The Line Came Back in November
Haute Lumière · Change That Stays · Chapter One
Chapter One
An improvement fades in the months after it is declared done, and those months are a stage of the work with their own measure and their own owner.
The Line Came Back in November
Haute Lumière · Change That Stays · Chapter One
It is ten past four on the last Friday in November, and the outpatients office on the second floor has emptied early. The clinic manager has the monthly report open on her screen and a mug of tea going cold beside the keyboard. The radiator under the window ticks. She is looking for one line, the rate at which patients miss their appointments, and she finds it on the third page. In March it sat near the top of the chart. By June, after four months of reminder calls, a rebooking desk and a new letter, it had fallen to the bottom third, and the project team had cake in the seminar room and a thank-you email from the director. She scrolls back through the summer. July, a little higher. September, higher again. November is where March was.
Nobody stopped the work. Nobody decided to go back. The reminder calls still happen, most days. The rebooking desk still exists, though it is staffed now by whoever is free. Her shoulders, which have been tight since the second page, know before she does that she will be asked about this at the next performance meeting, and that the honest answer is that she does not know when it went. She is a composite, drawn from the way these months are described in the research below and in the house's own practice, and no hospital is meant.
Ammanuel Santa Anna's canon has a passage for the moment she is in, and it is unusually exact about how quietly it happens.1 Writing about the stage of any organisational change at which the new way has to come home and live in ordinary days, it says:
The return threshold is where many transformations die — not dramatically, but quietly. The new insights are gradually forgotten. The old patterns reassert themselves. The boon, instead of transforming the daily life of the organization, becomes a nostalgic memory: "Remember when we..." The practitioner who is attentive to the return threshold knows that this is where the most patient and persistent work is needed — not the dramatic interventions of the departure phase, but the steady, unglamorous practice of embodying the new in every small decision and interaction.
This chapter makes one claim. Most improvements that fade were not rejected; they were left unread. The months after a project is declared done are a stage of the work in their own right, with their own measure and their own owner, and an organisation that treats them as the end of the work will keep finding its best results back where they started.
Every manager who has sat through a change course has met the three words. Unfreeze, change, refreeze. The model is usually attributed to the social psychologist Kurt Lewin and to a paper he published in 1947, Frontiers in Group Dynamics, the first article in the first issue of the journal Human Relations.2 Its appeal is obvious. It says that people and groups settle into a way of working, that the settled way must be loosened before anything new can take hold, and that the new way must then be made settled in its turn. The third word is the one the clinic manager needed. Refreezing is the step at which a change stops depending on the effort that introduced it.
The house's Luminous 100, its reading of the frameworks every consultancy sells, keeps the model and changes the metaphor. Ice is the wrong picture of an organisation, it says, because "organizations aren't frozen objects that need thawing but living systems in developmental homeostasis". And it replaces the violence of breaking the status quo with three gentler questions: "What is already softening? What is ready to flow? What new form is the living system yearning to take?" A living system does not stay where it is put. It keeps returning to the balance its forces hold it in, and the work of a change is to move the forces, not only the position.
The scholarship on Lewin has, in the last twenty years, moved in an unexpected direction. In 2016 Stephen Cummings, Todd Bridgman and Kenneth Brown compared what Lewin actually wrote about changing in three steps with how later textbooks presented it.3 Their conclusion was that he never developed such a model. The terms appear, briefly, in a long paper about group dynamics and field theory, and the tidy three-box diagram took form after his death, in the hands of the writers who built a new field called change management on top of it. Twelve years earlier Bernard Burnes had mounted a careful defence of Lewin's wider approach against its usual critics, who said it assumed organisations were stable, ignored power and suited only small changes.4 Burnes argued that the three steps were one part of a body of work that also included field theory, group dynamics and action research, and that the whole was still useful.
Both readings matter to a clinic manager on a Friday afternoon, for the same reason. Lewin's concern was that a group's level of work is held where it is by forces in balance, and that a change made without altering those forces will drift back.5 That is not a step at the end of a project. It is a description of what happens to every project, all the time, and it is the part of his thinking the three boxes least convey. A box called refreeze invites a team to do something once, at the end, and then stop. The forces that pulled the missed-appointment rate back up in July did not stop.
The canon has a sentence that keeps both halves of the matter in view, and it is a useful corrective to the way change courses talk about stability, as if it were the enemy. Living systems, it says, balance two imperatives, keeping a stable identity and adapting to new conditions: "Too much homeostasis becomes rigidity; too much evolution becomes chaos. The gift is in the dynamic balance." The clinic did not need less stability in July. It needed the new arrangement to become the thing its stability defended. A change has held when the organisation's ordinary pull towards balance works for it rather than against it.
A number nobody can find
Haute Lumière · Change That Stays · Chapter One
Most writing about change begins with a figure. Seventy per cent of change initiatives fail, it says, and the figure is attributed to consultancies, to business schools, to "the research". It appears in board papers, in proposals and in the opening slides of almost every change course. It is a useful figure for anyone selling help, because it makes failure the normal case.
In 2011 Mark Hughes went looking for its source.6 He reviewed five separate published instances in which a seventy per cent failure rate for organisational change was stated, traced each to its evidence, and found the same thing in every case: an absence of valid and reliable empirical evidence for the number. His conclusion was plain. A popular narrative of seventy per cent failure exists, and there is no valid and reliable empirical evidence to support it. He added that the idea of an inherent failure rate is itself hard to defend, because change is ambiguous, depends on context, is judged differently by different people, unfolds over time and is often not measured at all.
This chapter does not use the figure, and it is worth saying why the absence helps. A leader who believes that seventy per cent of changes fail will read the November chart as fate. One who knows that nobody has counted will read it as information about this change, in this clinic, in these months. The second reading is the useful one, because it asks what happened in July.
The figure survives for a reason that has little to do with evidence. It is comfortable. A change that fades inside a world where most changes fail is not anyone's fault, and a leader who has quoted the number at the start of a programme has, without meaning to, already written its excuse. The honest alternative is less comfortable and more useful. Nobody knows how often changes fail in general, but any organisation can know how its own last three changes fared, if somebody looks, and that history is a far better guide to the fourth.
There is a smaller and more careful literature on what happens after a change is introduced, and it is mostly written by people in health care, where a programme that fades can be measured in missed diagnoses and longer waits.
In 2005 a team led by David Buchanan reviewed what was then known about sustaining organisational change, and opened with a question the clinic manager could have written: why do some changes persist, while others decay?7 They defined sustainability as the process through which new working methods, performance goals and improvement trajectories are maintained for a period appropriate to the context. They noted how little attention it had received, and how often stability was treated in management writing as a symptom of inertia, a problem to be solved, rather than as an achievement. And they found that sustainability depended on factors at almost every level an organisation has: the change itself, individuals, managers, money, leadership, culture, politics, process, context and time. Their honest finding was that the relative weight of those factors could not be known in advance.
Read the clinic's year against their list and most of it is there. The change itself was simple enough to keep. The individuals who ran the rebooking desk were moved. The manager's attention went to winter. The money for the analyst's time ended with the project. Nobody would call any of those a decision about the missed-appointment rate, and together they decided it. That is what makes the review useful to a practitioner even though it cannot rank its factors: it is a list of the forces to watch, and most of them move on the day a project is declared finished.
The same year Mary Ann Scheirer reviewed nineteen empirical studies of health programmes in the United States and Canada.8 In fourteen of the seventeen studies that looked at whether programme activities continued, at least sixty per cent of sites reported keeping at least one component going. That is a finding about partial survival, not about success, and Scheirer was clear that the studies' methods had substantial limits. Seven years later Shannon Wiltsey Stirman and colleagues identified 125 studies of sustainability.9 About half relied on people's own reports, few used rigorous methods, and among those that did, only a small number reported that a programme had been fully sustained.
The grade on this evidence is modest and should be said. Most of the studies are retrospective and naturalistic, many measure whether an activity is still done rather than whether its benefit has lasted, and almost none can say why one site held and another did not. What they agree on is the shape the clinic manager is looking at. Changes rarely vanish. They thin. A component survives, the reminder calls go on most days, and the result drifts back while everyone can truthfully say the new way is still in place.
If the decay is not a decision, it has to be a mechanism, and the mechanism is ordinary. Consider what changed at the clinic in June. The project team, which had met every Tuesday, stopped meeting. The analyst who had produced the weekly chart went back to her other work, and the chart became a line on the third page of a monthly report. The two administrators who had run the rebooking desk were thanked and returned to their posts, and the desk was given to whoever was free. The director's attention, which had been on the missed-appointment rate since February, moved to the winter plan. Nothing in that list is a mistake. Every item is what an organisation does when a project ends.
John Kotter, writing about why transformation efforts fail, named one of the commonest errors as declaring victory too soon.10 The error he described was celebration taken as completion, the moment at which the energy that made a change possible is released before the change can stand without it. The cake in June was deserved. The problem was what the cake meant: that the work was over.
Santa Anna's canon describes the same release from the other side, as a habit an organisation can name about itself once it is willing to. One of its examples of a recurring pattern, the kind a leadership team can say aloud, reads like the clinic's year: "We have a pattern of launching initiatives with great enthusiasm and then quietly abandoning them when the next shiny thing appears." Nobody at the clinic would say they abandoned the missed-appointment work. They would say they moved on to the winter plan. The canon's point is that the two descriptions can be the same event.
And it says what a change looks like when the attention has gone and the structures have stayed. Writing about why reorganisations so often leave an organisation feeling exactly as it did, it puts the result in a sentence: "The furniture has moved, but the emotional field hasn't shifted." The rebooking desk is furniture. The habit of calling the patient whose letter came back is a field, and it was held in place for four months by a Tuesday meeting, a weekly chart and a director who asked. Remove those and the old balance reasserts itself. The canon's account of the return gives the plain version: "The organization reverts. The old patterns reassert themselves."
There is a second, quieter reason the line came back, and it belongs to how the improvement was measured. The weekly chart in spring showed one number, the outcome. It did not show whether the calls were being made on the days they mattered, the day before and the morning of. When the calls thinned in August, the outcome took weeks to move, and by the time it did, the cause was a month old and nobody remembered it. A change that is measured only by its result is read too late to keep. How a measure becomes one that both sides can believe is the subject of [Where the Flow Stops, the house's book on finding the gain an organisation already holds]({{BOOK:orgdev-flow}}); the narrower point here is that a measure has to outlive the project that set it, or the project's gain has no witness.
The holding read
Haute Lumière · Change That Stays · Chapter One
What keeps a change is less dramatic than what introduced it, and the canon is right that it is the more patient work. It begins before the cake. The practice this chapter describes is called the holding read, and it has four parts.
The first is a decision about what done means. A project is not finished when the number reaches its target. It is finished when the people who introduced the change have stepped back and the number has held without them, for a period long enough to include an ordinary month, a busy one and a holiday. At the clinic that would have meant keeping the project open, in a lighter form, through the summer, and closing it in the autumn if the line had stayed where June left it.
The second is an owner who was not on the project team. This sounds like a small point and turns out to be the largest. A project lead who reads her own result for drift is reading her own trophy, and both she and everyone around her will hear any rise as a criticism. An owner from the ordinary line of management, the person who will run the clinic next year, reads it as the clinic's number. In the canon's terms she is the accountability partner, and its coaching questions are the ones she asks each month: "How will you know you're staying on track? What will be your indicators that the vision is staying alive in daily practice?"
The third is a pair of measures, read together. One is the result, the missed-appointment rate, plotted every month against its history. The other is the practice, a simple count of whether the thing that produced the result is still being done: the share of patients called the day before. The practice measure moves first. When it falls, the owner asks what changed that week, while the answer is still in people's memories.
The fourth is a rule for what counts as drift, agreed in advance so that nobody argues about one bad month. A single high point is noise. Several points in a row moving towards the old range is the line coming home, and it triggers a conversation about the system: what has changed in the conditions that held the improvement, whether the desk is staffed, whether the calls have been squeezed by something more urgent. It is never a conversation about who let it slip. The canon's prompt for a group's regular look at its own agreements is the right register for that meeting: "How are we doing with our agreements? Where are we strong? Where have we drifted?"
Picture how the first of those conversations might go, at the clinic, if the holding read had been in place. It is the second week of August. The owner, who runs outpatients and was not on the project, has the two lines in front of her, and the practice line has fallen for three weeks in a row: fewer patients called the day before. The result line has not moved yet. She asks the two administrators, the booking supervisor and the analyst for half an hour, and she asks one question first: what changed in the last month that made the calls harder to make? The answer takes four minutes to arrive. The rebooking desk was given to whoever was free when one of the administrators went on leave, and whoever was free did not know that the call list was printed each afternoon at three. Nobody had decided to stop calling. The list was in a tray nobody now checked. It is moved to the shared screen that afternoon, and the practice line recovers the following week. The result line never moves at all, which is the point, and also the reason the conversation is so easy to skip.
Three mistakes are common. The holding read becomes a report that nobody discusses, which is the November page again with a new name. The owner is chosen from the project team, out of gratitude. And drift is met with a relaunch, a new campaign with new posters, which treats a mechanism as a lapse of enthusiasm and produces another June. Chapter 7 carries the holding read as working pages, with the agreement, the chart, the drift rule and the questions for each monthly conversation, and Chapter 6 places it in the ninety days after done.
The practice also answers the reason the line came back in the clinic's case, which was that the forces holding the improvement were removed one by one, each for a good reason. A holding read makes each removal visible as it happens. The day the rebooking desk is given to whoever is free, the practice measure dips, and someone asks. Often the answer is to keep the change and adjust it: a smaller desk, a different hour, calls made by text for the patients who prefer it. A routine that the people who perform it are allowed to adjust, the subject of Chapter 3, lasts longer than one that can only be kept or abandoned, a finding the sustainability studies repeat.
The forces most likely to reverse a change are often the least visible ones. One is the hours an organisation spends rescuing what was never maintained, which Chapter 2 follows into a packaging plant. Another is the person who carried the work and has since moved on, whose absence, Chapter 5 argues, is a test a change can be prepared to pass.
The article this chapter began as ended at the clinic's second November. The book can go further, into the finer points a manager meets when she tries a holding read in her own organisation, and the first is choosing the period.
A holding period has to include the conditions most likely to undo the change, or it proves nothing. For the clinic those were the summer holidays, when the booking team ran short, and the winter, when every department's attention moves to beds. A change in a warehouse needs a peak season in its holding period; a change in finance needs a year-end. The practical rule is to name, at the start, the two or three ordinary pressures the change has not yet met, and to keep the holding read open until it has met each of them once. That usually means somewhere between six months and a year. It rarely means the eight weeks most project plans allow.
The second point is the practice measure, which is the half of the pair most often skipped. It should be the thing that produced the result, counted as directly as possible: the share of patients called the day before, not the number of staff trained in calling. It should come from records the organisation already keeps, or from a count so simple that it costs a few minutes a week, because a practice measure that needs a new report will be the first thing dropped in a busy month. And it should be shown beside the result, on the same page, so that nobody reads one without the other. When the two lines disagree, the disagreement is the most useful thing on the page. A practice line that holds while the result drifts says the world has changed around the change: a new group of patients, a new referral route. A result that holds while the practice line falls says the result is being held up by something else, and that the change itself may already be fading.
The third point is the drift rule, which should be agreed with finance or whoever keeps the numbers before anyone has seen a bad month. The rule that works best in practice is borrowed from the way statisticians read process charts: one point outside the range the history shows, or several points in a row on the same side, moving towards the old level. The exact numbers matter less than two features. The rule is written down before the data arrive, and it is applied the same way whether the news is good or bad. A drift rule invented in the month the line rises will always find that this month is an exception.
The fourth point is what the drift conversation asks. It is held within a fortnight of the rule being met, it is short, and it has four questions in a fixed order. What has changed in the last two months in the conditions that held the improvement: the people, the hours, the equipment, the instructions, the demand? Is the practice still being done as it was, and if not, what is in its way? Does the change need adjusting to fit the new conditions, rather than restoring? And who will do what by when? The order keeps the conversation on the system. A conversation that begins with who let this slip ends with a relaunch, and the canon's whole account of the return threshold is a warning against that cycle.
The fifth is honesty about what a holding line can show. A number that holds after a project closes has not proved that the project caused it. Some of the improvement may have come from a mild spring, or from a fall that would have happened anyway after an unusually bad winter, the effect statisticians call regression towards the mean, which Where the Flow Stops reads in its chapter on measurement. The holding read is not a test of cause. It is a watch on whether the new arrangement is still working, and its value is in the early warning, not in the proof.
The sixth is when to stop. A holding read ends when the change has become ordinary: when the practice is written into the role descriptions of the people who do it, when it appears in induction for new staff, when its cost sits in the budget like any other running cost, and when the practice line has stayed put through the pressures named at the start. At that point the result line joins the department's ordinary report. The holding read was a stage, and the stage is over.
There is a last, quieter use. Every holding read, finished or abandoned, leaves a record of what happened to a change after it was declared done. Kept together, those records become the organisation's own history of change, the kind of evidence Hughes found missing from the literature, and one any leader can build for herself. Chapter 7 begins with a page for it: the last three changes, what each was meant to do, what it is doing now, and when anyone last looked.
It is ten past four on the last Friday of the following November. The clinic manager has the monthly report open, and the missed-appointment rate is on the first page now, with a second line beneath it showing how many patients were called the day before. Both lines are a little ragged. Both are where June left them. In August the second line dipped for three weeks, when a new booking system went in and nobody had told it about the call list, and the owner noticed and asked, and the list was moved.
She closes the report. The radiator ticks. There was no cake this year, and nobody has asked her about the line.
It stayed.
Chapter Two
An organisation that rewards the rescue and cannot see the prevention will quietly spend the time its improvements need to last.
The Fire That Never Started
Haute Lumière · Change That Stays · Chapter Two
It is twenty to seven on a Monday in October, and the stand-up at the packaging plant is held where it is always held, beside the whiteboard at the end of line three, under the strip lights and the noise of the extract fans. Eleven people in hi-vis, two coffees, one clipboard. The plant manager opens with thanks. At ten past three on Saturday morning the filler on line three seized, and the night crew stripped it, found the cracked bearing housing, drove forty minutes to fetch a part from the sister site and had the line running again by half past seven. There is a round of applause, and it is deserved. The shift lead goes slightly red.
The maintenance planner stands at the back, by the fire door. Line two, beside them, has not stopped unplanned for eleven months. He knows why, because he wrote the schedule that takes it down every second Thursday for four hours, and he fought for those four hours in every production meeting since January. He also knows that in August, when orders were high, the Thursday stops on line three were cancelled twice to make the numbers, and that the bearing housing on line three was on the list for the second of them. His jaw is set. Nobody mentions line two. Nobody has ever mentioned line two. He is a composite, drawn from the cases in the research below and from the house's own practice, and no plant is meant.
Ammanuel Santa Anna's canon describes the culture in that room without naming a factory, and it puts its finger on the story the applause tells.11 Writing about why so many organisations believe that real growth must hurt, it says:
For centuries, our dominant cultural narratives have linked growth to struggle, disruption, and the heroic overcoming of resistance. We celebrate the entrepreneur who slept on a factory floor, the athlete who trained through injury, the mystic who endured forty days of desert deprivation. These stories are not without truth — but they carry a hidden cost.
This chapter makes one claim. Improvements that depend on preventing problems fade when an organisation rewards the people who fix problems and cannot see the people who prevent them. The applause at twenty to seven is not a small matter of morale. It is a mechanism, and over a year it decides whether line two stays the way it is.
The clearest account of that mechanism comes from two system dynamicists at the Massachusetts Institute of Technology, Nelson Repenning and John Sterman, who spent a decade studying improvement programmes through more than a dozen in-depth case studies, in telecommunications, semiconductors, chemicals, oil, cars and recreational products.12 In 2001 they published the results in a paper with a title borrowed from one of their interviews: Nobody Ever Gets Credit for Fixing Problems That Never Happened.
Their starting point is almost too simple to state. The output of any process depends on two things: the time spent working, and the capability of the process the work is done with. When output falls short of what is wanted, there are two ways to close the gap. People can work harder, with overtime, more pressure and closer review. Or they can work smarter, by spending time on improvement: finding root causes, fixing equipment, training, maintenance. The two do not pay back in the same way. An hour of extra work produces an hour's extra output and then stops. An hour of improvement raises the output of every hour after it. In the authors' words, time spent on improving the capability of a process "typically yields the more enduring change."
The difficulty is time. Working harder pays immediately. Working smarter pays later, after a delay that can run to months, and in the meantime it costs output, because the hour spent on improvement is an hour not spent producing. Repenning and Sterman called the first pattern "better-before-worse" and the second "worse-before-better". A manager under pressure who tries both will see the first help this week and the second hurt this week, and will draw the obvious conclusion.
Then comes the step that turns a bad week into a bad decade. If the improvement time is cut to meet a target, capability does not fall at once; it erodes. Supervisors who defer preventive maintenance, the authors note, often enjoy "a grace period" in which output rises and costs fall, and only later, as the equipment wears, do yields and uptime drop. By then the cut is months old and nobody connects the two. The worse performance creates more pressure, the pressure cuts more improvement time, and the process slides into what they called the capability trap: a vicious cycle in which an organisation works ever harder on a process that is ever less able to do the work.
Walk the packaging plant's year through those loops and it reads like a calendar. In January the Thursday stops were agreed, and output on the lines that kept them dipped a little for a few weeks, as worse-before-better says it will. By spring line two had settled and was running cleaner than it had for years. In August orders rose, the production manager was short of output, and the quickest output on the site was the four hours a fortnight that line three spent stopped. Cancelling them was working harder, in the authors' sense, and it worked. September's figures were good. The bearing housing had been wearing all along, and the cancelled stop was the one on which a mechanic would have seen it. Nothing in September's figures said so.
Santa Anna's canon describes the same delay from the side of what it calls relational value, and the sentence fits a bearing housing as well as it fits trust: "The tragedy is subtle because the effects are delayed." The August decision to cancel two Thursday stops on line three was a good decision in August. Its cost arrived at ten past three on a Saturday in October, and when it arrived, it looked like bad luck.
The engineer's sentence
Haute Lumière · Change That Stays · Chapter Two
The title of the paper came from an engineer at a car company, who told the researchers: "Nobody ever gets credit for fixing problems that never happened." The sentence is funny and exact, and Repenning and Sterman placed it in the middle of the most uncomfortable part of their argument, which is about rewards.
As organisations grow more dependent on firefighting, they wrote, "they reward and promote those who, through heroic efforts, manage to save troubled projects or keep the line running." The learning, training and improvement work that would prevent the crisis in the first place goes unrewarded, because nobody can see a crisis that did not occur. Over time, senior management comes to consist of the people who were good at rescue, and they promote people like themselves. A project leader the researchers interviewed put it without embarrassment: "Our [company] culture rewards the heroes. Frankly, that's how I got where I've gotten."
The point is not that the night crew on line three should not have been thanked. They did a hard thing well at three in the morning. The point is what the stand-up teaches the eleven people standing in it about where reputations are made. Nobody at the plant will become production manager by keeping line two boring. Several people may become production manager by being very good on the nights when line three seizes, and the more often it seizes, the more chances they get.
Santa Anna's canon, writing about the people in any system who hold it together without being seen, names what such a system lacks: "Recognition of invisible labor. Systems that see and value the relational, emotional, and translational work that keystone positions entail — and that compensate and support this work rather than taking it for granted — create conditions in which keystones can sustain their contributions over time rather than burning out." The planner by the fire door is doing exactly that kind of work, in a technical register, and like the team leader of Chapter 5, whose weekly review vanished the month she was promoted, he is holding up something the organisation does not know it depends on. And the canon is plain about why it goes unseen. Of the steady practice that holds any system together, it says: "It is not glamorous. It is not heroic."
There is a darker turn in the research, and it explains why the trap is so hard to see from the inside. Managers caught in it, Repenning and Sterman found, tend to blame the people rather than the system. When improvement programmes stall because nobody has time to do them, the managers conclude that the workforce lacks discipline, increase monitoring, and so increase the pressure that caused the stall. In one product-development organisation they studied, the engineers' account of a failed programme was simple: "There just weren't enough hours in the day, and the work wasn't going to wait." The managers drew the opposite lesson from the same events. Faced with engineers who cut corners to meet their dates, they concluded that the engineers could not be trusted, and added monitoring, which added work, which produced more corners cut. The attribution confirmed itself. At the packaging plant the same loop is easy to imagine: a production manager who sees the planner's stops as a luxury, cancels them under pressure, sees a breakdown, and concludes that the maintenance team is not good enough to prevent breakdowns. One of the engineers summed up the new process in a line any planner would recognise: "The new process is a good one. Someday I'd like to work on a project that actually uses it."
The paper also describes two organisations that escaped, and the first is worth telling in some detail, because it shows both how the trap is broken and how quickly it closes again.
In 1991 an internal benchmarking study at Du Pont compared the company's maintenance record with the best performers in the chemicals industry.12 It found a paradox. Du Pont spent more on maintenance than the leaders and got less for it: 10 to 30 per cent more per dollar of plant value, with uptime some 10 to 15 per cent lower. Its mechanics worked more overtime, its spare-parts stores were overfull, and it still relied on expedited orders for critical parts. The pattern was the capability trap in a single page. Cost pressure had cut planned maintenance, breakdowns had risen, mechanics had been moved from planned work to repairs, and breakdowns had risen again.
A manager named Winston Ledet and a small team built a model of the problem, and then, realising that their own minds had changed because they had built it, turned it into a role-playing simulation they called the Manufacturing Game. The game let a room of mechanics and operations managers experience in a few hours what a plant experiences over months: that a planned-maintenance strategy makes costs rise and uptime fall before both improve. By the end of 1992 about twelve hundred people had played it. At the plants that put the programme in place by the end of 1993, the mean time between failures for pumps rose by an average of 12 per cent each time cumulative operating experience doubled, and direct maintenance costs fell by an average of 20 per cent. In twenty-three comparable plants that did not, the learning rate averaged 5 per cent, and costs rose.
These figures come from the company and its practitioners, reported by the researchers rather than independently audited, and they should be read as a well-documented case, not a trial. What makes the case useful is the sentence that follows the results. A member of the modelling team told the researchers: "As soon as you get the problems down, people will be taken away from the effort and the problems will go back up." And that, in part, is what happened. Mandated cost-cutting programmes led to downsizing across the company, weakening the very cycle of reinvestment that had produced the gain. The improvement was undone by the same logic that had created the need for it.
What broke the trap at Du Pont, in the authors' reading, was less the maintenance technique than a change in how managers explained their own problems. They quote a manager from one of the successful efforts who put the change in two sentences: "There are two theories. One says, 'there's a problem, let's fix it.' The other says 'we have a problem, someone is screwing up, let's go beat them up.'" The second theory is what the capability trap teaches. The first is what an organisation has to choose, deliberately, before any tool will help.
The second case, a BP refinery in Lima, Ohio, adds one detail that matters for anyone planning a change. When the refinery introduced the same learning laboratory in 1994, maintenance costs rose by 30 per cent in the first six months. Management, which had played the game, had been told this would happen, and did not abandon the programme. The worse-before-better dip, expected, became a sign the work had started. Unexpected, it would have looked like failure.
Two years before the Repenning and Sterman paper, the sociologist Leslie Perlow published a study of a software engineering team that shows the same mechanism at the scale of a single day.13 Over nine months she watched how the engineers used their time. They had too much to do and too little time to do it, a condition she called a time famine, and part of the cause was the team's own way of working: engineers were constantly interrupted by colleagues, and a crisis mentality, backed by a reward system built on individual heroics, kept the interruptions coming. When the way the team used its time was changed, its collective productivity improved.
Perlow's study is one team, observed closely, and it should be read as a vivid case rather than a general law. But it adds something the factory cases do not. The heroics were not only a response to crises. They were also a cause of the next ones, because each rescue interrupted the work that would have prevented the following crisis. A culture that rewards firefighting does not merely fail to reward prevention. It actively consumes the time prevention needs.
Santa Anna's canon describes the bodily side of that culture, and it is the side a leader can feel in her own week. Writing about how organisations hold time, it says: "Chronic urgency is a systemic sympathetic activator." The stand-up at twenty to seven is held at the speed of the fans. And the canon's passage on performance systems that run on activation explains why such a culture can look so productive for so long:
They generate productivity through activation — urgency, competition, anxiety about outcomes. This works in the short term. It produces measurable output. But it does so at the cost of the very capacities that values-based performance seeks to cultivate: creativity, relational depth, ethical sensitivity, and sustainable engagement.
The packaging plant produces measurable output in the weeks it cancels the Thursday stops. What it gives up is visible only on the nights it loses.
Counting what did not happen
Haute Lumière · Change That Stays · Chapter Two
The practice this chapter describes has two halves, and they answer the two halves of the trap: the time that is cut, and the credit that is never given.
The first half is the protected hour. Improvement and prevention time, the Thursday stops on line two, is set as a standing allocation, agreed with production in advance and treated as the last thing cut, not the first. When it has to be cut, as it sometimes will, the cut is decided by a named person, written down with the date and the reason, and given a date by which the time will be restored. That record turns an invisible erosion into a visible loan. This is a negotiation, and the production manager's side of it deserves to be stated fairly. She is measured on output this month, she has customers waiting, and four hours a fortnight on a line that is running well looks like four hours given away. She is not wrong that the stop costs output. She is wrong only about when the bill for skipping it arrives, and she is wrong because nothing she is shown tells her. The protected hour works when the person who gives up the output is shown the record of what the stops have found, and when her own targets allow for the stops instead of competing with them. Without that, the agreement lasts until the first busy month.
In October the planner would not have had to remember, privately, that the August stops on line three were cancelled. The record would have shown it, beside the date the bearing housing failed.
The protected hour comes with a warning, given out loud before the change begins, that results will get worse before they get better, and roughly for how long. The Lima refinery's managers held their nerve through a 30 per cent rise in costs because they had been told to expect one. A dip that has been named in advance arrives as an appointment rather than a verdict. The Cairn, another of the house's books, makes the same point about [an obstacle written down before it arrives]({{BOOK:rsd-rsd-and-self-confidence}}), and it holds for a plant as well as for a person.
The second half is the quiet ledger. Once a month, at the review where rescues are discussed, the team also reads a short list of things that did not happen and why. Line two ran eleven months without an unplanned stop. The new supplier's film jammed three times on trial and was rejected before it reached production. A mechanic noticed a hairline crack on a guard during a planned stop. Each line names the person or the practice responsible. The canon describes the same practice for the invisible work of holding a team together and calls it a naming ceremony, with an example sentence that could have been written for the planner: "I want to acknowledge that Marcus has been translating between engineering and design in ways that have prevented three conflicts this quarter." Prevented problems are hard to count, and the ledger does not pretend to count them exactly. What it does is make them part of the story an organisation tells about itself.
There are failure modes. The ledger becomes a list of claims that nobody checks, and people start writing down every breakdown they can imagine avoiding. The answer is to record only things with evidence: a stop that was planned and held, a defect found before it shipped, a part replaced on schedule and found worn. The protected hour becomes a ritual, held but not used, with the line taken down and nobody working on it. And leadership praises the ledger and goes on promoting the rescuers. That last is the decisive one, because promotions are the loudest message an organisation sends. Chapter 7 carries the protected hour and the quiet ledger as working pages, with the record of cuts and the monthly reading.
The practice changes what the stand-up is for. It still thanks the night crew, because they deserve it. It also, once a month, reads out line two. And it asks a question it has not asked before: of the rescues this month, which could have been prevented, and what would it have taken? In the canon's terms, it begins to treat prevention as a capacity worth investing in "even when individual heroism would be faster and easier."
The trap reaches beyond the factory floor, because the same arithmetic applies wherever an improvement needs time to keep. The missed-appointment work that slipped back at the outpatient clinic of Chapter 1 was undone by the same quiet reallocation of hours. And when the time is protected and the old way still returns, the cause often lies closer to the work, in the routine that people's hands still remember after the training has been forgotten, which is the subject of Chapter 3.
The article this chapter began as ended at the stand-up in March. The book can go further, into the finer points a manager meets when she tries to protect improvement time in her own operation, and the first is how much time to protect.
The honest starting point is the time the work already needs, not the time an improvement programme would like. For a maintenance planner that is the planned work the equipment requires; for an office it may be the hour a week the team spends on the problems that keep recurring. Begin there, protect it fully for a quarter, and expect the dip. Asking for more than the work needs, in the first quarter, invites the production manager to treat the whole allocation as negotiable.
The second point is the shape of the agreement. It works best when it is written as a standing fixture in the plan the operation already runs to, the production schedule, the rota, the shift plan, and not as a separate improvement plan that competes with it. A Thursday stop that is in the production schedule is a Thursday stop. A Thursday stop that is in the maintenance plan is a request, renewed every fortnight.
The third is the record of cuts, which is the heart of the practice. Each time protected time is given up, four things are written on a single page kept by the planner: the date, who decided, the reason, and the date by which the time will be restored. Nothing else. The page is shown at the monthly review beside the quiet ledger. Its purpose is not to shame anyone for a cut; some cuts are right. It is to make the loan visible, so that when a breakdown comes, the team can look back and see whether it followed a cut, and when, and the organisation can learn the delay between the two in its own plant. That delay is the number Repenning and Sterman's managers never knew, and it is the one that would have changed their decisions.
The fourth is the warning of the dip, which should be given with numbers wherever possible. After the first few weeks of a protected hour, the team will have a rough sense of what the stops cost in output. Write it down, show it to the people who will feel it, and say how long it is expected to last. When the dip comes, it will then be read against the forecast, not against the old normal. A dip within the forecast is the programme working. A dip beyond it is a question worth asking early.
The fifth is the evidence rule for the quiet ledger. Only things with a trace go on it: a stop that was planned and held, a part replaced and found worn, a defect caught before shipping, a complaint that did not recur because its cause was fixed. Each line names the person or the practice. A ledger of imagined disasters avoided is worse than none, because it teaches people to overclaim, and the first sceptical reader will dismiss the rest.
The sixth is the one that leadership must do itself, and it is about promotion. If the people who prevent problems are not, over a few years, among the people who are promoted, the ledger is decoration. A practical step is to add one question to every performance conversation for anyone whose work includes prevention: what did you prevent this year, and how do we know? Asked consistently, the question changes what people bring to the conversation. Within a year or two it begins to change who is seen as capable.
The seventh is a simple measure of whether the trap is loosening. In maintenance it is the share of work that was planned rather than reactive; in an office it is the share of the week spent on new work rather than rework, chasing and repair. It moves slowly. When it rises, the capability trap is opening. When it falls for several months in a row, something is being cut that the record of cuts has not caught.
And there will be genuine emergencies, in which the protected time has to go. The test of the practice is what happens after them. If the time is restored by the date on the record, the protected hour has survived its hardest case. If it is not, the next cut will be easier, and the one after that easier still, which is how the trap closes. Chapter 6 places the protected hour in the second month of the ninety days after done, and Chapter 7 gives the record of cuts and the quiet ledger as working pages.
It is twenty to seven on a Monday in March, and the stand-up is beside the whiteboard at the end of line three. The fans are as loud as ever. The plant manager opens with thanks to the afternoon crew for a clean changeover, and then reads, from a page clipped to the board, three lines. Line two: fourteen months without an unplanned stop. Line three: the Thursday stops have been held since November, and the second bearing housing was found worn on one of them and replaced. A mechanic: a crack in a guard, found on a planned stop.
Nobody applauds, quite. Somebody says nice one. The planner, at the back by the fire door, looks at the board.
On line three, nothing has happened all winter.
Chapter Three
A change that has been taught has changed the description of the work; the work itself answers to cues the training never touched.
What the Hands Still Remember
Haute Lumière · Change That Stays · Chapter Three
It is ten to nine on a Monday in November, and the claims office on the fifth floor smells of the weekend's carpet cleaning. The senior handler has been here eleven years. She has her coat on the back of the chair, a glass of water, and forty-one new claims in her queue from the weekend: two burst pipes, a dozen cars, the usual run of phones dropped in the usual places. In September the whole team spent a day learning a new routine for exactly this moment. Every new claim is to be read within the first hour and sorted into one of three tracks, and the simple ones get a short call to the claimant the same morning. Everyone agreed it was better. There is a laminated card with the three tracks on it propped against her monitor.
She opens the first claim. The screen opens, as it has for eleven years, on the assessment tab, and the queue is sorted, as it has been for eleven years, oldest first. Her hands do what they know. By ten she has assessed six claims in date order, thoroughly and well, and has called nobody. Her chest is a little tight in the way it always is on Mondays, and the routine she agreed with in September has not crossed her mind. 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 insurer is meant.
Ammanuel Santa Anna's canon has a passage for the gap between the card and her hands, and it is direct about where most changes are installed and where they actually live.14
This is why so many culture-change initiatives fail. They try to install a new culture — new values, new norms, new behaviors — as if culture were a software upgrade that could be loaded onto the organizational machine. But culture is not software. Culture is a living process, continuously self-creating through the daily interactions of the people within it. Installing new values on the lobby wall changes nothing if the patterns of interaction remain the same.
This chapter makes one claim. A change that has been written down, taught and agreed has changed the description of the work, and not yet the work. The work is performed by particular people at particular hours, in answer to cues that the training never touched, and a change that does not reach those cues will be performed the old way on the first busy morning.
For a long time organisational theory treated routines as the enemy of change. A routine was a fixed pattern, repeated, that kept an organisation doing what it had always done; changing an organisation meant breaking its routines and installing new ones. In 2003 Martha Feldman and Brian Pentland argued that this picture was wrong, or at least half wrong, in a paper that has since become one of the most cited in the field.15
Borrowing a distinction from the sociologist Bruno Latour, they described every routine as having two aspects. The ostensive aspect is the routine as an idea: the description, the procedure, what people would say the routine is if asked. The performative aspect is the routine as it is actually done, by specific people, at specific times and places. The two are related but never identical. The ostensive aspect guides performances and lets people account for them. The performances, in turn, create, maintain and modify the ostensive aspect. A routine changes when its performances change and the change is kept, and it stays the same when the performances keep reproducing the old pattern, whatever the procedure says.
Three years earlier Feldman had made the same point from the field, in a study of routines that changed as the people performing them responded to the results of earlier rounds.16 When the people doing a routine are seen as part of it, she wrote, routines turn out to hold a great deal of potential for change. They also turn out to hold a great deal of potential for staying the same, for the same reason. People adjust a routine to what happens to them while they perform it, and what happens to the senior handler at ten to nine on a Monday is forty-one claims and a screen that opens on the old tab.
Read the claims office through that distinction and the September day looks different. The training changed the ostensive routine, thoroughly. Everyone could describe the three tracks. The laminated card is the ostensive routine, propped against a monitor. The performative routine, the thing that happens when a claim arrives, was not touched at all, because the performance answers to things that the training did not change: the default tab, the sort order, the Monday volume, the handler's eleven years. The canon's account of how old patterns return names the moment it is most likely: "The old patterns will always return in moments of stress." Monday morning is the moment of stress the new routine was designed for, and it is also the moment the old routine is strongest.
What the hands answer to
Haute Lumière · Change That Stays · Chapter Three
The psychology of habit gives the same answer at the level of a single person, and it is more precise than the organisational version about why agreement is not enough.
In 2007 Wendy Wood and David Neal set out a model of how habits work and how they interact with what people intend.17 Habits, they argued, form through the slow learning of associations between a response and the features of the context in which it is usually performed: the place, the time, the preceding action. Once a habit is formed, seeing the context triggers the response directly, without passing through a goal. The habit associations accrue slowly and do not shift much with a person's current intentions or with the occasional exception. That is the sentence a change programme should read twice. A training day changes intentions. It does not change the associations between the assessment tab, the date-ordered queue and the handler's first move, which were built over eleven years of Mondays.
The model also explains why the usual remedy disappoints. When a new routine slips, the instinct is to remind people of it: another email, a refresher session, a second laminated card. A reminder works on the goal, and the goal was never the problem. The senior handler wants to sort the claims within the hour; she said so in September and she would say so now. What she meets at ten to nine is not a goal but a context, the same context that has cued the same first move several thousand times, and the reminder is not in it. Wood and Neal's own conclusion points the other way: the most promising route to changing a habit runs through the cues that trigger it.
How long the new associations take to form is a question with an honest and unhelpful answer. In a study published in 2010, Phillippa Lally and colleagues asked ninety-six volunteers to choose an eating, drinking or exercise behaviour and do it daily, in the same context, for twelve weeks, and to report each day how automatic it felt.18 For those whose data could be modelled, the time to reach most of the eventual automaticity ranged from eighteen days to 254. Missing one day did not materially set the process back. The study is small, the behaviours were chosen by the volunteers and simple, and automaticity was self-reported, so it is a guide to the shape of habit formation rather than to any office. The shape is the useful part. New habits form along a curve that flattens slowly, they need a consistent context to form at all, and the time varies so much between people that any programme assuming a fixed number of weeks will be wrong for many of them.
Santa Anna's canon, writing about how to work with patterns that live in the body as well as the mind, makes the practical inference in a single instruction: "Use environmental design (change the physical space, the routines, the social networks) to support new consciousness." In the claims office, the environment is the screen. The Tuning Fork, another of the house's books, makes [the argument that a body learns by repetition in small doses]({{BOOK:rsd-module-one-introduction-for-practitioners}}), and a team learns the same way, one Monday at a time, if the Mondays are arranged so that the new move is the easy one.
The clearest evidence that a written routine is not the same thing as a performed one comes from operating theatres, and it comes in a pair.
In 2009 a team led by Alex Haynes reported the results of the World Health Organization's surgical safety checklist in eight hospitals in eight cities, from Toronto to Ifakara.19 The checklist has nineteen items, designed to improve communication and consistency in the theatre. The team collected data on 3,733 consecutive patients before the checklist was introduced and 3,955 after. The rate of death fell from 1.5 per cent to 0.8 per cent, and inpatient complications from 11.0 per cent to 7.0 per cent. It is a before-and-after study, not a randomised trial, and the hospitals knew they were being watched. But the effect was large, and checklists were adopted across the world within a few years.
In 2014 David Urbach and colleagues reported what happened when the province of Ontario, in Canada, encouraged universal adoption.20 Using administrative data from 101 hospitals, they compared about 109,000 procedures in the three months before each hospital adopted the checklist with about 106,000 in the three months after. The adjusted risk of death moved from 0.71 per cent to 0.65 per cent, and complications from 3.86 per cent to 3.82 per cent. Neither change was statistically significant.
The two studies can be argued about at length, and they have been. Ontario's hospitals began from lower complication rates, the follow-up was short, and the data cannot show how the checklist was actually used in each theatre. The reading this chapter takes is narrower, and it is one the data allow. The same nineteen items, introduced with great care and attention in one setting and as a policy across a province in another, did not produce the same results. The artefact was the same. Whether the routine changed is a different question, and the paper alone cannot answer it.
Pentland and Feldman gave that difference a name in the title of a later paper, and the title is the argument: Designing Routines: On the Folly of Designing Artifacts, While Hoping for Patterns of Action.21 A checklist, a laminated card, a new screen, a procedure manual, are all artefacts. They can be designed. A pattern of action, the thing people actually do at ten to nine on a Monday, cannot be designed in the same way; it can only be influenced, and then watched, and then influenced again.
If the routine lives in its performances, then the question for anyone who wants a change to hold is what each performance answers to. In most offices the answer is a short list of ordinary things, and almost none of them is in the training.
The first is the default. The screen opens on the assessment tab because it always has; the queue sorts oldest first because someone set it that way in a previous system. Defaults are the strongest instruction in any workplace and the least discussed, because nobody experiences them as instructions. The second is the measure people are read against. If the weekly report still counts claims assessed, and not claims sorted within the hour, then the old routine is what the organisation is still asking for, in the only language it repeats every week. The third is the order of the day: what happens first, what comes after the stand-up, which task fills the gap before the first meeting. The fourth is who is praised, and for what. The fifth is the room: who sits beside whom, whose experience is in earshot when a new handler is unsure.
The canon's chapter on rituals explains why such ordinary things carry so much weight. Writing about the recurring practices that make a group feel safe, it says that rituals "create predictable containers for safety — recurring moments in the collective life where the nervous system can anticipate what is coming and, over time, begin to settle in advance of the experience itself." The old routine is such a container. Its predictability is part of why it feels right on a hard morning, and a new routine that asks a tired person to give up a predictable container for an unfamiliar one will lose, unless the new one is made predictable too: the same first move, at the same moment, every Monday, until it becomes the thing the body expects.
The sustainability research read in Chapter 1 adds one finding that bears directly on this. In Mary Ann Scheirer's review of programme sustainability, one of the factors that most consistently helped a programme last was that it could be modified over time.22 That fits the performative view exactly. A routine that the people doing it are allowed to adjust is a routine they are performing, and therefore keeping. A routine that can only be followed or broken will, under pressure, be broken.
There is a telling detail in most offices that try a change like this, and it is worth looking for. The handlers who joined after September tend to perform the new routine more readily than the experienced ones. It is not that they are more willing. They have no eleven years of Mondays; the triage view is the only Monday they know, once someone has shown them how to set it as their default. The experienced handlers carry the old cues in their hands, which is also why they are the best placed to say which cues need changing.
Walking the routine at its busiest hour
Haute Lumière · Change That Stays · Chapter Three
The practice this chapter describes is simple to state and takes patience to do. It is to walk the routine where it is performed, at the hour it is hardest, and to change what the performance answers to rather than what the procedure says.
It begins with the hour. Most changes are observed, if they are observed at all, on a quiet afternoon in the week after training, when the new routine is fresh and the volume is low. The useful observation is at the routine's worst moment: Monday at nine in the claims office, month-end in finance, shift change on a ward, the last hour before the van leaves in a warehouse. Someone sits beside the work, with permission and without a clipboard of judgements, and writes down what is actually done, step by step, and what each step was prompted by. Not what should have prompted it. What did.
Picture the first walk in the claims office, on the last Monday in November. The team leader sits beside the senior handler at ten to nine, with her agreement, and writes down what happens. The screen opens on the assessment tab, and she clicks into it. The queue is in date order, and she starts at the top. The team leader asks, after the third claim, what made her start at the top. She thinks for a moment and says that she has always started at the top, and that the triage view is two clicks away, and that on a Monday two clicks feels like a decision. Neither of them had thought of the clicks before. By ten the team leader has seven things written down, and only one of them is about the handler.
Then the list of cues. For each step in the old routine that came back, the walker and the person doing the work ask together what prompted it: the default, the sort order, the report, the habit of the person at the next desk. The person doing the work is the expert here, and what the person who does the work already knows is, as [Where the Flow Stops argues of a nurse who fetches linen from another ward]({{BOOK:orgdev-flow}}), usually the best information an organisation has and the least collected.
Then the changes, and they are made to the cues. The screen opens on the triage view. The queue sorts by track, not by date. The weekly report counts claims sorted within the hour, beside claims assessed. The Monday stand-up begins with the three tracks, so that the first move of the week is the new one. None of this is training. All of it is the environment the canon's instruction describes.
Then the adjustments, and they belong to the team. The handlers will find that the three tracks need a fourth for claims involving a vulnerable customer, or that the same-morning call works for cars and not for burst pipes. The canon's advice on complex change applies at this scale: "Run small experiments rather than large programs." Each adjustment is written down with its reason, so that the ostensive routine catches up with the performances instead of drifting away from them.
And then the walk is repeated, at the same hour, a month later and again two months after that, because the first walk shows where the old routine was strongest and the later walks show whether the new one has taken its place. The canon describes the moment that tells a team the change has held, and it is the right measure for the third walk: "The first time the organization faces a challenge of the kind that would previously have triggered the old patterns — and responds differently."
The mistakes are predictable. The walk is done once and on a quiet day. The cues are listed and the training is repeated anyway, because training is what a change programme knows how to buy. The team's adjustments are treated as deviations and corrected back to the procedure, which teaches the team that the routine is not theirs. And the time to walk and adjust is the first thing cut when the queue is long, which is the trap of Chapter 2, in another office. Chapter 7 carries the walk as a working page, with the cue list and the artefact inventory.
Sometimes the walk turns up a handler who did not forget the new routine at all. She chose the old one, because she believes the new one will fail the customers she worries about, and she may be right. That is a different matter from habit, and it calls for a different conversation, the one in which an objection is heard as information about what a change will break, which Chapter 4 describes.
The article this chapter began as ended at the claims office in February. The book can go further, into the finer points a manager meets when she walks a routine in her own organisation, and the first is choosing the hour.
The hour to walk is the one at which the old routine is most likely to return, and the people who do the work can always name it. Ask them. In a claims office it is Monday morning; in a warehouse, the last hour before the vans; in a hospital, shift change; in finance, the three days before month-end; in a restaurant, the first sitting on a Saturday. A walk at any other hour shows the routine as it is performed when nothing is pressing, which is the version the training already described.
The second point is the walker's stance. The walk is done with the agreement of the person whose work is watched, and it is explained beforehand, in plain words: the purpose is to find what the routine answers to, not to check whether she is following it. The walker writes down what happens and, where it is not obvious, asks what prompted a step, after the step and never during it. She does not correct anything during the walk, however tempting it is. A walk that turns into coaching produces the performance the walker wants to see and teaches the team that walks are inspections.
The third is the record. For each step, three columns: what was done, what prompted it, and whether it was the new routine or the old. The middle column is the one that matters, and it fills with ordinary things: the screen opened here, the list was in this order, she was asked a question by the person at the next desk, the report counts this. It helps to write the prompts in the person's own words.
The fourth is the inventory of artefacts, which turns the walk's middle column into a list of things that can be changed. The categories are nearly always the same: the defaults of any system the work passes through, the order in which work arrives or is listed, the forms and templates people fill in, the reports that count the work and the measures people are read against, the room and who sits where, the first move of the day or the shift, and what is praised. Chapter 7 gives the inventory as a page, with a column for who owns each artefact, because changing a default usually needs someone outside the team.
The fifth is the order of changes. Start with the cheapest cue that prompted the most old steps. Defaults and sort orders often cost minutes to change and remove the largest share of the old routine at a stroke. Reports and measures come next, because they take longer to agree and are read every week. The room and the order of the day come last, because they affect other teams. Train only after the cues have changed, and train briefly, because the work now teaches the routine.
The sixth is how to treat adaptations. The team will change the new routine as they perform it; Feldman's whole argument is that this is how routines live. Each adaptation is written down with its reason, and within a week the team leader decides with the team whether it becomes part of the routine. Most should. A few will undo what the change was for, and those are discussed openly, with the reason the change was made. What should never happen is an adaptation discovered months later and corrected without discussion, which is how a team learns to hide its improvements.
The seventh is using newcomers as informants. People who joined after the change have only ever performed the new routine, or have been taught the old one by colleagues without anyone noticing. Asking them, in their second or third week, what they were shown and by whom is one of the quickest ways to see which routine the team is actually passing on.
The eighth is that the walk sometimes shows the old routine was right. A step the new design removed may turn out to be the one that caught a fault the designers had not seen. When that happens, the honest response is to change the design and say so. A change that cannot learn from its own walk will lose the people performing it, and they will be right.
And the last is how to tell habit from choice. A person performing the old routine out of habit will usually say, when asked, that she meant to do the new one and does not know why she did not. A person who has chosen the old routine will usually say why, and the reason will often be specific. The first is a matter for cues, and this chapter. The second is a matter for the conversation in Chapter 4. Chapter 6 places the walk in the second month of the ninety days after done, and Chapter 7 gives the walk sheet and the artefact inventory as working pages.
It is ten to nine on a Monday in February, and the claims office smells of wet coats. The senior handler has forty-four claims from the weekend, a lot of them storm damage. She opens the first. The screen opens on the triage view, and the queue is sorted into four tracks, the fourth one the team added in December for customers who need a slower call. She sorts the first twelve in twenty minutes, makes four calls before half past nine, and starts on the assessments.
The laminated card is still propped against her monitor. She has not looked at it since before Christmas.
Her hands know the way now.
Chapter Four
What an organisation calls resistance is often its most accurate forecast of how a change will come undone.
We Tried This in 2016
Haute Lumière · Change That Stays · Chapter Four
It is half past two on a Wednesday in January, and the canteen at the bakery still smells of the morning's bread. The briefing on the new production planning system has just ended. For an hour the operations manager, eight months into the job, has shown thirty people how the system will schedule every run from the sales forecast, so that the ovens are loaded by data rather than by the paper sheet the night supervisor has kept on a clipboard for as long as anyone can remember. There were questions about passwords and about training dates. There were no objections. Now the night supervisor, twenty-six years on the site, is standing by the urn with a mug of tea, and as the operations manager passes she says, not unkindly, We did this in 2016. It lasted till Easter.
The operations manager smiles and says something about how much the software has improved. Her face is warm. Walking back to her office she has already decided what the sentence was: resistance, from the person she was warned about, and a problem to be managed. She will route the training round the night shift for a few weeks and let the results persuade them. She is a composite, as is the supervisor, drawn from the way these afternoons are described in the research below and in the house's own practice, and no bakery is meant.
Ammanuel Santa Anna's canon has a sentence for the moment by the urn, and it is one of the most practically useful in his writing on organisations.23 In a chapter on what unmourned loss does to a workplace, it says:
Change resistance. One of the most common complaints in organizational life is that "people resist change." Entire consulting practices are built on helping organizations overcome this resistance. But here is a truth that most change management frameworks miss: much of what presents as resistance to change is actually unprocessed grief from the last change.
This chapter makes one claim. What an organisation calls resistance is very often the most accurate forecast it holds of how a change will come undone, carried by the people who watched the last one fail. A change that routes round that forecast meets it again months later, as the old way returning, and a change that hears it first can be designed not to.
The word resistance has a long history in organisational writing, and in 2008 Jeffrey Ford, Laurie Ford and Angelo D'Amelio pointed out that the history had been written almost entirely from one side.24 The prevailing view, they argued, tells a story that favours the people leading a change. It treats resistance as an irrational and dysfunctional reaction, located "over there", in the people on the receiving end. They set out to tell the rest of the story, and it has two parts. The first is that change agents contribute to resistance through their own actions and inactions. The second is that resistance can be a resource for change.
The first part is uncomfortable in a familiar way. The paper's examples of what leaders do to produce the resistance they then complain about are, as remembered from the full text, ordinary ones: agreements made during earlier changes and later broken, trust damaged and never repaired, explanations that answered the questions leaders wanted asked rather than the ones people had.25 The bakery's 2016 system was introduced, presumably, by someone who promised that the night shift's knowledge would be built into it, and then left before Easter. The supervisor's sentence by the urn is in part a report on that promise.
The second part is the one this chapter is about. If resistance is a resource, it is because an objection carries information the change agent does not have. It may be information about the work, about the past, or about the people who will have to carry the change on the nights nobody from the project is watching. Sandy Piderit, writing in 2000, added a refinement that helps in practice.26 People's responses to a change, she argued, are rarely just for or against. They are ambivalent, with thoughts, feelings and intentions that need not point the same way. The night supervisor may think the system is a good idea, feel tired at the prospect of another one, and intend to keep her clipboard just in case. Calling all of that resistance throws most of it away.
Santa Anna's canon describes the defensive reflex in organisational terms that a leader can recognise in herself: "When a change initiative threatens the organization’s identity, the organization behaves like a body encountering a pathogen. It resists, neutralizes, or isolates the “foreign” element." The canon means the organisation resisting the change. But the operations manager walking back to her office is doing the same thing to the supervisor. She has classified the objection as foreign and is already planning to isolate it.
Seventy-nine studies
Haute Lumière · Change That Stays · Chapter Four
The question of why people react to change as they do has been studied quantitatively for sixty years, and in 2011 Shaul Oreg, Maria Vakola and Achilles Armenakis reviewed the whole of that literature.27 Of the studies published between 1948 and 2007, seventy-nine met their criteria. From them the authors built a model with three parts. People's reactions are attitudes with three dimensions: what they think, what they feel and what they intend to do. The reactions have antecedents of two kinds. Some exist before the change arrives: the characteristics of the people receiving it and the internal context of the organisation. Others belong to the change itself: how it is carried out, what benefit or harm people expect from it, and what it actually changes. And the reactions have consequences, for work and for people's lives.
The model is a synthesis of correlational studies, and it cannot say what causes what in any particular bakery. But its shape is a corrective to the canteen. Of the things it finds associated with how people react, only one group is about the person. The rest are about the organisation's history, the way the change is introduced and what it will really do to people's work. A leader who hears an objection and thinks only about the objector is reading one box of a three-box model.
Read the bakery against it. The internal context includes 2016. The change process includes a briefing at which there were no objections, which a more experienced manager might have read as a warning rather than a success. The perceived harm is specific and, as it turns out, accurate: the night supervisor knows that the supermarket's Thursday orders change after the forecast is run, and that in 2016 the system scheduled the Thursday bake from figures that were wrong by the time the ovens were lit. Her clipboard is where the correction lives. She has no reason to think the new system knows any of that, because nobody has asked her.
The silence at the briefing deserves a second look. Thirty people heard a plan that changes how the ovens are loaded, and the only questions were about passwords. A room that has no objections to a large change has usually decided that objections are not wanted, or not useful, or not safe, and has saved them for the corridor. The canteen by the urn is the bakery's version of what [Where the Flow Stops calls the meeting after the meeting]({{BOOK:orgdev-flow}}), the second conversation in which an organisation says what it actually thinks. The supervisor's sentence was not the start of a resistance campaign. It was the only part of the briefing that told the truth about the plan's prospects.
The canon's claim about grief deserves more than its first sentence, because the rest of the passage explains the supervisor's tone better than any account of her personality. Unmourned loss, it says, becomes the resistance leaders find so frustrating, "not because people are opposed to the new direction, but because they haven't been allowed to complete their relationship with the old one."
What did the night shift lose in 2016? Perhaps less than a layoff and more than nothing. A way of working that they were good at was declared obsolete in a briefing. A sheet that one of them had refined over years was replaced by a screen. When the screen failed at Easter, the sheet came back without acknowledgment, and the people who had kept it going through the failure were neither thanked nor asked what they had learned. None of that is dramatic. All of it is still in the room in January, by the urn.
The Luminous 100, the house's reading of the frameworks consultancies sell, keeps the distinction that best explains this, from William Bridges: the model "Distinguishes between change (external, situational) and transition (internal, psychological)."28 The change in 2016 happened in a week. The transition, the inner work of letting go of the old way, was never given any time at all, and so it was never finished. The 100's own reading of Bridges says what an unfinished ending could become if it were honoured: "Living-systems awareness shows that the Ending isn't death but composting". The nutrients of the old way, the knowledge in the clipboard, are exactly what the new system needs.
The canon describes a practice for making that history visible, and it is simpler than it sounds. A team draws, on a long sheet of paper, a timeline of the significant losses and changes it has lived through, and stands in front of it. In the canon's account of one such session, seeing them all together "revealed the sheer volume of unprocessed grief the system was carrying." A leader new to an organisation, as the operations manager is, has not lived through any of it. Everyone else has. That asymmetry is most of what she heard by the urn.
If the operations manager asked around, quietly, about 2016, she would probably find a story with three parts, because stories like it usually have them. The system was introduced by a manager who left soon afterwards. It worked well on the day shift, where the orders were steady, and badly on nights, where the Thursday changes landed. And when it was abandoned, nobody said so: the clipboard came back, and the system was left running on a screen in the office, unused, until the licence lapsed. Each part is a lesson about this year's change. The first says who will be left holding it. The second says where it will fail. The third says how the failure will be hidden.
There is a further layer, and it is the one most useful to a leader because it turns a complaint into a question. The developmental psychologists Robert Kegan and Lisa Lahey called it immunity to change.29 Santa Anna's canon summarises their idea with care: "what looks like resistance to change is often a brilliantly organized self-protective system." People who do not change, the canon goes on, are often running a hidden commitment, a quiet dedication to protecting themselves or others from something they fear would happen if they did.
The canon's own use of the idea is the important part. It "reframes resistance not as an obstacle to be overcome but as information to be understood", and it gives the questions that follow: "What is this person protecting? What developmental achievement is at stake? What would need to be true for this edge to feel safe enough to approach?"
Ask the first of those about the night supervisor and the answer arrives quickly. She is protecting the Thursday bake. More precisely, she is protecting the supermarket order, the customers on the shelf on Friday morning, and the twelve people on her shift who in 2016 spent four Thursday nights re-baking a run the system had got wrong. Her clipboard is a safety system, built by hand, and her objection is a report that the new design has no equivalent. The canon's chapter on the parts of any system that guard it puts the organisational version gently: "these all arose for good reasons. They've been protecting the organization from genuine dangers." The danger the clipboard guards against is real. It is also, almost certainly, one the new system could be configured to guard against too, if anyone knew it existed.
Not every objection is right, and it would be a disservice to pretend otherwise. Sometimes the person by the urn is protecting a way of working that suits her and costs everyone else, or defending a skill the new system genuinely makes unnecessary. The canon's questions still help, because they separate the two. An objection that protects a real danger names the danger, specifically, when asked. An objection that protects a position tends to stay general. Either way the leader learns something she needs, and either way the person who objected has been treated as someone whose view of the work counts. That second effect lasts long after the particular argument is forgotten.
That is the link between this chapter and the routine that comes back under pressure, read in Chapter 3 in an office where the screen still opened on the old tab. In that office the old way returned because nobody changed what the work answered to. In the bakery, if the objection is routed round, the old way will return for a better reason. The clipboard will come back on the first Thursday the forecast is wrong, because it is the only thing on site that works on Thursdays.
The conversation before the design
Haute Lumière · Change That Stays · Chapter Four
The practice this chapter describes is a conversation, held before a change's design is fixed, with the people most likely to say it will fail. It is called the objection conversation, and its order matters more than its wording.
It is held early, one to one or in a very small group, and somewhere that is not the canteen after a briefing. The operations manager asks the supervisor for an hour, and says plainly why: that she heard the sentence about 2016, that she thinks the supervisor may know something the project does not, and that she would rather hear it now than in March.
Then three questions, in order. The first is about the work: what will this break? Not whether the system is a good idea, which invites an argument, but where, specifically, it will fail on a real night. The answer will be concrete, and it will often be right. The second is about the past: what happened last time, and what did it cost the people who were there? This is the question that lets the transition be spoken, and it should be asked with genuine curiosity, and without any defence of whoever ran 2016. The third is about what to keep: what does the old way do that the new one must also do? This is where the clipboard's knowledge enters the design: the Thursday correction, the order of the runs when an oven is down, the warning signs that the forecast is off.
Picture it held. It is nine on a Sunday evening, at the start of the night shift, in the supervisor's small office off the bakery floor, because that is where she is most at ease and the operations manager has come to her. The clipboard is on the desk. The first question takes a while to land, because the supervisor expects to be persuaded, and when she realises she is being asked, she picks up the clipboard and turns it round. The Thursday column has two sets of figures in it, one in black and one in pencil. The pencil is the supermarket's call-off, which arrives at four in the afternoon, after the forecast has run. In 2016 the system had no pencil. The second question takes longer, and the answer is not about software at all; it is about four nights of re-baking and a manager who said thank you to the day shift. The third question produces a list of six things the clipboard does, and two of them are things nobody in the project had heard of.
What happens next decides whether the conversation was worth having. The design changes where the objection was right, visibly, and the person who raised it is told, and credited by name where it counts. Where the design will not change, the reason is given plainly, once. And the objector is asked to help with the part she knows best: to walk the Thursday bake with the project team, or to test the system on her shift before anyone else's. The canon's advice for when a change meets its old patterns in a group is the right register for every step of this: "Shadow does not yield to judgment. It yields to patience, to safety, to the repeated demonstration that honesty is valued even when it is imperfect, incomplete, and trembling."
The mistakes are easy to see in advance. The conversation is held after the design is fixed, so that it can only produce disappointment. It becomes a persuasion session, in which the leader explains why the objection is wrong. The leader hears the forecast and ignores the history, or hears the history and treats it as therapy and the forecast as noise. Or the objector's knowledge is taken into the design and her name is left out of it, which teaches the whole shift that speaking up is a way to have your work taken. Chapter 7 carries the conversation as a run sheet, with the three questions, the ways each is commonly misheard and a page for recording what changed.
There is a quieter gain, and it is often the largest. The person who said a change would fail, and was asked why, and saw the design change, is very often the person who will keep it going when the project team has moved on. That is not a guarantee. It is a pattern worth watching for, because the question of who carries a change once its sponsor has gone is the one on which most improvements finally turn, and, as Chapter 5 finds, the person who carried it is very often not the person who launched it.
The article this chapter began as ended in the canteen in April. The book can go further, into the finer points a leader meets when she holds these conversations in her own organisation, and the first is whom to ask.
Three groups, and they overlap. The people with the longest memory of the work, because they carry the history of the last change. The people already described, in the project's own conversations, as likely to resist, because the description usually means they have said something the project did not want to hear. And the people who were silent at the briefing, especially the experienced ones, because silence from someone who knows the work is rarely agreement. In a team of thirty, that is usually four to six conversations. They take an hour each, and they are the cheapest hours in the whole change.
The second point is timing, and the rule is strict: before the design is fixed. A conversation held after the system is configured can still produce goodwill, but it cannot produce a better design, and the person asked will know it within ten minutes. If the design is already fixed when the leader arrives, as the operations manager's nearly was, the honest move is to say so and to ask which parts most need to change before go-live.
The third is the place. The conversation goes to the objector, on her ground and at her hours: the night supervisor's office at the start of a shift, a quiet corner of the floor, not the leader's office and never a meeting room booked for the purpose. The difference in what is said is large.
The fourth is how each question is commonly misheard. What will this break? is heard as an invitation to argue that the change is a bad idea, and some people will take it that way at first. The leader steers back, gently, to specifics: where, on which night, with which order. What happened last time? is heard as an invitation to complain, and it is, partly, which is why it is asked. The leader listens to the end without defending anyone, and then asks what the person learned. What does the old way do that the new one must also do? is the question people most often cannot answer straight away, because the old way's functions are habits to them, not functions. Asking them to walk through a hard night, step by step, usually brings the answer out.
The fifth is sorting what comes back. Where the Flow Stops divides what an organisation says in its corridors into three kinds of sentence: forecasts of what will happen, preferences about what someone would do instead, and grievances about what something has cost before. The same sorting works here. Forecasts go into the design, tested where possible. Preferences are weighed and answered. Grievances are taken seriously as history and separated from the forecast they usually contain; the history is acknowledged, and the forecast is tested.
The sixth is the group version, for a team with a heavy history. The canon's loss timeline, a long sheet on which a team marks the changes and losses it has lived through, is a way of letting a whole team say what the supervisor said by the urn. It needs a careful facilitator, an hour and a half, and a clear promise about what happens next: that what the timeline shows will be read before the new design is fixed. It is not therapy and should not be run as therapy. It is the internal context of Oreg's model, drawn on a wall.
The seventh is what to do when the objection is wrong. It still gets an answer, with the reason, and the person is still thanked. If an objection is wrong in a way that shows a fear rather than a fact, the canon's questions apply: what is this person protecting, and what would need to be true for the change to feel safe? Often the answer is a guarantee the change can give: that a skill will still be needed, that a role will not be cut, that the old way will be kept available for a month as a fallback.
The eighth is the record. Every change to the design that came from an objection is written down with the name of the person who raised it, if she agrees, and shown when the change is introduced. This page, more than any other in the project, tells the rest of the team whether speaking up is worth it.
And the last is the invitation. The objectors who were right are asked to help where their knowledge is greatest: to walk the routine at its hardest hour, as Chapter 3 describes, or to test the new way on their own shift. Some will decline, and that is their right. Those who accept are often the people who keep the change once the project has gone, which Chapter 5 takes up. Chapter 6 places the objection conversations in the first month of the ninety days after done, before the walks, and Chapter 7 gives the run sheet and the record as working pages.
It is half past two on a Wednesday in April, and the canteen smells of bread. The system went live in March on the day shift and in the second week on nights, with a Thursday correction built into it that the night supervisor wrote out by hand and the project team turned into a rule. On the first Thursday the forecast was wrong, and the rule caught it, and the run was right. The clipboard is in the supervisor's locker. She says she is keeping it for a while.
The operations manager is at the urn. The supervisor stops beside her with a mug and looks at the rota on the wall.
It's past Easter, she says.
Chapter Five
A change carried by one person is held on loan against her presence, and the loan is called in the day she leaves.
The Tuesday She Was Gone
Haute Lumière · Change That Stays · Chapter Five
It is five past eight on a Tuesday in April, and the complaints team at the contact centre is logging on. The room is open-plan, fourteen desks under a long window that faces the car park, and the light is the thin spring light that makes every screen look grey. For a year and a half, at half past eight every Tuesday, the team has met in the small room by the lifts for forty minutes. Five complaints closed the week before, read aloud, one question asked of each: what would have stopped this reaching us? One change agreed, written on the whiteboard, checked the following Tuesday. The number of customers who had to call twice fell over that year, and the regional manager mentioned the team by name at a conference.
The senior adviser, four years on the team, looks at her calendar. There is nothing at half past eight. She checks again, and then she understands. The team leader who started the Tuesday review was promoted in March to run the centre's other site, and her calendar went with her, and the Tuesday invitation, which had always come from her, went with the calendar. The new team leader is capable and kind and has never heard of the review. The senior adviser has brought her printouts of five complaints, as she does every week. Her stomach drops in the way it does when a door she expected to be open is locked. She is a composite, as are her colleagues, drawn from the way these weeks are described in the research below and in the house's own practice, and no contact centre is meant.
Ammanuel Santa Anna's canon tells a story about a different kind of place, and it is the clearest account in his writing of what happens when the person who holds something leaves.30 It is about a woman named Eleanor, whose porch was where her neighbourhood gathered every Saturday, and about what followed when she moved away:
When Eleanor had a stroke at seventy-eight and moved to a care facility forty miles away, the neighborhood didn't collapse overnight. But over the following months, something shifted. The Saturday gatherings dissolved. The young mother's depression deepened without her weekly anchor. The firefighter stopped talking to his neighbors. The teenage girl ran away from home. The engineer stopped eating properly.
This chapter makes one claim. A change carried by one person is held on loan against her presence, and it is repaid, usually without anyone noticing, on the day she leaves. A change lasts when its keeping has been moved, while she is still there, into the parts of the organisation that remain when she goes.
Every organisation knows the figure the research calls a champion: the person who believes in a change, carries it, argues for it in meetings and keeps it going when others lose interest. The literature on champions is large and has been growing fast. In 2018 Edward Miech and colleagues reviewed it for health care, where it is largest, and found 199 research articles published between 1980 and 2016.31 More were published in the last two years of their review than in its first thirty. The use of the word varied widely, and most studies recorded only whether a champion was present or absent. But the individual studies consistently found champions to be a positive influence on whether an implementation worked, and the four that had randomly assigned champions to some sites and not others showed that the question could be studied experimentally.
The sustainability research says the same thing from the other end. When Mary Ann Scheirer reviewed nineteen studies of whether health programmes lasted, she found five factors with consistent support across them, and one was that "a 'champion' is present."32 The others were that a programme could be modified over time, that it fitted its organisation's mission and procedures, that its benefits to staff or clients were readily seen, and that it had support from outside.
So the complaints team's experience is not a story about champions being a bad idea. The team leader was the reason the review existed, and for eighteen months she was the reason it worked. The difficulty is what the evidence leaves unsaid. A champion helps a change to start and to last while she is present. Very little in the research measures what happens after she goes, and the complaints team is finding out.
Systems that depend on one
Haute Lumière · Change That Stays · Chapter Five
Santa Anna's canon uses an ecologist's word for people like the team leader. A keystone species is one whose influence on its ecosystem is out of all proportion to its numbers, and the canon carries the idea into human systems: the person whose presence holds a group together, often without a title that says so. Its account of what makes such systems fragile is stated without softening: "systems that depend on a single keystone are fragile systems." And it says, in the same passage, what the resilient ones do instead. They have "distributed their keystone functions across many people, many positions, and many practices."
The canon is also honest about why organisations drift towards the fragile arrangement, and the explanation is not neglect. It is that concentration is easier: "Human systems naturally tend toward keystone concentration — it is more efficient, more comfortable, and more familiar for a system to rely on a few exceptionally capable individuals than to develop distributed capacity across many." The complaints team had every reason to let the team leader carry the review. She was good at it. She chose the cases well, kept the forty minutes to forty, and knew how to turn a complaint into a question nobody felt accused by. Spreading those skills would have been slower, and in any given week, less good.
There is a caution in the canon that matters as much as the diagnosis, because the obvious response to a lost champion is to look for a new one. It calls this the hero trap, the habit of an individualistic culture that reduces a living system to its most visible person: "Find the keystone person and everything will be fine." The point, it says, is that keystone influence "is relational and systemic", emerging from a person, a position and a context together. A new team leader who is asked to become the old one will either fail or succeed in the same fragile way.
In 2004 Trisha Greenhalgh and colleagues reviewed what was known about how innovations spread and are sustained in health services, one of the most thorough syntheses of its kind.33 Their model is detailed and resists summary, but one of its threads, as remembered from the full text, bears directly on the complaints team: an innovation lasts when it becomes routine, built into the organisation's ordinary procedures, budgets and roles, rather than remaining a special project dependent on the people who introduced it.34 That is the organisational form of the canon's distributed keystone. The review needed to stop being the team leader's meeting and become the team's.
If a change is to be moved from a person into a place, someone has to know what the person was actually doing, and this is harder than it sounds, because much of what a champion does is invisible to her as well as to everyone else.
List what the team leader did for the Tuesday review, and the list is longer than ran a meeting. She sent the invitation, from her own calendar. She booked the small room, by asking the facilities coordinator, whom she knew. She chose the five complaints, on Monday afternoon, and she chose them to be varied, never five about the same fault. She protected the forty minutes from the centre's shift planner, who wanted the team on the phones at half past eight, by agreeing a later start to the morning's calls with the duty manager. She wrote the agreed change on the whiteboard and photographed it. She checked it on the following Tuesday. Once a month she told the regional manager what had changed, so that the work was seen above her. And she made the room safe enough that advisers would read out a complaint they had handled badly. Seven of those eight tasks are mechanical. The eighth is not, and it is the one most often mistaken for the whole.
When she was promoted, the handover was a good one by most standards. Her successor received the team's targets, the rota, the open cases and a page on each adviser. The Tuesday review was mentioned, in a line, as a weekly huddle, works well. None of the eight tasks was written down, because none of them had ever been thought of as a task. They were how she worked.
It is worth being fair to the new team leader, because the easy reading of April is that she dropped something valuable. She did not know it was there. In her first month she was learning fourteen people, a new set of targets and a site she had never worked at, and the one meeting on her predecessor's list that sounded optional was the huddle. When the senior adviser finally mentions the review, in the second week of April, the new team leader is glad to hear of it and asks what it involves, and the answer takes twenty minutes, because nobody has ever had to explain it. That conversation is the handover that should have happened in March. It is also, if it goes well, the moment the review stops belonging to one person, because for the first time the people who kept it have to say what keeping it means.
Santa Anna's canon describes this failure in its own setting, a governance practice that depends on a few committed people, and its description fits the complaints team closely: "conscious governance, practiced without adequate support, burns out. The facilitators exhaust themselves. The group's enthusiasm wanes. The organizational culture reasserts its old patterns, and the conscious governance experiment becomes another well-intentioned initiative that didn't survive contact with institutional gravity." Institutional gravity is the right phrase for what removed the review. Nothing pushed it out. It lost the one force holding it up.
The same mechanism appears wherever a change slips back, in a different coat each time. At the outpatient clinic of Chapter 1, the forces that held it were withdrawn one by one when the project closed, each for a good reason.
In a bakery, the person most likely to keep a new system going turned out, in Chapter 4, to be the night supervisor who had said it would fail. In each case the question is the same: what is holding this up, and will it still be there next quarter?
The change literature has an older word for what the complaints team needed. John Kotter, in his account of why transformation efforts fail, named as the last of his errors the failure to anchor a change in the organisation's culture, by which he meant, as remembered, showing people how the new way connected to results, and making sure the next generation of leaders embodied it.35 The second half of that is the part the complaints team missed. The team leader's successor had not been chosen for the review, told about it, or asked to keep it.
The sustainability studies support a broader version of the same idea. Shannon Wiltsey Stirman and her colleagues, reviewing 125 studies, grouped the influences on whether a programme lasted under four headings: the organisational context, the capacity to keep it going, the processes by which it was carried out, and the features of the programme itself.36 Only one of those headings is about people, and even it is about capacity, which can be built, rather than about any particular person.
The canon adds the dimension the studies find hardest to measure, which is time. Of the kind of culture in which keystone functions are shared rather than concentrated, it says: "A keystone culture is not built in a quarter, a year, or even a strategic plan cycle." The Tuesday review ran for eighteen months, which is long enough to change a number and not long enough, on its own, to change who keeps it.
Rehearsing her absence
Haute Lumière · Change That Stays · Chapter Five
The practice this chapter describes has three parts, and the most important is done while the champion is still in post, because that is the only time it can be done well. It is called the rehearsal of absence.
The first part is the carrier's inventory. The champion and one or two colleagues sit down for an hour and list everything she does that keeps the change going, in the plainest terms: who sends the invitation, who books the room, who chooses the cases, who protects the time, who checks the follow-up, who tells the people above. The list should include the relational work, the eighth task, described as concretely as possible: reads the first complaint herself, and it is always one of her own mistakes. Most champions are surprised by the length of the list, and most find at least one task they had never noticed doing.
The second part is moving each task into something that remains. The invitation comes from a team calendar, not a personal one. The room is a standing booking. The choice of cases rotates among the advisers, with a short note on how to choose. The forty minutes are written into the shift plan as a fixture, agreed with the duty manager, so that protecting them is not anyone's personal favour. The monthly line to the regional manager becomes a line in the team's report. And the relational work is shared deliberately, by having different people lead the review, with the champion beside them, until several can do it. None of these moves is clever. Together they are the difference between a meeting someone runs and a meeting a team holds. In the canon's terms, the keystone functions have been distributed across many people, many positions and many practices.
The third part is the absence itself, rehearsed. Once the tasks have been moved, the champion steps away from the change for two or three weeks while still in post: a holiday, a secondment, a deliberate gap. The team runs the review without her. When she returns, they sit down together and read what happened. What ran without her? What stopped? What was done, but differently, and is the difference better or worse? Whatever stopped is the next thing to move. A change that survives a rehearsed absence is a change that is ready for a real one.
Two further habits help. When the champion does leave, her successor inherits the change explicitly, in a conversation about it and not in a line of a handover note, and is asked whether she will keep it, and what she would change. And the people who keep it are named, regularly, for keeping it, in the way Chapter 2 describes for the prevented problems a plant learned to count: maintenance is visible only if someone makes it so.
The mistakes are familiar. The inventory is skipped because the champion is busy and the change is working. The tasks are moved on paper and the champion goes on doing them, so that the rehearsal of absence is the first time anyone else tries. Or the absence is never rehearsed, because it feels like a vote of no confidence, and the real absence, when it comes, is the first test. Chapter 7 carries the inventory, the list of moves and the absence review as working pages.
There is a gain in this practice that has nothing to do with the change itself. The champion who has handed her change to the place is free to carry the next one. The team that has kept a practice without its founder knows something about itself that it did not know before: that the improvement was theirs.
The article this chapter began as ended in the small room by the lifts in July. The book can go further, into the finer points a leader meets when she moves a change from a person into a place, and the first is when to begin.
The best time is at the start, when the change is introduced, by naming from the first week a second keeper beside the champion. The next best time is the moment the change has visibly worked, when the champion has most credibility and least reason to fear that sharing it will look like a loss of confidence. The worst time is when the champion's departure has been announced, because by then everyone, including her, is busy with the handover of everything else, and the change gets a line.
The second point is how to run the carrier's inventory so that it finds the invisible tasks. A list made from memory in a meeting finds the visible ones. Two methods find the rest. The first is a week's diary: for five working days the champion notes every action, however small, that she took because of the change, from the email she sent to the room she booked to the colleague she reassured in a corridor. The second is the question asked of the team rather than of her: what does she do for this that you would notice if she stopped? The team's answers are usually different from hers, and the differences are where the relational work is.
The third is matching each task to a structure that remains. The structures are few and ordinary, and most organisations already have them: a team calendar or shared mailbox in place of a personal one, a standing booking in place of a favour, a rota in place of a person, a fixture in the shift plan or the production schedule, a line in a regular report, a sentence in the role description of whoever holds the post, and a page in the induction for new staff. Every task on the inventory should end with the name of the structure that now holds it, and the date it moved.
The fourth is the relational task, which cannot be moved into a structure and has to be shared between people. The method that works is the oldest one: the champion leads, then co-leads, then sits beside someone else who leads, then is absent. Each stage lasts several weeks. During the co-leading stage the champion should say less than she wants to, and in the sitting-beside stage she should speak last, if at all. The canon's distributed keystone is built this way, one person at a time.
The fifth is the absence itself. It should last long enough for the change to meet one of its ordinary pressures without the champion: two weeks for a weekly practice, a full month for a monthly one. During it she is not consulted, which needs saying out loud, because the team's first instinct will be to message her and hers will be to answer. Before she goes, the team agrees what it will write down: each week, what ran, what did not, and what was done differently.
The sixth is the absence review, held in the week she returns. Four questions, in order. What ran without her? What stopped, and why? What was done differently, and is it better? What task is still hers that should not be? The last question produces the next round of moves. Many teams need two rehearsals before a change survives a real absence, and that is a good result, not a disappointing one, because the second rehearsal costs nothing and the real departure, unrehearsed, can cost the change.
The seventh is the successor conversation, when the real departure comes. The incoming leader is told about the change in a conversation of its own, with the inventory, the record of moves and the last absence review on the table. She is asked whether she will keep it and what she would change, and her answer is taken seriously. A successor who chooses to keep a practice is a keeper. A successor who inherits it in a line is a caretaker until the first busy month.
And the last is the hardest to accept: not every change should last. Some practices outlive their purpose, and keeping them out of loyalty to the person who began them is its own kind of fragility. The canon's account of living systems names the balance between keeping an identity and adapting to new conditions, and a change that is retired deliberately, with its reasons and its lessons written down, has not failed. What fails is the change that fades without anyone deciding, because then nobody learns whether it should have lasted. Chapter 6 places the carrier's inventory and the rehearsal of absence in the third month of the ninety days after done, and Chapter 7 gives the inventory, the moves and the absence review as working pages.
It is five past eight on a Tuesday in July, and the complaints team is logging on under a window full of summer. At half past eight the team calendar sends its reminder. The small room by the lifts is booked, as it is every Tuesday, until December. This week the senior adviser has chosen the five complaints, and one of the newer advisers will lead.
At twenty past eight a message arrives on the team channel from the centre's other site. It is from the old team leader, who has started a review of her own there, and wants to know how the team chooses its cases now. The senior adviser types a reply, attaches the note on choosing, and picks up her printouts.
The door to the small room is open.
Chapter Six
Five practices set in order, from reading the record of past changes to rehearsing the absence of the person who carries this one.
The Ninety Days After Done
Haute Lumière · Change That Stays · Chapter Six
It is a quarter to eight on a Monday in late spring, and the operations director of a firm of some six hundred people is at her desk before anyone else on the floor. On Friday the firm celebrated the end of a year-long improvement programme: a new way of scheduling work across its three sites, shorter lead times, fewer late orders, a short speech by the chief executive and a buffet in the atrium. The programme team has been thanked and is being released back to its old jobs this week. On her desk is the closing report, which says the programme is complete. Beside it is a list she wrote on Sunday evening of the firm's last three improvement programmes and what became of each. Two of them she can no longer find in any report.
She is a composite, drawn from the way these 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 who has just been told that a change is finished.37 It describes the first ninety days after done in order, with who is in the room, what each step needs, roughly how long it takes, the decisions that change the result, the mistakes that are 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 nothing here is a new programme. The ninety days after done are the programme's last stage, carried on in a lighter form by the people who will own the work from now on, and they should never be announced as an initiative of their own. The canon names why programmes so often leave an organisation unchanged: "Because most change initiatives operate at the level of structure and behavior while leaving the deeper levels of collective emotion, somatic patterning, and meaning-making untouched." The ninety days are spent at those deeper levels, in quiet ways: the record of what happened before, the objections nobody voiced, the cues in the work, the time and credit the practice needs, and the people who carry it.
The second is that urgency is the wrong fuel. The Luminous 100's reading of Kotter's eight steps replaces the first of them, the creation of urgency, with something gentler, on the ground that "the most sustainable change energy comes not from fear of what we'll lose but from the pull of what we're becoming." A change that has to be kept alive by alarm will be lost the first month the alarm is turned down.
The five practices in this book are the holding read (Chapter 1), the protected hour and the quiet ledger (Chapter 2), the routine walk (Chapter 3), the objection conversation (Chapter 4) and the rehearsal of absence (Chapter 5). In the ninety days after done they run in a different sequence: read the record, hear the objections, walk the routine, protect the time, move the carrier's work, and read the holding line throughout.
The reasoning follows the way changes are lost. A change is most at risk from the forces that undid the last one, so the first ten days are spent finding out what those were. The people most able to say how this one will come undone are those who objected, or stayed silent, so they are heard next, while the design can still move. Only then does it make sense to walk the routine, because the walk is guided by what the objectors said and goes better with them in it. Protecting improvement time comes after the walk, because the walk shows how much time the new routine actually needs. Moving the champion's work into the organisation comes last, because it depends on knowing what the practice now is. And the holding read starts on day eleven and never stops, because it is the instrument that tells the director whether any of the rest is working.
Four roles carry the ninety days, and they should be named before anything else is done.
The sponsor is the most senior person who will be asked, a year from now, whether the change held. Often that is the director who sponsored the programme; sometimes it is the person who runs the operation day to day. Her job is to protect the time the other roles need, to attend the drift conversations, and to keep the improvement on the first page of whatever report she reads. Expect two to three hours a week.
The keeper owns the holding read. She runs the work the change touched and was not on the programme team, for the reason Chapter 1 gives: a project lead reading her own result is reading her own trophy. Expect an hour a week, more in the months a drift rule is met.
The champion is the person who carried the change: the programme lead or, more often, the team leader or analyst who made it work in practice. For ninety days her job reverses. She stops doing the change and starts handing it over, which is harder.
The convenor keeps the record: the record of changes, the objection pages, the walk sheets, the record of cuts, the quiet ledger, the carrier's inventory, the holding chart. About two hours a week.
Around them sits a small group of four to six: the keeper, one or two people from the front line of the work, one person with a long memory of the organisation's earlier changes, and one recent joiner. The canon, writing about any practice that depends on a few committed people, is plain about why a group is needed and not a single owner: "The community of practice is not a luxury." Without it, it goes on, "the practice dies of isolation."
One decision changes the result more than any other at this stage. The sponsor must decide whether the programme is still open. If the closing report has been signed and the programme team released, the ninety days will have no standing and will be treated as optional. The better course is to tell the organisation, plainly, that the programme closes in ninety days, and that the last stage is about keeping what it found.
The first ten days change nothing visible. They gather four inputs, and the sponsor and the convenor do most of the gathering.
The first is the record of changes: the last three improvements the organisation made to work of the same kind, what each was meant to do, what it is doing now, and when anyone last looked. The convenor finds the numbers where they exist. The person with the long memory supplies the rest. The page usually takes a few hours and is usually sobering, and it should be read without blame, because Chapter 1's argument is that most fading is a mechanism, not a decision.
The second is a short list of the practices that produced this change's result: the specific things people now do that they did not do before. The champion writes it, and the front-line members correct it. These become the practice measures of the holding read.
The third is the history of the result. Finance or the analyst copies the last one to two years of the result measure, with the programme period marked, dates the copy and files it unaltered. It is the history against which the holding line will be read.
The fourth is the list of people to hear: those with the longest memory of the work, those the programme team described as resistant, and those who were silent at the launch briefings. Chapter 4 explains why each.
At the end of the ten days the group meets for the first time and writes, on a single page, its first guess at how this change is most likely to be lost: which force from the record of changes is most likely to repeat, which practice is most fragile, who is carrying most of it. The page is labelled a guess and kept for day ninety.
What goes wrong at this stage. The record of changes is skipped because it feels like dwelling on failure. The programme team writes the practice list alone, and it describes the change as designed rather than as performed. And the sponsor treats the first ten days as a formality before the real work, which teaches the group the same.
Days eleven to thirty: hearing the objections
Haute Lumière · Change That Stays · Chapter Six
In the third week the sponsor or the keeper holds the objection conversations of Chapter 4, one to one, on the other person's ground and at her hours. Four to six, an hour each. The three questions, in order: what will this break, what happened last time, and what does the old way do that the new one must also do.
On the same days the holding read begins. The keeper and finance agree, on one page, what done means, the result measure and its history, the practice measure, the period, and the drift rule, and they sign it before the first monthly figure is read.
The decision that changes the result is what is done with what the objectors said. Some of it will require changing the new way, after the programme team thought the design was finished. If the changes are made, and visibly credited to the people who raised them, the rest of the ninety days run on goodwill. If they are not, the objectors' forecasts will come true, and on schedule.
What goes wrong. The conversations become persuasion. The forecasts are noted and filed. The holding agreement is written by the programme team, which wants to close, rather than the keeper, who has to live with it. The early signs are easy to read: an objector who agreed to talk and then becomes quieter, and a holding agreement whose period happens to end before the change's first hard season.
How to read what comes back. By day thirty the convenor's record should show a handful of specific forecasts, each with what was done about it, and a signed holding agreement. The forecasts are the most valuable thing the ninety days will produce, because they are the organisation's own account of how this change will be lost.
Overlapping with the conversations, the group walks the routine at its hardest hour, as Chapter 3 describes: Monday morning, month-end, shift change, whatever the front-line members name. One walk per site or team, with an objector from the conversations in the walk wherever one agreed. Each walk records what was done, what prompted it, and whether it was the new routine or the old.
The walks feed the artefact inventory: the defaults, sort orders, forms, reports, rooms, first moves and forms of praise that the performance answers to. The group changes the cheapest cue first and walks again in four weeks.
The decision that changes the result is whether the team's adaptations are welcomed. The new routine will already have changed in the weeks since launch. If the walk treats each change as a deviation, the team learns to hide its adjustments, and the routine on paper and the routine performed begin to drift apart. If the walk records each change and decides with the team, within a week, which to keep, the routine becomes the team's.
What goes wrong. The walk is done on a quiet afternoon. The cues are listed and the training is repeated. A default that needs another department to change it is left because the request is awkward. The early sign is a second walk that finds the same old steps for the same reasons.
By the fifth week the walks will have shown how much time the new routine needs to be kept and improved: the planned stops, the review hour, the weekly look at what recurred. The sponsor agrees that time with whoever owns output, as a standing fixture in the plan the operation already runs to, as Chapter 2 describes, and with a warning of the dip it will cause and roughly for how long.
The convenor opens the record of cuts. The keeper opens the quiet ledger and reads it, once a month, at the review where rescues are discussed.
The decision that changes the result is the first cut. There will be a busy week in which the protected time is wanted for output, and someone will decide. If the cut is written down with a date for restoring the time, and the time is restored on that date, the practice has survived its hardest case. If not, the second cut will be easier.
What goes wrong. The protected time is held but not used. The ledger fills with imagined disasters. The stand-up goes on praising only the rescues. The early sign is the share of planned against reactive work, which Chapter 2 suggests as a simple measure; if it falls for two months running, something is being cut that the record has not caught.
In the eighth week the champion runs the carrier's inventory of Chapter 5: a week's diary of every action she takes for the change, and the team's answer to what they would notice if she stopped. Each task is matched to a structure that will remain, a team calendar, a standing booking, a rota, a fixture in the plan, a line in a report, a sentence in a role description, a page in induction, and moved, with the date.
The relational part of her work is shared by leading, co-leading and sitting beside. Then, between days seventy and eighty, she is absent from the change for two weeks while still in post, and not consulted. The team keeps a short record of what ran, what stopped and what was done differently.
The decision that changes the result is whether the absence really happens. It is easy to postpone, because it feels like a vote of no confidence and because the change is working. An absence that is postponed beyond the ninety days rarely happens at all, and the real departure becomes the first test.
What goes wrong. The tasks are moved on paper and the champion goes on doing them. The team messages her during the absence and she answers. The early sign is an inventory that contains no relational tasks, which means it was written from memory and has missed the work that matters most.
Days sixty-one to ninety: the holding line and the absence review
Haute Lumière · Change That Stays · Chapter Six
In the last month the holding read has its first real test. By now there are two or three monthly figures since launch on the holding chart, beside the practice measure. If the drift rule has been met, the keeper holds the drift conversation of Chapter 1 within a fortnight: what has changed in the conditions, is the practice still being done, does the change need adjusting, and who will do what by when. If it has not been met, the keeper says so, briefly, at the monthly review, and reads the practice line aloud.
In the week the champion returns, the group holds the absence review: what ran without her, what stopped and why, what was done differently and whether it is better, and what is still hers that should not be.
What this step does not settle is how a gain that has been kept 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, and they are decided with the people who will be covered; for an organisation working with Luminous Prosperity they are the substance of the engagement, not of a book. What the ninety days can do is make that later decision honest. A gain that fades in its seventh month is not one both sides can sign, and the measurement conversation of Where the Flow Stops assumes a gain that has been held long enough to be read.
What goes wrong. The holding line is read and not discussed. The first drift is met with a relaunch. The absence review is cancelled because the change survived. The canon describes the most deceptive failure of all, and it is the one to watch for in the last month: a practice that continues in name after its purpose has drained away. "The form may persist, but the life has gone out of it." The sign is a meeting that is still held and no longer changes anything.
The group meets on day ninety with seven things on the table: the record of changes, the first-guess page from day ten, the objection pages with what was done about each forecast, the walk sheets and the artefact inventory, the record of cuts and the quiet ledger, the carrier's inventory with the absence review, and the holding chart.
It asks a short list of questions, in order. Where was the first guess right, and where wrong? Which forecasts from the objection conversations have already come true, and were they caught? Is the new routine performed at the hardest hour, and what still prompts the old one? Has the protected time been cut, and was it restored? What ran without the champion, and what is still hers? What does the holding line show, beside the practice line? And then the question that matters most for the next ninety days: which force from the record of changes is still present?
What not to conclude is as important. Ninety days cannot show that a change has held; they can show whether the conditions for holding it are in place. The holding read continues, usually for the rest of the year, until the change has met each of the pressures named at the start. The canon puts the test in a sentence that fits this day exactly. The question, it says, is whether the organisation "has developed enough capacity, enough practice, enough somatic grounding in the new way of being, to recognize the old pattern when it arises and choose differently."
The programme can now be closed, and the closing report should say so plainly: what was found, what is being held, by whom, how it is read, and when the holding read ends.
If the whole of this chapter is too much to begin with, three things can be done on the first Monday after a change is declared done, and none of them needs anyone's permission.
Write down the last three changes your part of the organisation made to work like this one, and what each is doing now. If you do not know, find out for one of them.
Choose one practice that produced this change's result, and find a way to count, each week, whether it is still being done. Put the count beside the result.
And ask the person who carried the change what she does for it that nobody else would notice if she stopped. Write the answer down.
The canon says of the small practices that hold a culture together what could be said of all five in this book: "These are not extraordinary practices. They are small, ordinary, and profoundly consequential." Kept for ninety days, and then for the ninety after that, they are how a change becomes the way things are done.
It is a quarter to eight on a Monday in late summer, and the operations director is at her desk before anyone else on the floor. The closing report she signed on Friday is in the tray. It says the programme is closed and names a keeper at each site. On the wall by her door the holding chart has five monthly points on it since launch, and beneath them a second line, the share of orders scheduled by the new rule, which dipped in July when one site's planner was on leave and came back the week after.
Her Sunday list of the firm's last three programmes is still on the desk. She adds a fourth line to it, with today's date, and leaves the rest of the line blank.
Chapter Seven
Fourteen working pages, the agreements, the walk sheet, the run sheets and the reviews, ready to carry into a room.
The Field Manual
Haute Lumière · Change That Stays · Chapter Seven
The chapters of this book describe five practices as people doing them.38 This chapter holds the working pages: the agreements, run sheets, question sets and one-page records 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 ninety days after done, 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. Read the work as it is performed, not as it is described. And answer everything you collect: an objection heard and not answered, a cut recorded and not restored, a drift seen and not discussed, teaches the opposite of what the practice is for.
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.
From Chapter 1. Used in days one to ten, and kept afterwards as the organisation's own history of change.
List the last three improvements made to work of the same kind as this one.
| Change | When introduced | What it was meant to do | What it is doing now | When anyone last looked | What undid it, if anything |
|---|---|---|---|---|---|
Reading it. Look down the last column. The force that undid an earlier change is the one most likely to undo this one. Write it at the top of page 2.
From Chapter 1. Agreed and signed between days eleven and thirty, before the first monthly figure is read.
| The change, in plain words | |
| What done means (the condition for closing, not the target) | |
| The keeper (runs the work; was not on the programme team) | |
| Result measure, and the record it comes from | |
| History used, and the ordinary range it shows | |
| Practice measure (the thing that produced the result), and how it is counted | |
| The pressures the change has not yet met (season, peak, holiday, year-end) | |
| Holding period (until each pressure has been met once) | |
| Drift rule (agreed now, applied the same way to good and bad news) | |
| The force most likely to undo it (from page 1) | |
| Date agreed | |
| Signed |
An example drift rule, to be adapted: one monthly point outside the range of the history, or three in a row on the same side moving towards the old level.
From Chapter 1. Read monthly by the keeper; the conversation is held within a fortnight of the drift rule being met.
The chart. Plot the result measure each month against its history, with the programme period marked. Beneath it, on the same page and the same time axis, plot the practice measure.
Reading the two lines together.
| Result line | Practice line | What it usually means |
|---|---|---|
| Holds | Holds | The change is being kept. Say so at the review, briefly. |
| Drifts | Holds | The conditions around the change have shifted. Look at demand, mix, inputs. |
| Holds | Falls | Something else is holding the result up. The change itself may be fading. |
| Drifts | Falls | The change is going home. Hold the conversation now. |
The drift conversation (thirty minutes, four questions, in order).
Avoid. Opening with who let this slip. A relaunch with new posters. Changing the drift rule in the month it is met.
Closing the holding read. When the practice is in the role descriptions, in induction and in the budget, and the lines have held through each pressure named on page 2, the result joins the ordinary report. Add a line to page 1.
From Chapter 2. Opened once improvement time is protected (days thirty-one to seventy), kept by the planner or keeper, read at the monthly review.
The protected time. What it is: How much, how often: Where it is written in the operation's own plan:
| Date | Time given up | Who decided | Reason | To be restored by | Restored on |
|---|---|---|---|---|---|
The dip, forecast in advance. Expected cost in output or time: Expected to last until: Told to:
Reading it. A cut is not a fault. An unrestored cut is the beginning of the capability trap. Read the page beside any breakdown or recurrence: did it follow a cut, and how long after?
5 · The quiet ledger
Haute Lumière · Change That Stays · Chapter Seven
From Chapter 2. Read once a month, at the review where rescues are discussed.
Only things with a trace: a planned stop held, a part replaced and found worn, a defect caught before it shipped, a complaint that did not recur because its cause was fixed.
| Month | What did not happen | The evidence | Who or what prevented it |
|---|---|---|---|
One measure of the trap loosening. Each month, the share of work that was planned rather than reactive (or, in an office, new work rather than rework and chasing): per cent.
For performance conversations. What did you prevent this year, and how do we know?
From Chapter 3. Used in days twenty-one to fifty, at the routine's hardest hour; repeat after four weeks.
The hour: (ask the people who do the work which hour it is) Whose work: Agreed with them beforehand: yes / no
Before starting, say: I am here to find out what the work answers to, not to check whether you follow the routine. Do not correct anything during the walk. Ask what prompted a step after it, never during.
| Step | What was done | What prompted it (in their words) | New routine or old? |
|---|---|---|---|
Afterwards (ten minutes, with the person). Which steps felt like the new way? Which felt like the old way, and what made the old way easier at that moment? What have you changed about the new routine, and why?
Reading it. The third column is the input to page 7. A person who meant to do the new routine and did not is meeting a cue. A person who chose the old one and can say why is raising an objection; take it to page 8.
From Chapter 3. Filled from the walk sheets.
| Kind of artefact | What it is here | Prompts old routine or new? | Who can change it | Changed on |
|---|---|---|---|---|
| Defaults of a system (opening screen, standard option) | ||||
| Order in which work arrives or is listed | ||||
| Forms and templates | ||||
| Reports, and the measures people are read against | ||||
| The room: who sits where, who is in earshot | ||||
| The first move of the day or shift | ||||
| What is praised, and where |
The order of changes. The cheapest cue that prompted the most old steps first. Then reports and measures. Then the room and the order of the day. Train only after the cues have changed, and briefly.
Adaptations. Each change the team makes to the new routine is written here with its reason and decided with the team within a week.
| Date | Adaptation | Reason | Kept, changed or reversed, and why |
|---|---|---|---|
From Chapter 4. Used in days eleven to thirty, before the design is fixed, with four to six people.
Whom to ask. The people with the longest memory of the work. Those already described as likely to resist. The experienced people who were silent at the briefing.
Where and when. On their ground, at their hours. An hour. Say why: I think you may know something the project does not, and I would rather hear it now.
The three questions, in order.
| Question | Commonly misheard as | Steer back to |
|---|---|---|
| What will this break? | An invitation to argue against the change | Where, on which day, with which piece of work |
| What happened last time, and what did it cost the people who were there? | An invitation to complain | Listen to the end; do not defend anyone; then ask what they learned |
| What does the old way do that the new one must also do? | A question with an obvious answer | Walk through a hard day, step by step |
If the objection seems to protect a fear rather than a fact, the canon's questions, in its words: What is this person protecting? and What would need to be true for this edge to feel safe enough to approach?
Sorting what you hear.
| Forecast (what will happen) | Preference (what I would do instead) | Grievance (what this has cost before) |
|---|---|---|
Forecasts go into the design, tested where possible. Preferences are weighed and answered. Grievances are acknowledged as history, and the forecast inside each is tested.
The record of what changed because of an objection.
| Raised by (with permission) | The point | What changed in the design | Date | Told to them on |
|---|---|---|---|---|
Avoid. Holding the conversation after the design is fixed. Persuading. Taking the knowledge and leaving out the name.
From Chapter 4, after the canon's practice. Optional, for a team with a heavy history of change. About ninety minutes, with a careful facilitator.
It is not therapy, and should not be run as therapy.
10 · The carrier's inventory and the moves
Haute Lumière · Change That Stays · Chapter Seven
From Chapter 5. Used in days fifty-one to eighty.
Two ways to find the tasks. A week's diary kept by the champion of every action taken because of the change. And the question asked of the team: what does she do for this that you would notice if she stopped?
| Task | Visible or relational? | Structure that will hold it | Who now does it | Moved on |
|---|---|---|---|---|
| Sends the invitation | Team calendar | |||
| Books the room | Standing booking | |||
| Chooses the material | Rota, with a note on how to choose | |||
| Protects the time | Fixture in the shift plan or schedule | |||
| Checks the follow-up | ||||
| Tells the people above | Line in a regular report | |||
| Makes it safe to speak | Relational | Shared: lead, co-lead, sit beside, absent | ||
Structures that remain. A team calendar or shared mailbox. A standing booking. A rota. A fixture in the operation's plan. A line in a regular report. A sentence in a role description. A page in induction.
From Chapter 5. The champion steps away for two weeks (a weekly practice) or a month (a monthly one), is not consulted, and the team records each week what ran. The review is held in the week she returns, about an hour.
| Week | What ran | What did not | What was done differently |
|---|---|---|---|
The review questions, in order.
The answers to question 4 go back onto page 10. Two rehearsals before a real departure is a good result.
The successor conversation, when the real departure comes. The inventory, the moves and the last absence review on the table. Will you keep this? What would you change?
From Chapter 6.
| Days | Practice | Who | Time it takes | What it produces | Early sign it is going wrong |
|---|---|---|---|---|---|
| 1–10 | Read the record: past changes, the practice list, a dated copy of the result's history, whom to hear (pages 1, 2) | Sponsor, convenor, long-memory member | Sponsor about 3 hours a week | A first-guess page: how this change is most likely to be lost | The record of changes is skipped |
| 11–30 | The objection conversations; the holding agreement signed (pages 2, 8, 9) | Sponsor or keeper; keeper and finance | An hour per conversation; an hour for the agreement | Forecasts, design changes, a signed agreement | An objector goes quiet |
| 21–50 | The routine walk and the artefact inventory (pages 6, 7) | Working group, with an objector | An hour per walk; repeat in four weeks | Cues changed; adaptations decided | The second walk finds the same old steps |
| 31–70 | The protected hour and the quiet ledger (pages 4, 5) | Sponsor and the owner of output; keeper | Monthly reading, 15 minutes | Time held; prevention made visible | A cut not restored |
| 51–80 | The carrier's inventory, the moves, the rehearsed absence (pages 10, 11) | Champion, team | A week's diary; two weeks' absence | Tasks in structures; an absence review | The inventory has no relational tasks |
| 11–90 and after | The holding read (pages 2, 3) | Keeper | An hour a month | A chart, two lines, drift conversations | The chart is read and not discussed |
From Chapter 6. A meeting of the group and the sponsor, about two hours.
Bring the record of changes, the first-guess page, the objection records, the walk sheets and artefact inventory, the record of cuts and the quiet ledger, the carrier's inventory and absence review, and the holding chart.
Ask, in order:
And the canon's prompt, for the group's own agreements: How are we doing with our agreements? Where are we strong? Where have we drifted?
What not to conclude. That the change has held (ninety days show whether the conditions for holding it are in place, not whether it has met every pressure). That one good month is a kept gain. That a meeting still held is a practice still alive.
Closing the programme. The closing report names what was found, what is being held, by whom, how it is read, and when the holding read ends.
From Chapter 6. For anyone who wants to begin with less.
Most improvements that fade were not rejected. They were left unread in the months after they were declared done, and those months can be given an owner and a measure.
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.
This book sets out ninety days for one change. Which of your own are quietly coming undone, and what keeping them would return, is read from your records and your people. That begins with a conversation.