
It is ten past four on a Thursday, and the claims manager is in the same glass room with the door open. There is one email on the screen, from her director, and it is short. The form will have a section for claims involving more than one property. The telephone line will close in the summer rather than the spring, and for policyholders over a certain age it will not close at all. And would she bring two of her team to the design meeting next week, the ones who wrote the part about the flood weekend.
Her throat is fine. At half past four she reads the email aloud to the team, all of it, including the part about the two of them.
Then she asks the question on the card again, about something else.
Chapter Four
Every answer a manager gives teaches the next question, and returning a question is design, not withdrawal.
Giving the Work Back
Haute Lumière · The Manager's Week · Chapter Four
It is ten to ten on a Monday, and the manager of a customer support team at a software company has just answered the fourth question of the morning. It arrived, like the other three, as a message in the corner of her screen: a customer wants a refund outside the policy window, and can the team give it. She has answered this question, in nearly these words, perhaps a dozen times since the summer. The answer is usually yes, with a note on the account. She types it again. The message disappears, and another arrives, about a different customer, and it is the same question.
Beside her keyboard is a notebook with the week's priorities written on the first page, under Friday's date. The first of them is the proposal to her own manager for how the team should be organised next year. It is half a page long. Across the open-plan floor, her twelve people are working steadily and well, and every forty minutes or so one of them waits for her before doing the next thing.
She feels it in her stomach, a small clench each time the corner of the screen lights, and something like relief each time she answers, because answering is quick and she is good at it. She is a composite, drawn from the way these mornings are described in the research below and in the house's own practice, and no firm is meant. Most managers who were promoted for being good at the work have had her Monday.
Ammanuel Santa Anna's canon describes what each of her answers does, and it does not blame her for giving it. Writing of the manager who steps in from genuine anxiety, it says:
And every one of these moves, repeated over time, creates the very dependency it fears. The team that's always rescued never learns to rescue itself.
This chapter makes one claim. A good share of a manager's week is spent on questions that belong to her team, and each answer, however kind, teaches the next question. Giving the work back is not stepping away from the team. It is the most useful piece of design a manager does, and it returns hours to her week and judgement to her people at the same time.
The most useful distinction for her Monday was drawn by Ronald Heifetz, a psychiatrist who came to the study of leadership through medicine and taught it at Harvard's Kennedy School. In Leadership Without Easy Answers (1994) he separated two kinds of problem.24 A technical problem is one for which the knowledge to solve it already exists: someone, somewhere, knows the answer, and the task is to apply it. An adaptive problem is one that cannot be solved by expertise alone, because solving it requires the people who have the problem to learn something, change something in how they work, or give something up.
Santa Anna's canon, which builds on Heifetz in its account of leadership, puts the consequence in a sentence:
Heifetz argued that the most common failure of leadership is treating adaptive challenges as if they were technical ones — searching for an expert, a plan, a fix — when what is actually needed is the slower, more uncomfortable work of collective learning and transformation.
The refund question looks technical. It has an answer, and she knows it. But the reason it arrives a dozen times is not technical at all. The team knows the answer as well as she does. What it lacks is the authority to give it and the confidence that giving it will be backed. That is an adaptive problem: it can only be solved by the team taking on a judgement it currently hands upward, and by the manager letting it. Every time she answers, she solves the technical problem and preserves the adaptive one.
The Luminous 100, the house's reading of the frameworks every consultancy sells, states Heifetz's prescription in a line that is easy to agree with and hard to practise: "The leader's job is to create the conditions for others to do adaptive work — not to provide the answer." In 1997 Heifetz and Donald Laurie set out the same idea for a business audience in the Harvard Business Review under the title The Work of Leadership, and one of the principles they named has stayed in use ever since: give the work back to the people.25 The phrase is theirs. What it asks of a manager is the subject of the rest of this chapter.
It helps to look at the refund question from the other side of the floor, because the team's reasons for asking are better than they look. The person who sends the message knows the answer is probably yes. She also knows that if she gives the refund herself and it turns out to be wrong, she will be the one explaining it, and that if she asks, the manager will be. Asking is not a lack of confidence in her own judgement. It is an accurate reading of where the risk currently sits. Heifetz's point is that the leader who wants the team to carry the judgement has to move the risk with it, openly, or the team will go on doing the sensible thing, which is to hand both upward.
The heat, and the holding
Haute Lumière · The Manager's Week · Chapter Four
Giving work back is uncomfortable for both parties, and Heifetz did not pretend otherwise. Adaptive work produces distress, because it asks people to do something they are not yet sure they can. His prescription was to manage that distress rather than remove it: enough to keep people working on the problem, not so much that they stop. The Luminous 100's summary is exact about this: Heifetz "Deliberately raises the heat (productive disequilibrium) while keeping it within a tolerable range."
He borrowed a term from developmental psychology for the conditions that make this possible, the holding environment, and the canon explains why it is not optional:
Without a sufficiently strong holding environment (one that provides both challenge and support), people will default to their existing patterns rather than risk the vulnerability of genuine change.
For the support team, the existing pattern is the message in the corner of the screen. If the manager simply stops answering, the heat rises far past the tolerable: twelve people suddenly responsible for refunds they have never been allowed to give, with no idea where the limit is or whether they will be blamed for crossing it. Some will freeze, and some will give everything away. The holding is what makes the returned question bearable. It is a stated limit, a known person to ask when the limit is unclear, and, above all, the certainty that a reasonable decision made inside the limit will be backed even if it turns out to be wrong.
The canon describes the shape of that holding in its account of autonomy, in a sentence that corrects the most common misreading of empowerment: "Autonomy is not the absence of guidance; it is the presence of choice within meaningful constraints." The constraint is not the opposite of giving work back. It is what makes it possible.
If giving work back were only a matter of a manager's restraint, it would be a question of character. The research on teams suggests it is more a matter of design, and that design matters more than anything a manager does in the moment.
In 2001 Ruth Wageman, a student and colleague of the Harvard psychologist Richard Hackman, published a field study of self-managing teams that compared two kinds of thing a leader does.26 The first was design: how the team was set up, what it was responsible for, what its boundaries and resources were. The second was hands-on coaching: what the leader did day to day to help the team work. Both mattered for how well the team managed itself, how good its members' relationships were and how satisfied they were. But, the abstract reports, "only leaders' design activities affect team task performance." And the two interacted: "well-designed teams are helped more by effective coaching—and undermined less by ineffective coaching—than are poorly designed teams."
It is one study, of teams in one kind of service work, and its strength is the comparison it makes possible rather than any figure. But the finding reorders a manager's week. The answers in the corner of the screen are coaching of a sort, help given in the moment. Wageman's teams suggest that no quantity of good answers would move the team's performance as much as one change in its design: deciding, once, that refunds within a stated limit are the team's to give.
For the support team, design is a short list of decisions that no message in the corner of the screen can make. Which kinds of case are the team's to decide, and up to what limit. Who on the team holds the hard cases when the limit is unclear. When the team looks back at its decisions together, and how. Each of these is a choice the manager makes once, and revisits every few months, rather than a thing she does forty times a week. Wageman's finding is that choices of this kind, made well, do more for the work than any amount of skilful answering, and that they make the answering, when it is still needed, work better too.
Hackman spent much of his career on what such design consists of. In Leading Teams (2002) he set out a short list of conditions under which teams tend to do well: a real team with stable membership and a shared task, a compelling direction, a structure that enables the work, a supportive organisational context, and coaching that is available when it is needed.27 The list is given here in outline, from the book's argument. In 2005 Hackman and Wageman added a theory of team coaching that is unusually practical about timing.28 It identifies, in the abstract's words, "the specific times in the task performance process when coaching interventions are most likely to have their intended effects," and the conditions under which coaching is and is not likely to help. The general point is that help is most useful at the beginning of a piece of work, at its natural midpoint and at its end, and least useful as a constant stream in the middle, which is exactly the stream in the corner of her screen.
There is a reason the manager keeps answering, and it is not that she does not know any of this. Answering feels good, and not answering feels dangerous. Santa Anna's canon is unusually direct about where that comes from, and it places it in the manager's body rather than her judgement:
You cannot empower others if control is your nervous system's default. You cannot build autonomy in your team if ambiguity makes your stomach clench. You cannot foster voice if disagreement feels like threat.
The clench in her stomach when the screen lights is not a weakness to be overcome by resolve. It is a forecast, of the kind managers are paid to make: if this goes wrong, it will be hers. The difficulty is that the forecast is about the one refund, and the cost is in the dozen. Each answer protects her from a small and visible risk, a refund given wrongly, by preserving a large and invisible one, a team that cannot act without her and a manager whose week is full of questions she has already answered.
The cost is also borne by the team, and it is not only time. The canon is careful here, because it does not want the opposite error either: "Empowerment has shadows. It can become abandonment when the manager steps back too far." A team handed authority without a limit, a person to ask or the promise of backing has not been empowered. It has been left. But a team never handed authority at all learns something worse than dependence. It learns that its own judgement is not trusted, and in time it stops forming one. The Luminous 100's summary of situational leadership puts the balance plainly: "No single best leadership style — the best style depends on the maturity of the person for the specific task." Some of the twelve are ready for the refund decision tomorrow. One or two are not, yet, and need the limit drawn more tightly.
Situational leadership's old insight is that the same manager should not treat everyone on a team the same way, and it applies here with some force. The newest person on the support team, three weeks in, is not ready to hold a refund outside the window, and handing it to her on the same terms as the others would be the abandonment the canon warns of. For her, giving the work back means something smaller: being asked what she would do, before being told, and having her answer taken seriously when it is right. The decision she holds this month is narrower than the one her colleagues hold. The direction is the same.
The cost lands in her own week too, and this is where giving work back connects to everything else a manager is asked to do. Every message answered is a minute taken from the half-page proposal that only she can write, the kind of work Chapter 1 found missing from a director's week, and from the note upward of Chapter 3, which only she can send. The questions in the corner of the screen are not separate from her capacity to lead. They are where it goes.
The question returned
Haute Lumière · The Manager's Week · Chapter Four
Picture the same manager a fortnight later, having decided to do one thing differently. For one week, she kept a list. Every time someone brought her a question or a decision, she wrote one line: what it was, who brought it, and one of two marks. A T if it was genuinely technical, something only she knew and the team could not be expected to: a policy change from legal, an account she had personal history with. An A if the team had the knowledge and lacked the authority or the confidence: the refunds, the tone of a reply to an angry customer, whether to escalate a bug.
On Friday she read the list back. The As outnumbered the Ts by a long way, and most of them fell into three kinds. She took the three kinds to her Monday team meeting, one at a time, and for each one she did the same four things. She said that the decision was now the team's. She stated the limit, as a number or a rule, and what lay outside it. She named who to ask when a case was unclear, which was usually not her. And she said, out loud, that a reasonable decision made inside the limit would be backed by her, including when it turned out to be wrong, and that they would look at the hard cases together at the end of each week, not in the moment.
The first returned decision was worse than hers would have been. A refund was refused that she would have given, to a customer with a long history. She did not reverse it. She brought it to the Friday review, where the team looked at the case together and moved the rule slightly, and the person who had refused it was the one who proposed the change. The second week there were fewer questions in the corner of the screen. The fourth week there were almost none about refunds, and two new ones about something she had not thought of, which the team now saw because it was no longer spending its attention waiting.
The Ts she dealt with differently, and the difference matters. A technical question that arrives repeatedly is a sign that an answer exists in one head and not on a page. She wrote the answers to the five commonest on a single shared page, in plain words, with the date, and told the team that the page was theirs to correct. Within a month two of the answers had been improved by people on the team, who knew the edge cases better than she did. A technical problem, Heifetz's distinction implies, still has to be solved once; what it does not need is to be solved by the same person every time.
She also found, reading the list, that a few of the As were questions she had been keeping for herself without meaning to, because she liked answering them. The tone of replies to angry customers was one. She was very good at it, and answering it was the best part of her morning. It went back to the team too, with a page of her best replies attached, and the Friday review took over the teaching.
The order of what she did matters more than the list. She returned the work in kinds, not one question at a time, because a decision returned once is a refusal and a decision returned as a class is a design. She stated the limit before anyone needed it. She moved her help from the stream in the middle to the review at the end, which is where Hackman and Wageman's theory says help lands. And she let the first wrong decision stand where it could be undone, because a manager who reverses the first returned decision has returned nothing.
Chapter 7 carries the escalation sort as a working page, with the two marks, the four things said when a decision is returned and the weekly review, and the section below takes up the signs that a team has been handed too much. The gain it produces can be counted in records a firm already keeps: in the time a customer waits for a decision, in how often decisions are escalated, in the hours the manager spends on the work only she can do. Whether those hours and that waiting are worth something an organisation would sign for, and how such a measure is agreed before anyone knows what it will say, is the question of Chapter 5; this book does not describe how a measured gain is divided.
The article this chapter began as ended with a one-word reply. The book can go further, into the finer points a manager meets when she gives work back to her own team, and the first of them is the list.
The list is kept for one ordinary week, and every question or decision brought to the manager gets one line: what it was, who brought it, how it arrived, and one of two marks. T means technical in Heifetz's sense: the knowledge to answer it exists, and it is genuinely the manager's, either because only she holds it or because the decision belongs to her role. A means the team has the knowledge and lacks the authority, the limit or the confidence. Most managers find the As outnumber the Ts, and that the As fall into a small number of kinds. The kinds are what matter, because a decision returned once is a refusal, and a decision returned as a class is a design.
The second point is what to say when a class of decision is returned, and it is the same four things every time. The decision is now the team's. The limit, stated as a number or a rule, and what lies outside it. The person to ask when a case is unclear, which should usually not be the manager. And the promise, said aloud, that a reasonable decision made inside the limit will be backed, including when it turns out to be wrong. The fourth sentence is the one managers leave out, and it is the one that moves the risk. Without it the team will go on asking, sensibly, because asking is safer.
The third is the review. Hackman and Wageman's account of coaching places help at the beginning of a piece of work, at its midpoint and at its end, and least usefully as a constant stream in the middle. Returning work moves the manager's help from the stream to a review: a short, regular look at the hard cases, held with the team, in which the question is never who got this wrong but what does this case tell us about the limit. The first review is where the practice either becomes part of the team's work or quietly fails, and what decides it is how the manager treats the first decision she would have made differently. If she reverses it, she has returned nothing. If she lets it stand where it can stand, and asks the team what it teaches, she has returned the work and the learning with it.
The fourth is the technical questions. A T that arrives repeatedly is not a sign that the team needs the manager. It is a sign that an answer exists in one head and not on a page. The manager writes the commonest answers down, in plain words and with a date, and gives the page to the team to correct. Within a few weeks the page is usually better than her answers were, because the people using it know the edge cases.
The fifth is the people who are not ready. Giving work back is not the same for everyone on a team, and the canon's warning that empowerment can become abandonment applies most to the newest. For a person three weeks into the job, giving the work back means being asked what she would do before being told, and having her answer taken seriously when it is right. The class of decision she holds is narrower, and widens as she shows she can hold it. A manager who returns the same decision to everyone on the same terms has not designed anything; she has stepped away.
There are mistakes worth naming. Returning a decision without a limit, which raises the heat past what anyone can bear. Returning it without the promise of backing, which leaves the risk where it was. Reversing the first decision. Returning the questions she dislikes and keeping the ones she enjoys, which most managers find, reading the list honestly, they have been doing. And returning work without telling her own manager, who may go on bringing every question in that class to her and undo the design from above.
And the decision that changes the result most is the limit. Set too tight, it returns nothing; set too loose, it returns too much. The practical rule is to set it close to what the manager would have approved anyway, and to widen it at each review while it holds. Chapter 6 places the returned work in the second half of a ninety-day plan, after the manager has read her own week, and Chapter 7 gives the list, the four sentences and the review as working pages.
It is ten to ten on a Monday, and the corner of the support manager's screen lights. It is a message from one of the twelve, and it is not a question. It says that the team gave three refunds outside the window over the weekend, all within the limit, and that one of them was for a customer who had written to the chief executive, and here is what they said to him.
She reads it twice. The notebook beside the keyboard is open at the proposal for next year, which is four pages long now, and has a section in it on which decisions the team should own.
She replies with one word, thanks, and goes back to page four.
Chapter Five
A manager's capacity shows in what her team can do without her, and it can be counted honestly.
What the Team Does Without You
Haute Lumière · The Manager's Week · Chapter Five
It is half past five on a Friday, and the manager of a structural engineering team in a firm of consulting engineers is writing a handover note. On Monday she begins a fortnight of leave, the first she has taken in three years that is longer than a long weekend. Her team is nine engineers and a technician. There are four live projects. The note is meant to say who decides what while she is away.
She has written the projects at the top of the page and stopped. Beside two of them she can write a name without thinking. Beside the third she has written a name and crossed it out and written her own mobile number. The fourth is the one with the difficult client, and she has written nothing, because the honest answer is that every decision on it for a year has gone through her, and she is not sure anyone else knows what she knows about it.
A pot plant on the windowsill has been watered by somebody else all week, which she noticed this afternoon for the first time. Her chest is tight in a way she associates with packing. She is a composite, drawn from the way these Fridays are described in the research below and in the house's own practice, and no firm is meant. Most managers who have tried to take a real holiday have written her note.
Ammanuel Santa Anna's canon, in a book about management as the work of growing people, describes a manager who would have found the note easy, and puts the difference between them in two sentences: "My job is not to produce results through people. My job is to produce people who produce results." And it ends its introduction with the question that her blank fourth line is asking:
What would it mean for you to measure your success as a manager not by what your team produces while you're present, but by what they become after you're gone?
This chapter makes one claim, and it closes the argument the four before it have made. A manager's capacity is best measured by what her team can do without her, and that is not a sentiment. It shows in records an organisation already keeps: in who stays, who grows into a larger role, how decisions are made when the manager is absent, and what the work looks like a year after she has moved on. Measured honestly, agreed before anyone knows the answer, it is a gain an organisation can see and share in.
The best-known modern attempt to answer whether managers matter at all began, at Google, with a hypothesis that many engineers there held in the early 2000s, and that the company's own account states without embarrassment: "Managers don't actually matter."30 The company, which had briefly experimented with removing engineering managers altogether, set out to test it in a multi-year study it called Project Oxygen.
The study's own summary, published on Google's re:Work site, describes the data: "We analyzed over 10,000 data points about our managers, taken from performance reviews and feedback surveys," together with interviews. The hypothesis did not survive. "Teams with highly effective managers were more likely to achieve better results, be happier, and experience lower turnover." The study then named the behaviours that distinguished the managers rated most effective, first eight and later ten, and the first on the list, in Google's wording, is "Be a good coach." Others follow it: "Empower the team without micromanaging," "Support career development and discuss performance," "Make strong decisions." David Garvin wrote the project up for the Harvard Business Review in 2013 under a title that is the story in one line, How Google Sold Its Engineers on Management.31
Read the list beside the handover note and it becomes less abstract. Be a good coach and Support career development are not about the projects the team delivers this month. They are about what the team will be able to deliver when the manager is not there to help: whether the engineer who might hold the difficult client has been given a piece of it before, whether anyone has talked to her about wanting more, whether the manager has ever let her take a decision on it and lived with the result. The blank line on the note is a year's worth of those conversations not had, written in one place.
The Luminous 100, the house's reading of the frameworks every consultancy sells, makes the same observation about the best-researched theory of leadership, that its practitioners grow people who can lead in their turn, "which is why transformational leaders produce more transformational leaders rather than creating dependency." Google's list is that sentence broken into behaviours a team can rate.
A word on the strength of this evidence. Project Oxygen is one company's internal analysis of its own managers, published in summary by the company itself. Its measures are largely ratings by the people managed, and its findings are associations within one unusual firm. It is not a controlled study and it was not designed to be published as one. Its value is in what it was trying to disprove, and in the fact that a firm of engineers, sceptical of management on principle, measured its way to the opposite view.
The value of a boss
Haute Lumière · The Manager's Week · Chapter Five
Two studies from economics put numbers to the same question, from the records firms keep about their own work.
In 2015 Edward Lazear, Kathryn Shaw and Christopher Stanton used a company's data on the measured productivity of service workers, and on which supervisor each worked under, to estimate how much the supervisor mattered.32 They found the effect "to be large." In their abstract's terms: "Replacing a boss who is in the lower 10% of boss quality with one who is in the upper 10% of boss quality increases a team's total output by more than adding one worker to a nine-member team would." And, in a finding that bears directly on the handover note, "Workers assigned to better bosses are less likely to leave the firm."
In 2021 Mitchell Hoffman and Steven Tadelis looked at what they called people management skills, measured through employees' survey ratings of their managers, at a large high-technology firm.33 Such skills, they found, "have a strong negative relation to employee turnover," and several research designs, including following workers who joined the firm or changed managers, supported a causal reading. The same abstract is careful about what they did not find: these skills "do not consistently improve most observed nonattrition outcomes." The better people managers kept their people. They did not, on the measures the firm recorded, always lift everything else.
It is worth translating the first finding into the terms of the engineering office, because the abstract's arithmetic is unusually concrete. A team of nine with a much better boss produces more than a team of ten with a much worse one. Nobody in the firm would hesitate to hire a tenth engineer if the work justified it. Very few firms think of the development of their managers as a decision of the same size, because the extra output does not arrive as a person with a desk. It arrives as fewer mistakes, faster decisions, and people who stay.
That honest limit is worth more than a large claim, and it points to where the measurement of a manager's work should begin. Turnover is recorded, dated and counted by every organisation with a payroll. Its cost is real and is usually known roughly by finance. And both studies agree that it moves with the quality of the person managing.
Gallup, the polling firm, has made a much larger claim, and it is often repeated. In 2015 Randall Beck and Jim Harter wrote that "managers account for at least 70% of the variance in employee engagement scores across business units," citing Gallup's report State of the American Manager.34 The figure is Gallup's estimate from its own engagement surveys, and the article does not describe how it was calculated. It may well be roughly right. But it is a claim about the variance in a survey score rather than about work done, made by a firm that sells the survey, and it is best held as an estimate from an interested party until its method can be read.
Santa Anna's canon tells the story of two managers, and it is worth reading closely, because it describes exactly what the handover note will test. Both led teams of similar size in the same industry. The first, Marcus, was outstanding by every conventional measure: targets met, projects on time, reviews meticulous. The second, Priya, had good figures that were "not always the best in the division." Then each was promoted. Of the first, the canon writes: "When Marcus was promoted, his team functioned well for about three months. Then, slowly, it began to deteriorate. Not dramatically — just a gradual dimming." Its explanation is one sentence: "Because that intelligence had lived in Marcus, not in the team." Priya's team, by contrast, went on producing results after she left, because what she had built was in the people.
The story is the canon's own, told as an account of two people it met, and it should be read as illustration rather than evidence. But it has one feature that makes it useful to an organisation, which is that the difference between the two managers was invisible while they were present. Both teams hit their numbers. Only the absence told them apart. That is the same thing the economists found from the other side: the value of a manager shows up in what her people do, including whether they stay, and some of it shows up only over time.
The canon names the cost of leaving it unmeasured in a sentence from its book on how wisdom passes between people: "It is the silence of a retirement party where decades of institutional wisdom walk out the door in a single afternoon and no one thinks to ask what is being lost." The fourth project on the handover note is that silence in small. What the manager knows about the difficult client exists in one head. If she left tomorrow for good, the firm would find out what it was worth by losing it.
The canon is equally honest about a less comfortable pattern, one that measurement can expose and good intentions often hide. Managers, it warns, tend to "extend more trust, more autonomy, and more developmental opportunities to people who look like them, think like them, or remind them of their younger selves." A manager who believes she develops her whole team may find, when she looks at who got the stretch projects in a year, that she developed three of them. The Cairn, another of the house's books, reads [the second file of recognition other people keep about us](https://hautelumiere.life/editions/rsd-rsd-and-self-confidence/page/), and why recognition works best as evidence rather than as warmth. Who was given the chance to grow is a matter of record.
If a manager's capacity to grow her people is worth something to an organisation, it can be measured, and the measurement can be honest. What makes it honest is the same discipline the house has written about at length in Where the Flow Stops: the measure is agreed before anyone knows what it will say.
For the development of a team, the records usually already exist. Departures, and which of them the organisation regretted, are in the payroll and the exit records. Internal moves and promotions, and where the people promoted came from, are in the personnel system. The time a new joiner takes to work without supervision can be read from the dates on which they first lead a piece of work. And the decisions made below the manager, and how they held up, can be recorded by the team itself over a period long enough to see a pattern. None of these is a perfect measure of development. Each is a proxy, and the canon's caution about proxies holds: "trust itself is not a number," and neither is a person's growth. But proxies agreed in advance, read over a long enough history and checked by someone without a stake in the answer, can be believed by both sides.
One of those records needs reading the right way round. A person who leaves the team for a larger role elsewhere in the organisation is a loss to the team and a gain to the firm, and a manager measured only on her own team's departures will be tempted into what the canon calls "relationship hoarding (leaders keeping talented people on their teams beyond what is useful, rather than allowing them to move where they would contribute most)". Internal moves upward belong on the credit side of the ledger, not the debit.
Two further rules keep them honest. Each measure travels with a second one that would move if it improved for the wrong reason: lower turnover beside the quality of the work, so that a manager cannot keep people by asking little of them; more decisions made below beside how many of them were later reversed. And the history is fixed before anything changes, a year or two of each record copied and dated, so that the improvement is read against the ordinary range of the past rather than against the worst recent month.
The question of who calculates deserves its own sentence, because it is the one most often skipped. A manager is not a disinterested reader of her own team's turnover, and nor is the human resources partner whose programme paid for her development. Neither is dishonest. Neither should be the last person to touch the number. A calculation done, or checked, by someone whose standing does not depend on the answer, and shown to the team in the same form the organisation sees it, is what lets each side believe a figure the other produced.
The gain such a measure finds is not small, and it is one an organisation already pays for without seeing. Every regretted departure has a cost finance can estimate, in recruitment and in the months a replacement takes to reach the same level. Every decision that waits for one person is time somebody else spends waiting. And every manager who cannot take a fortnight away is a single point of failure the organisation has not priced. Which records, which history, which allowances and which second measures are right for a given organisation are questions for that organisation and the people it covers, and they are the substance of an engagement rather than a book. How a measured gain is then shared is not described here at all.
The house's interest in this is stated once, plainly. Luminous Prosperity works with organisations to find the gains they already hold, where capacity and knowledge are spent without being seen, and to share in what is found. This book's subject is the part of that work that sits in a manager's week, and in the questions a team could already answer for itself, which Chapter 4 followed.
The fortnight away
Haute Lumière · The Manager's Week · Chapter Five
Picture the engineering manager on the Friday before, a little earlier in the afternoon, having decided to use her leave as a test rather than an ordeal.
Three weeks before the leave, she wrote the handover note in draft and showed it to her team, blank fourth line and all. She asked them who could hold the difficult client for a fortnight. Two engineers offered, and she chose the one who knew the client less well, because the other already held a project, and spent four of the remaining afternoons with her on the client's history: what had gone wrong in the past, who on the client's side decided what, which questions to answer in writing and which on the telephone. She wrote the limits on one page: what the engineer could agree on her own, what needed the technical director, and the one situation in which she should be rung, which was a safety matter and nothing else.
She asked the team to keep, while she was away, a short log of the decisions they made that they would normally have brought to her, one line each: what, who, and why. She said she would read it when she came back, and that she would not reverse anything in it that could stand, and that she would not read her email. Then she went.
She noticed, on the first Monday of the leave, that her chest was still tight, and on the Wednesday that it had loosened. The canon's book on management would not be surprised. It describes the manager's urge to step back in as a body's forecast that control is slipping, and it is clear that the forecast is information rather than an instruction. Two weeks without acting on it was, for her, the first evidence in three years that the forecast was usually wrong.
When she came back the log had fourteen lines. Twelve were decisions she would have made the same way. One she would have made differently, and on reflection the engineer's was better. One was a mistake, a small one, already corrected by the team before she returned. The difficult client had rung twice and been dealt with, and had written to the technical director to say how helpful the engineer had been.
Chapter 7 carries the fortnight away as a working page, with the draft handover, the limits, the log and the reading of it on return, beside the development ledger that records who in a team was given which chances to grow. The log is also the first entry in a record that most teams never keep. Read beside the handover note from three weeks earlier, it shows which decisions moved, to whom, and how they held. Kept every time the manager is away, and read alongside the team's departures, promotions and the time new joiners take to work alone, it becomes the kind of history against which a manager's development of her people can be read honestly, a year from now, by someone who does not have to take her word for it.
Its commonest failures are the obvious ones. A manager who reads her email on the beach has tested nothing. A manager who reverses the first decision in the log on her return has taught the team to wait next time. And a manager who chooses the holder of the difficult client by who is already strongest has developed nobody, which is the pattern the count of a manager's real span in Chapter 2 tends to expose and the diary of her week in Chapter 1 tends to explain.
The article this chapter began as ended with a laptop left at work. The book can go further, into the finer points an organisation meets when it tries to measure what its managers build in their people, and the first of them is when to begin.
The measure is agreed before anything changes, and the history it will be read against is fixed first. Before any manager changes how she works, someone in finance or in human resources, together with the person who will later do the calculation, copies one to two years of each record that bears on the team, dates the copy and files it unaltered. The records are usually these: departures, with a note of which the organisation regretted; internal moves and promotions, with where each person came from; the dates on which new joiners first led a piece of work without supervision; sickness absence; and whatever the team's own work is measured by, its output or its quality. Nothing is decided about the copy yet. It is the history any later improvement will be read against, fixed before anyone could move it. Where the Flow Stops gives the reason in full: once the work has changed, the baseline cannot be recreated.
The second point is the pairing of measures. Each measure of development travels with a second one that would move if the first improved for the wrong reason. Fewer departures, beside the quality of the work, so that a manager cannot keep her people by asking little of them. More decisions made below the manager, beside how many were later reversed and at what cost. Faster time to independence for new joiners, beside the errors they made in their first year. Internal promotions out of the team, beside how well the people promoted did in their new roles. The second measure is not there because the manager is suspected. It is there so that nobody can later say the first was gamed.
The third is the decision log. Departures and promotions are slow, and an organisation that waits for them will wait a year for its first reading. The log of decisions kept by a team during its manager's absence, and kept again each time she is away, gives an earlier signal: which decisions moved, to whom, and how they held. It is the one record in this list that most organisations do not already keep, and it costs a team a line a decision.
The fourth is who calculates. A manager is not a disinterested reader of her own team's turnover, and nor is the human resources partner whose development programme she attended. The calculation is done, or checked, by someone whose standing does not depend on the answer, and the result is shown to the team in the same form the organisation sees it. This is the step most often skipped, and the one that most decides whether either side will believe the number.
The fifth is how to read it. Plot each measure against the ordinary range of its history and treat a point as news only when it falls outside that range, or when several in a row move the same way. One quiet quarter for departures is not yet a gain; people leave in seasons, and a team of ten has too few departures in any quarter for one to mean much. Read over two years, against two years of history, with the second measures beside them, the same figures can say a great deal.
There are mistakes worth naming. A baseline taken from the worst recent period, which flatters the first months and punishes every period after. An adjustment proposed after the results are known. A measure used alone. Internal moves out of the team counted as losses, which rewards the hoarding the canon warns of. And a calculation done privately and announced, which nobody on either side has reason to believe.
What this chapter does not settle is how a measured gain is then shared with the people who produced it: the proportions, the timing, the thresholds and the terms. Those belong to each organisation and the people it covers, and they are decided with them; for an organisation working with Luminous Prosperity, they are the substance of the engagement, not of a book. What the measurement can do is make that later decision honest, by producing a gain both sides already believe. Chapter 6 places the fixing of the history in the first ten days of a ninety-day plan and the measurement conversation in the last thirty, and Chapter 7 gives the records, the pairs and the one-page agreement.
It is half past five on a Friday, and the engineering manager has been back for four days. On the screen is the handover note from before her leave. She is rewriting it as a standing page, because the team asked for one: four projects, four names, and beside the difficult client the engineer's name, written without crossing out.
The log of fourteen decisions is pinned to the board by the door, where the team can see it. The pot plant on the windowsill has been watered.
She closes the note, and for the first time in three years she does not take the laptop home.
Chapter Six
Five practices set in order, from reading the week to measuring what the team can do alone.
The First Ninety Days
Haute Lumière · The Manager's Week · Chapter Six
It is a quarter to eight on a Monday, and the head of a department of some sixty people has arrived before anyone else, which she does not usually do. She has six managers reporting to her, each with a team of their own, and a manager above her who sits on the executive. She has read the first five chapters of a book like this one, or lived them, and she has recognised her week in the first, the charge nurses in her own deputies in the second, herself in the glass room of the third, the corner of the screen in the fourth and, most uncomfortably, the blank line of the handover note in the fifth. She has a notebook, a coffee and her calendar for the week, which has thirty-eight entries in it.
She is a composite, drawn from the way these first weeks are described in the research the earlier chapters read and in the house's own practice, and no firm is meant. This chapter is written to her, and to anyone in her position: a person who manages, often a person who manages managers, and who wants to begin. It sets out the first ninety days in order: who is involved, what each step needs, roughly how long it takes, the decisions that change the result, the mistakes made at each step and how to see them early, how to read what comes back, and what to do on the first Monday. The working pages it refers to are gathered in Chapter 7.
Two things should be said before the plan. The first is that it can be misread as a programme, and it should not be. Ammanuel Santa Anna's canon, in the book on management that the earlier chapters have quoted most, warns against exactly this, in a sentence that a ninety-day plan should be honest enough to print:35 "The answer isn't a technique. It's not a set of best practices or a leadership framework you can implement in ninety days." What ninety days can do is establish five practices and show whether they are being kept. The orientation underneath them, the canon's view that a manager's work is to grow people who can do the work, takes longer, and the practices are how it begins. The second is that the order matters more than any single step. The chapters were written in the order a reader meets the problem. The ninety days run in the order the practices depend on each other.
The five practices in this book are the week read back (Chapter 1), the span counted (Chapter 2), the note upward (Chapter 3), the work returned (Chapter 4) and the team measured by what it does without its manager (Chapter 5). In the first ninety days they run in this sequence: read the week and count the span together, then change the week, then reshape the span, then carry knowledge up, then return the work, then measure and test.
The reasoning is simple. Every later practice spends hours the manager does not yet have. Returning work to a team takes a weekly review, the patience to let a first decision stand and the time to write down the answers that live in one head. Carrying knowledge up takes an hour a decision. A manager who tries either inside the week of Chapter 1 will find no room for it, and will conclude that the practice does not work. So the first change is to the week itself, and the first change to the week is to read it, because a manager who protects her mornings without having read her week will protect the wrong ones.
The span comes next because the week cannot be changed much while the structure keeps writing it. A manager holding seventy-four people, or six managers who each bring her every hard case, will find her protected mornings filled again within a month. The note upward follows once there is time to listen, and the returned work follows the note because the two share a skill: hearing what a team knows and treating it as information. The measurement is agreed last, because it measures what the other four have produced. Its first move is made on day ten, though: the history against which any change will be read is copied and fixed before anything changes, for the reason Chapter 5 gives.
The canon puts the cost of getting this wrong in four words that apply to every layer of the department: "Disempowerment is a lineage." Managers manage the way they were managed. A head of department who changes her own week, her own span and her own habit of answering will find that her six managers begin to change theirs, often before she has asked. One who asks them to change theirs while keeping her own will find they do exactly what she does.
Four roles carry the first ninety days, and they should be named before anything else is done.
The manager is the person whose week, span and team are being looked at. In this chapter she manages managers, so the practices run at two levels at once: in her own week, and in the weeks of the six people who report to her. Expect it to take three to four hours a week of her time beyond her ordinary work in the first month, most of it spent reading and talking, and less after that. Much of it is time she will recover from her own week within the month.
Her own manager is the person who wrote some of her hours without meaning to and who decides the structure above her. He does not need to do much. He needs to know what she is doing and why, to read the reading-back of her week in the first fortnight, and to reply to the notes upward when they come. A manager who begins without telling her own manager will find the first hours she protects quietly reclaimed from above.
A partner from finance or human resources holds the records the measurement will later read: departures and which were regretted, internal moves and promotions, absence, the dates on which new joiners first worked unsupervised, and whatever measures the team's own work. The partner's first job, before day eleven, is to copy and date one to two years of each. Later, the partner will either do the calculation or find someone with no stake in it who can.
The convenor keeps the record: the diary and its reading, the span count, the notes upward and their replies, the classes of work returned and the decisions made inside them, the log kept during the manager's absence. Often this is a deputy or one of the six managers. The convenor needs about an hour a week, and the one quality that matters is that people trust the convenor to write down what happened rather than what the manager would like to have happened.
One decision changes the result more than any other at this stage. It is whether the manager shows her reading-back to her own manager, including the hours he wrote. The canon's book on executive perception asks leaders to imagine the middle of their organisation, and the most useful single act a person in the middle can do is to make that imagining unnecessary, by showing the view.
The first ten days change nothing visible. They gather four inputs, and the manager gathers three of them herself.
The first is the week. For one ordinary week she keeps the diary of Chapter 1, in quarter hours, at the end of each block, with the three marks: who asked, who else could have done it, whether it moved one of her few written priorities. In the same week, on the same page, she keeps the list of Chapter 4: every question or decision brought to her, one line each, marked T or A. Keeping both on one page is easier than keeping them separately, because most of the As arrive as blocks in the diary.
The second is the span. In the same week she counts the four spans of Chapter 2 for her own position: the formal span, the daily span, the working span (the different people who brought her something that needed her), and the unheld span (the people she is responsible for and has had no working contact with in a month). For a head of department the formal span is six; the working span is usually much larger, because the managers' people find their way to her, and the count often shows it.
The third is the decisions of the last quarter that waited for her. With the convenor, she lists the decisions in her area that took longer than they should have, and for each one asks, privately and without blame, where it waited. Most managers who do this find that a good number waited in their own diary.
The fourth is the history. Before day eleven, the partner from finance or human resources copies out one to two years of each record that bears on the department's people and work, dates the copy and files it unaltered. Nothing is decided about it yet. It is the history the measurement conversation will read in the last month, fixed before any practice could move it.
At the end of the ten days the manager writes, on a single page, a first guess at where her capacity is going and where it is being held up: a sentence on the week, a sentence on the span, a sentence on the decisions that wait. It is labelled as a guess and kept for day ninety.
What goes wrong at this stage. The commonest mistake is to keep the diary for a day and call it a week; a single day is always unusual. The second is to start changing the week in the middle of reading it, which makes the reading meaningless. The third is to announce an initiative. A manager who tells her department she is starting a programme on capacity will find her week full of people asking about it. The early sign is a question in a corridor: what is this called? If people are asking what it is called, it has been named, and the manager should stop naming it.
Days eleven to thirty: changing the week
Haute Lumière · The Manager's Week · Chapter Six
In the third week the manager reads her week back, with the convenor, using the sorting of Chapter 7, and then she does four things in order.
She tells the department, in a sentence or two at a meeting she already holds, what the reading showed and what she is going to change. She protects two or three longer blocks a week for the work only her position can do, and says when she is reachable instead. She hands one recurring piece of work to a named person, with the authority to decide, chosen because the diary showed her presence added little. And she looks hard at the standing meetings: for each one, the question is when it last changed a decision, and what would be lost if it met half as often with a written note between.
On the same days she shows the reading-back to her own manager, with the hours he wrote marked plainly and without complaint, and asks him one question: whether there are any of those hours he would rather she spent differently. Most managers who do this are surprised by the answer.
The decision that changes the result is what she protects the blocks for. The temptation is to fill them with the work she enjoys or the work that is most overdue. The diary is the guide: the work that only her position can do, which in a head of department is usually the thinking that needs the whole department in view, the conversations with her own manager and her peers, and the conversations with whichever of her six managers is ready for something larger.
What goes wrong. She protects the blocks and then gives them up for the first meeting that asks. She cancels standing meetings without asking what they did, and loses the one that was quietly holding two teams together. She hands work to the strongest person rather than the one who would learn most. The early signs are visible in the diary itself, kept again for one week at the end of the month. If the protected blocks have been filled, the reason will be written beside them, and it is usually a request from above or a question from below that belonged somewhere else. Both are the subjects of the next steps.
How to read what comes back. By day thirty, the second diary week should show some longer blocks used for the work only she can do, fewer hours asked for by people who could have decided without her, and at least one standing meeting changed. It will not show a quiet week. It is not meant to.
Overlapping with the changed week, the manager turns to the span: first her own, then, with each of her six managers, theirs.
For her own, the count from the first ten days usually shows that the managers' people are reaching her directly, often because her managers are too stretched to hold them, or because she has been answering. For each of the six, the manager and the convenor run the same count in an hour: the four spans, the people outside the chart, and Gulick's three questions about how varied the work is, how much time the manager has, and how far apart the people are in space and hours. The arithmetic of Graicunas is used once, if at all, to show a director why a span of twelve people who work closely together is larger than it looks.
The count usually shows one of three things, and each has its own answer, as Chapter 2 sets out. A span too wide for the work, which is a structural decision for the manager's own manager. Relationships held by the wrong person, which she can often move herself. Or informal leaders already doing the work without time or authority, in which case the answer is to make the structure true: name the groups they already hold, give them protected time, and give them the decisions they are already being asked to take.
The decision that changes the result is who holds the newest person in each group. The newest person in any span is the one most likely to be unheld, and the one whose first months most decide whether she stays. Every new group drawn should have a named answer to that question.
What goes wrong. The count is taken from the chart. The new groups are drawn without asking the people in them, and split pairs who have worked together for years. Informal leaders are given groups without time. The conversation with the director becomes a request for headcount and is refused before the evidence is read. The early sign is that the people asked to hold groups say yes too quickly, which usually means they think they have no choice. The manager should ask each of them, privately, what they will stop doing to make the time.
How to read what comes back. By day fifty, each of the six managers should have a count of their own span on one page, and at least one structural change should have been made or proposed with the evidence attached.
By the second month there is usually a significant decision from above to carry down: a change to a process, a budget, a structure, a system. The manager uses the first one to begin the habit of Chapter 3, and asks each of her six managers to do the same with their teams.
She carries the decision down first, in person, in her own words: what was decided, why, what she thinks of it, and what she does not yet know. She asks the question on the card, what do you know about this that the people who decided it might not?, and listens. She sorts what comes back into forecasts, preferences and grievances, honours all three in the room and carries the forecasts up, within a week, on one page with three headings: what the team understood and accepts, what it will cost them, and what they know that the decision may not have accounted for. She asks for one thing, a reply. And she names the people who saw what the decision missed.
The decision that changes the result is speed. A forecast that arrives while the design of a change is still open is information. The same forecast a month later is a complaint. The week is the practical limit.
What goes wrong. The decision is carried down by email. Nothing is carried up because the decision is made. Everything is carried up, unsorted, and the director reads it as resistance. The director does not reply. The early sign of the last is plain, and the manager's answer is to send the note anyway for three weeks and then raise the habit, with the unanswered notes in hand, in her own next conversation with her manager. The canon's governance chapter describes an experiment in its own terms, and the same spirit fits here: "The commitment is not to succeed — it is to learn."
How to read what comes back. By day sixty, the convenor's record should hold two or three notes upward from the manager and some from her six managers, with their replies, and at least one decision from above that changed in its design because of what a note carried.
With the week changed and the span reshaped, there is room for the practice that most changes what the department can do without her. The manager takes the list of As from the first ten days and groups them into kinds. For each kind, at a meeting with the people who will hold it, she says the four things of Chapter 4: the decision is now theirs; here is the limit and what lies outside it; here is who to ask when a case is unclear; and a reasonable decision made inside the limit will be backed, including when it turns out to be wrong. She writes down the answers to the commonest technical questions on a page the team can correct. And she sets a short review at the end of each week for the hard cases.
Her six managers do the same with their own lists, and the manager asks each of them, at the end of the first month, which class of decision they returned and what happened to the first one they would have made differently.
The decision that changes the result is what the manager does with the first returned decision she would have made differently. If she reverses it, she has returned nothing. If she lets it stand where it can stand and takes it to the review, asking what it teaches about the limit, she has returned the work and the learning with it.
What goes wrong. A decision is returned without a limit, and the team either freezes or gives everything away. A decision is returned without the promise of backing, and the questions keep coming. The manager returns the questions she dislikes and keeps the ones she enjoys. Her own manager goes on bringing questions in a returned class directly to her, and undoes the design from above. And for the newest people the limit is set as wide as for the rest, which is the abandonment the canon warns of. The early signs are in the list, kept again for a week at the end of the period: if the As in a returned class have not fallen, something is keeping the risk where it was.
How to read what comes back. By day seventy the list should be shorter, the page of technical answers should have been corrected at least once by the team, and the review should have moved at least one limit.
Days sixty-one to ninety: measuring and testing
Haute Lumière · The Manager's Week · Chapter Six
In the last month two things happen, and the second depends on the first.
The first is the measurement conversation of Chapter 5. The partner from finance or human resources leads it. The manager, one or two of her six managers, someone who speaks for the people whose work is being measured, and whoever will do or check the calculation are in the room. The manager's own manager attends and says little. The conversation reads the history fixed before day eleven and agrees, on one page, the records that will be read (departures and which were regretted, internal moves and promotions, time to independence for new joiners, decisions made below the manager and how they held), the second measure that travels with each, the things that move the numbers for reasons nobody controls and how each will be allowed for, who calculates and who checks, and how the result will be shown to the department. It usually takes two meetings of ninety minutes.
The second is the test. Some time in the last month the manager takes a week away, or a fortnight if she can, with a handover written in draft three weeks before and shown to the people who will hold her decisions, limits on one page, one stated reason to be rung, and a log kept by her managers of every decision they made that they would normally have brought to her. She does not read her email. If leave is not possible in the period, a week in which she withdraws from every decision not on the list of her own is the next best thing, and is recorded as such.
The decision that changes the result is what she does with the log on her return. She reads it, takes the decisions she would have made differently to the next review, and reverses nothing that can stand. The log goes into the record beside the history, and is kept again each time she is away.
What goes wrong. The baseline is taken from a bad recent period, because it makes the gain look larger. An adjustment is proposed after results are known. A single measure is used alone. Internal moves out of the department are counted as losses. The calculation is done privately and announced. And the manager reads her email during the week away, which tests nothing. The early sign of most of these is the same: someone in the measurement conversation goes quiet. The manager's job at that step is to notice who has stopped speaking.
What this step does not settle is how a measured gain is then shared with the people who produced it: the proportions, the timing, the thresholds and the terms. Those depend on the organisation, its people and its history; they are decided with the people who will be covered, and for an organisation working with Luminous Prosperity they are the substance of the engagement, not of a book. What the first ninety days can do is make that later decision honest, by producing a gain both sides already believe.
The manager and the convenor meet on day ninety with the first-guess page from day ten, the two diary weeks, the span counts, the notes upward and their replies, the list of returned work and the decisions made inside it, the log of the week away, and the signed measurement page.
They ask a short list of questions, in order. Where was the first guess right, and where wrong? How much of the week is now spent on the work only the position can do, and what filled the protected blocks when they were filled? Which spans changed, and who now holds the newest people? Which decisions from above changed because a note carried something up in time? Which classes of decision now sit with the team, and how did the first ones hold? What did the department do during the week away, and what did the manager learn from the log? And then the question that matters most for the next ninety days: which of the six managers have begun to change their own weeks, and which have not?
What not to conclude is as important. Ninety days are too short to say the department's capacity has changed; they are long enough to say whether the practices are being kept. Departures and promotions move slowly, and one quiet quarter is not yet a gain. And a department that ran perfectly during the week away may have had an unusually quiet week, which is why the log is kept every time.
The canon asks a manager a question in its introduction that fits day ninety better than any measure: "Can you embrace the mastery of becoming increasingly unnecessary?" Ninety days will not answer it. They will show whether the answer has begun to change.
If the whole of this chapter is too much to begin with, three things can be done on the first Monday, and none of them needs anyone's permission.
Keep a diary of today, at the end of each block, with one mark beside each entry: who asked for it.
Each time someone brings you a question today, write it down, and mark whether they could have answered it themselves if they had known they were allowed to.
And look at your calendar for next week and find one block, of two hours or more, for the work only your position can do. Write one word in it, and tell the people who would otherwise fill it when you will be reachable instead.
The canon's book on management ends its opening with a sentence that is uncomfortable to read on a Monday, because it removes the comfort of thinking that a busy week is a neutral one: "Every day, in every interaction, you are either growing people or you are not." The ninety days are a way of finding out which.
It is a quarter to eight on a Monday, and the head of department is not the first in; two of her managers are already at their desks. Her calendar has twenty-nine entries. Two mornings have a single word in them. The standing meeting on Wednesdays now happens on alternate Wednesdays, with a page between. On the shared drive there is a page of answers that her team has corrected eleven times, and in the convenor's folder a log of nineteen decisions from the week she was away.
None of these has a name. Nobody has announced anything. She opens the notebook, writes what she is least sure about this week, and goes to find out.
Chapter Seven
Fifteen working pages, the diary, the counts, the note upward, the returned decision and the agreement, ready to carry into a week.
The Field Manual
Haute Lumière · The Manager's Week · Chapter Seven
The chapters of this book describe five practices as people doing them.36 This chapter holds the working pages: the diary, the counts, the question card, the one-page note, the lists and the agreements that the earlier chapters point to and do not print. Each page is meant to be photocopied, filled in by hand and kept in a folder that the convenor of Chapter 6 looks after. Nothing here needs software.
The pages follow the order of the first ninety days, not the order of the chapters. Each opens with a line saying where it comes from and when it is used. Where a page gives a script, the words are a starting point; say them in your own voice. Where it gives a time, the time is a planning guide from practice, not a finding.
Two rules govern the whole manual. Write down what happened, not what you wish had happened. A diary kept to look good, a span counted from the chart and a log tidied before it is read all teach the opposite of what the practice is for. And answer everything you collect: a note upward with no reply, a returned decision reversed, a log read and not discussed, each tells people their knowledge stops where it was written down.
What these pages do not contain is the design of any arrangement for sharing a gain: proportions, thresholds, timing or terms. Those belong to each organisation and the people it covers, and they are decided separately. The canon's book on management is plain about what pages like these cannot do on their own: "The tools without the inner work are shells." Use them as a way of seeing, not as a substitute for it.
From Chapter 1. Kept for one ordinary week in days one to ten, and again for one week at the end of each month.
Write each entry at the end of the block it describes, not at the end of the day. A block is a quarter of an hour, or one meeting, or one unplanned conversation. Record the conversations nobody booked as well as the ones in the calendar.
Week of:
My two or three written priorities this quarter:
| Day and time | What the block actually held | Who asked for it (me · team · my manager · peer · outside) | Who else could have done it | Moved a priority? (yes / no) |
|---|---|---|---|---|
Reading it. Do not read it until Friday. A diary read on Tuesday changes Wednesday, and then the week is no longer ordinary.
From Chapter 1. Used on the Friday of each diary week, about an hour, with the convenor if you wish.
Count the blocks in each group and write the totals.
| Question | Count of blocks | What it suggests |
|---|---|---|
| Asked for by me | ||
| Asked for by my team | ||
| Asked for by my own manager | ||
| Asked for by peers | ||
| Asked for by people outside | ||
| Could have been done by someone else | ||
| …and that person would have learned from it | ||
| Moved one of my written priorities | ||
| Longest unbroken block spent on a priority (minutes) |
Then answer four questions underneath.
Reading it. Expect most hours to have been asked for by other people. That is ordinary. The question is whether the hours that only your position can fill are among them. Show this page to your own manager, with his hours marked plainly and without complaint.
From Chapter 1. Used once in days eleven to thirty.
| Meeting | How often, how long | Who attends | Last decision it changed, and when | What would be lost at half the frequency, with a note between | Keep · halve · change · end |
|---|---|---|---|---|---|
Watch for. A meeting that changes no decisions but holds two teams together. It is doing work; change its form, not its existence. Ask the people who attend before you end anything.
From Chapter 2. Used in days one to ten for your own position, and in days twenty-one to fifty with each manager who reports to you.
Position:
Week of:
| Span | How it is counted | Number |
|---|---|---|
| Formal span | Names on the chart who report to this position | |
| Daily span | Of those, how many are at work while you are, on an ordinary day | |
| Working span | Different people who brought you something needing a decision or a conversation this week | |
| Unheld span | People you are responsible for with no working contact in the last month | |
| Off-chart relationships | Contractors, people on dotted lines, informal leads, peers whose work is the other half of yours |
Gulick's three questions. Answer each in a sentence.
Reading it. If the unheld span is above a handful, write the names, and beside each one the name of the person who in practice holds them now, if anyone does. Those names are usually the informal leaders. If the working span is much larger than the formal span, people are reaching past someone to get to you, and the next question is why.
5 · The arithmetic, for one conversation
Haute Lumière · The Manager's Week · Chapter Seven
From Chapter 2, after Graicunas (1933). Used only to show a director why load grows faster than headcount.
For a manager with n direct reports, Graicunas counted the possible relationships as:
n × (2 to the power n, halved, + n − 1)
| Direct reports | Relationships (Graicunas) |
|---|---|
| 2 | 6 |
| 3 | 18 |
| 4 | 44 |
| 5 | 100 |
| 6 | 222 |
What it does not tell you. The right span for any position. Most possible groups never form, and nobody holds these numbers consciously. Its use is to make one point: a team that works closely together asks more of its manager than its headcount suggests.
From Chapter 2. Used in days twenty-one to fifty, once the counts are done.
Tick the pattern the count shows, then write the proposal in three sentences.
| Pattern | Answer | Whose decision |
|---|---|---|
| The span is too wide for the work | Another manager, or a split | Your own manager |
| Relationships are held by the wrong person | Move them | Often yours |
| Informal leaders are already doing the work | Name the groups they hold; give protected time and the decisions they already take | Yours, with your manager's agreement |
For each new group.
| Group lead | People in the group | Protected time per week | Decisions the lead now takes | Who holds the newest person |
|---|---|---|---|---|
Before you propose it. Ask each person who will lead a group, privately, what they will stop doing to make the time. Ask the people inside the old structure who works well with whom. Do not split a pair who have held a shift together for years without asking them first.
From Chapter 3. Used every time a significant decision is carried down, from day thirty-one.
Carry it down first, in person. Say four things in your own words: what was decided; why; what you think of it; what you do not yet know.
Then ask the card. Write it once and ask it the same way each time:
What do you know about this that the people who decided it might not?
Sort what you hear. Put each answer in one column while you listen or straight afterwards.
| Forecast (what will happen) | Preference (what I would have done instead) | Grievance (what this has cost us before) |
|---|---|---|
Forecasts go up as evidence. Preferences are noted and honoured in the room. Grievances are separated into their forecast, which goes up, and their history, which is a conversation for another day.
From Chapter 3. Sent within a week of carrying a decision down.
To:
Decision:
Carried to the team on:
1 · What the team understood and accepts.
2 · What it will cost them, said plainly.
3 · What they know that the decision may not have accounted for. Give enough detail that someone designing the change could act on it, and name the people who saw it, with their permission.
I am asking for one thing: a reply.
Mistakes to watch for. Sending it by email without the conversation first. Carrying the grief up instead of the forecasts. Asking for a reversal. Writing it in the language of the slides. Stopping after one unanswered note: send it for three weeks, then take the unanswered notes to your own manager as the subject of a conversation.
From Chapter 4. Kept for one ordinary week in days one to ten, on the same page as the diary if you wish, and again for a week at the end of days forty-one to seventy.
| Day and time | What was brought to me | Who brought it | How it arrived (in person · message · meeting) | T or A |
|---|---|---|---|---|
T means technical: the knowledge exists, and the decision is genuinely mine, by role or because only I hold the knowledge. A means the team has the knowledge and lacks the authority, the limit or the confidence.
Reading it. Group the As into kinds. Most lists fall into three or four. The kinds are what you return. Then look for the As you have been keeping because you enjoy answering them; there are usually one or two. For the Ts that recur, write the answers on a shared page with the date and give it to the team to correct.
10 · Returning a class of decision
Haute Lumière · The Manager's Week · Chapter Seven
From Chapter 4. Used once for each kind of decision returned, from day forty-one, at a meeting with the people who will hold it.
The class of decision:
Say four things, in this order, every time.
For the newest people. Write a narrower limit, or the smaller form of the decision: being asked what they would do before being told.
The weekly review (twenty to thirty minutes).
| Hard case | Decision made | Would I have decided differently? | What it tells us about the limit | Change to the limit, if any |
|---|---|---|---|---|
Rules. Never reverse a returned decision that can stand. Ask what a case teaches about the limit, never who got it wrong. Widen the limit at each review while it holds. Tell your own manager which classes you have returned, so that he does not undo the design from above.
From Chapter 5. Kept from day forty-one and read every quarter.
| Person | Joined the team | Stretch pieces of work given this quarter | Decisions they now hold | Conversation about what they want next (date) | Moved to a larger role (date, where) |
|---|---|---|---|---|---|
Reading it. Look at the third column across the whole team. If the stretch pieces have gone to the same few people, write down why, honestly. Treat a move to a larger role elsewhere in the organisation as a credit, not a loss.
From Chapter 5. Prepared three weeks before a period of leave, and read on return.
Three weeks before. Write the handover in draft and show it to the team, blank lines and all.
| Piece of work | Who holds it while I am away | What they may decide alone | What needs someone else, and who | Hours together before I go |
|---|---|---|---|---|
The one reason to ring me:
The log, kept by the team while I am away. One line for every decision that would normally have come to me.
| Date | Decision | Who made it | Why |
|---|---|---|---|
On return. Read the log in full. Take the decisions you would have made differently to the next review. Reverse nothing that can stand. File the log beside the history in the measurement folder, and keep it again each time you are away.
Mistakes to watch for. Reading email. Choosing the holder of the hardest piece of work by who is already strongest. Reversing the first decision in the log.
From Chapter 5. The history it reads is copied and dated before day eleven; the conversation is held in days sixty-one to ninety.
Who is in the room. The person who holds the records (finance or human resources). The manager. Someone who speaks for the people whose work is measured. The person who will do the calculation and, if that person reports to one of the others, someone who will check it.
The questions, in order.
The pairs.
| Measure | Record it comes from | Travels with | Record it comes from |
|---|---|---|---|
| Regretted departures | Payroll and exit records | Quality of the team's work | |
| Internal moves and promotions out of the team | Personnel system | How the people promoted did in their new roles | |
| Time for a new joiner to lead work unsupervised | Dates of first unsupervised work | Errors in the first year | |
| Decisions made below the manager | The decision log (page 12) | Decisions later reversed, and their cost |
The one-page agreement. Fill it in together and sign it before anything changes, or, if it has begun, from today.
| What is changing | |
| Measures and their records | |
| Second measures and their records | |
| History used, and the ordinary range it shows | |
| Things nobody controls, and how each is allowed for | |
| Who calculates; who checks | |
| How and how often the result is shown, and to whom | |
| What happens if a record or system changes | |
| Date agreed | |
| Signed |
Reading the figures. Plot each against the ordinary range of the history. Treat a point as news only when it falls outside the range or when several in a row move the same way. One quiet quarter is not yet a gain.
From Chapter 6.
| Days | Practice | Who | Time it takes | What it produces | Early sign it is going wrong |
|---|---|---|---|---|---|
| 1–10 | Read: the diary, the escalation list, the span count, the decisions that waited; fix a dated copy of the history (pages 1, 4, 9) | Manager; convenor; records partner | About an hour a day for a week | A first-guess page on where capacity goes | People asking what the new initiative is called |
| 11–30 | Change the week: tell the team, protect blocks, hand one piece of work on, review standing meetings; show the reading to your own manager (pages 2, 3) | Manager | Two hours to read; one meeting | Longer blocks used for the work only the position can do | Protected blocks filled again within a fortnight |
| 21–50 | Reshape the span, your own and each manager's (pages 4 to 6) | Manager; each manager below | An hour per count | A structure that matches the week; a named holder for each newest person | People saying yes to leading a group too quickly |
| 31–60 | Carry knowledge up (pages 7, 8) | Manager; each manager below | An hour per decision | Notes upward, replies, a design that changed | Notes unanswered |
| 41–70 | Return the work (pages 9, 10) | Manager; each manager below | One meeting per class; a weekly review | Classes of decision held by the team | The As in a returned class do not fall |
| 61–90 | Measure and test (pages 11 to 13) | Records partner leads; manager; team; calculator | Two meetings of 90 minutes; a week away | A signed agreement; a decision log | Someone in the room goes quiet; email read during leave |
From Chapter 6. A meeting of the manager and the convenor, about two hours.
Bring the first-guess page, both diary weeks, the span counts, the notes upward and their replies, the escalation lists, the development ledger, the log of the week away and the signed agreement. Ask, in order:
What not to conclude. That capacity has changed (ninety days show whether practices are kept). That one quiet quarter of departures is a gain. That a smooth week away proves anything on its own.
For anyone who wants to begin with less, on the first Monday.
A manager's week is mostly written by other people. Read back, it shows where an organisation's capacity goes, and how much of it could be given back.
Luminous Prosperity Inc. takes this work into organisations as a gainshare. The flow is opened from the inside, with the people who already hold what the firm knows. The gain is counted from records the firm already keeps, against a measure both sides sign before anyone knows the answer, and the house is paid as a share of what is found: between thirty and ten per cent of it.
The field manual carries one manager through one week. Reading the weeks of a whole layer of managers, and counting what returning the work is worth, is done in the building. That begins with a conversation.