TL;DR
- The decision this article is about: keeping the strongest people in the retained population onside after a restructure, while the entire HR apparatus is pointed at the people who are leaving.
- When doing nothing is right: when no high-performer in the layers above the affected population has signalled ambivalence, and the design genuinely leaves their role, scope and reporting line intact.
- What has to be true for any of this to work: the affected population is honestly defined, the leaders who stay are briefed before their teams, and someone owns the risk of unplanned attrition as a tracked item, not a vibe.
- How the approaches split: proactive retention of named individuals, structural reassurance for the layer above, and quiet sensing across the retained population, used in that order of priority.
- The decision rule: if the top layer of the retained population isn't personally reassured by a named person they trust before the announcement lands, the plan is missing its most important step.
- The outcome to expect: the plan reads as a success two weeks after announcement. The real test is whether the high-performers you kept are still in role two quarters later, working at the same level, on the same kind of work.
The room where the plan is signed off, and the silence nobody planned for
A CHRO at a mid-cap industrial business opens the final version of the reorganisation pack the night before the leadership off-site. The structure is settled. The selection grid is signed. The legal review is closed. The comms timeline is locked. She scrolls to the retention risk page and finds it's the section the team never built. Everyone worked on who is leaving. Nobody worked on who might leave because of the way they're being kept.
She closes the laptop and walks through the question that should have been on the page from the start: who in the layers we aren't touching is recalculating right now, on the basis of a silence we've not even noticed we are running. This is the work that decides whether the reorganisation pays back. The risk the plan names is the departure list. The risk the plan misses is the people it never intended to lose.
The real issue isn't the structure. The real issue is the people who stay, who were never told the plan was coming, and who decide what to do about it on the strength of nothing more than the absence of a conversation they were expecting to have.
Best tools for HR Strategy
When you do not actually need to act yet
There are four honest stages here, and the first one is real. Not every restructure needs a retention motion built around it. The job is to know which stage you're in, and not to behave as if a stage-three problem can be solved with stage-one thinking.
Stage one: the current setup is genuinely fine. A team of forty has been reorganised twice in three years and the leadership group has held together both times. The two directors who carry the technical knowledge were consulted during the design. Their reporting line is unchanged. Their remit is wider, not narrower. The people at the next layer down have had honest one-to-ones about what the new structure means for them. In this situation, doing nothing is the right call. A retention motion layered on top of an already-honest process signals doubt to a population that has no reason to doubt. Read the room. If the room is steady, leave it.
Stage two: the friction you can feel but can't yet name. A division of two hundred has been through a quiet stretch where the most respected senior managers have stopped volunteering for cross-functional work. The restructure is small, two layers deep, and affects a single function. The senior managers above the affected layer haven't been told what the change means for them, only that it's happening. Their behaviour is shifting before the announcement. Nothing is broken. Something is loose. This is the stage where a structured conversation with the top of the retained population prevents the looseness from setting into a story. The story, once it sets, is the thing you spend the next year trying to walk back.
Stage three: real risk of unplanned attrition. A business of six hundred has cut a layer out of the commercial function. The heads of the two remaining commercial teams were not in the room when the decision was made. One of them has been in the role for less than a year and has options. The other has equity that vests next spring. Both have been told informally by their own people, who have been told by theirs, who have been told by HR. The plan has not been announced, but the plan is already in the building. At this stage, not acting is itself a decision, and the decision is to let the strongest people in the retained population read the silence as the answer. Some of them will read it correctly. Some won't.
Stage four: the edge case where the restructure itself is the secondary risk. A business of a thousand plus is mid-process on a strategic divestment that will reshape the entire organisation. The restructure is happening because the divestment is happening. The retention risk isn't the people in the layers being cut. It's the leadership group two levels above the cut, who can see that the company they joined is no longer the company they're in. At this stage, the question is whether the restructure is the right vehicle for the conversation at all. Sometimes the right answer is to pause the reorganisation and have the strategic conversation first, so that the structural conversation has a context to sit inside.
Five questions you ask yourself at 11pm
Should I be reassuring the retained population, or am I panicking? The honest answer is that you can't tell from inside your own head. You can tell from the behaviour of the two or three most senior people in the layers above the affected population. If they're behaving normally, you're reassuring yourself unnecessarily. If they're behaving differently, you're not panicking early enough. The mistake is treating your own anxiety as data. It isn't. Their behaviour is.
Who in the retained population can I least afford to lose? Write the names down. Not the role. The names. The five to eight people whose departure would change the kind of business you're running rather than the shape of it. If you can't write the names down, you've not thought hard enough about what this restructure is for. The whole point of the structure you're building is to be the kind of business those people want to work in.
How much do I tell them before the design is final? Tell them the direction, not the destination. Tell them the criteria, not the conclusions. Tell them what you're trying to solve for, and what you're explicitly not willing to trade. A senior manager who hears only that the structure is changing will fill the rest with the worst version of the silence. A senior manager who hears what the plan is for, and what is off the table, can hold the uncertainty. The detail that's not yet ready isn't the part that needs to be said.
What does the layer above the affected population actually need? Not a message. A conversation. Specifically, a conversation with a named person they trust, in which three things are true: their own role is being talked about, the criteria being applied are clear, and the person they're talking to is allowed to be honest. Reassurance that comes through a town hall, or through their manager's manager, doesn't land. Reassurance that comes through a person they already trust does. The mechanism is the relationship, not the content.
What is the cost if I lose one of them? Not the recruitment cost. The cost of the project they were going to lead. The cost of the customer they were the contact for. The cost of the technical knowledge that walked out of the building with them. If you can't price it, you don't yet understand what you're protecting. If you can price it, you've a number that makes the conversation worth having.
Three honest categories the approaches split into
Proactive retention of named individuals. This is the work of identifying, in advance, the small number of people in the retained population whose departure would change the kind of business you're running. The retention motion is built around them as individuals. A senior leader is assigned. A conversation is held before the announcement. The conversation names the role, the scope and the trajectory. Where appropriate, it also names the economic terms. The motion is built around the principle that the strongest people in any organisation are the ones with the most options, and they will read a restructure as a signal about whether their options are getting better or worse. The motion succeeds when those people make a public, visible commitment to the next phase. It fails when it's treated as a bribe. The named individuals can tell the difference. So can their teams. A retention conversation that feels transactional will produce a transactional outcome.
Structural reassurance for the layer above the affected population. This is the work of making the case for the restructure to the people who aren't affected, in a way that answers the question they're actually asking, which isn't "why is this happening" but "what does this mean for me". The motion succeeds when the layer above the affected population can explain the restructure to their own teams in a way that's honest, specific and bounded. It fails when the explanation is too generic to be useful. Generic reassurance produces generic anxiety. Specific reassurance produces specific commitment.
Quiet sensing across the retained population. This is the work of listening for signals, before the announcement, in the population that has not been formally engaged. The motion succeeds when the sensing surfaces two or three named individuals who are quietly recalculating on the strength of rumour, and who would benefit from a conversation before the announcement rather than after. It fails when the sensing becomes surveillance. People can tell when they're being listened to in order to be managed. The motion only works when the listening is genuine, and when the things that surface are acted on, not filed.
Five diagnostic questions to self-assess against
Who, by name, would I lose sleep over losing? If you can't answer this with five to eight names, the question to ask yourself is whether the design has been tested against the people it's meant to keep. The way to answer it's to walk the org chart, layer by layer, and write down the names whose departure would change the kind of business you're running. The names should be specific. A role isn't a name.
How recently have those people been in a room with the leader they most trust? If the answer is more than four to six weeks, the gap is itself the problem. The restructure will land into a relationship that has been allowed to cool. The way to answer this is to ask the question for each name: when did they last sit with someone who could tell them the truth. If the answer is no-one, the work to do isn't the restructure.
What is the most generous interpretation a high-performer in the retained population could put on the silence? Write it down. If the most generous reading is bad, the actual reading will be worse. The way to answer this is to write the email you're not sending, in the voice of the senior manager who is currently not in the loop, and to ask yourself whether the absence they're interpreting could plausibly mean they're about to be cut.
What story is the layer above the affected population already telling itself? The story will exist. It will have been written without your input. The way to answer this is to ask the question of the two or three most senior people in that layer, in a setting where they can be honest, and to listen to the answer rather than to the answer you were hoping for.
What is the explicit criterion the retained population can apply to test whether they're affected? If the criterion isn't articulated, every individual in the retained population will apply their own. The way to answer this is to write the criterion down, in one sentence, and to test it against three hypothetical individuals. If the criterion produces three different answers, the criterion isn't yet a criterion.
Five Sequencing Approaches, Reviewed
Designing the structure fully before selecting anyone into it
The structure is drawn, layer by layer, on paper. The selection happens once the structure is signed. This is the textbook approach, and for good reason: the people who are selected into the new structure need to know that the structure is the answer to a question, not a justification for a decision that was already made. The motion earns its place when the criteria for selection are stable, when the volume of change is large enough that a partial design would force two redesigns, and when the legal and consultation obligations require a settled design before conversations can begin. The motion falls short when the design is treated as a constraint rather than a hypothesis. A structure that has not been tested against the people who will live inside it's a structure that will be re-designed twice. The cost of the approach is the cost of getting the design wrong twice.
Selecting the leadership first and letting them design beneath
The top one or two layers of the new structure are selected first. They're then given the authority, and the time, to design the layers beneath them. This earns its place when the strength of the retained organisation is in its leadership rather than in its process, and when the leaders selected have a track record of designing teams that work. The motion is also the right one when the criteria for selection at the lower layers genuinely depend on who is selected at the top, which is more often true than leadership teams admit. The motion falls short when the selected leaders don't, in fact, have the skill to design. A leader who has been successful as an operator isn't, by that fact alone, a leader who can design a structure. The cost of the approach is the cost of selecting the wrong top layer. A bad structure underneath a good leader can be fixed. A bad leader underneath a good structure can't.
Phasing function by function
The restructure is rolled out one function at a time. Each function is settled before the next one begins. The motion earns its place when the volume of change is large enough that the organisation can't absorb it all at once, when the interdependencies between functions are limited, and when the consultation obligations are easier to discharge one function at a time. The motion falls short when the phasing itself becomes the story. A phased restructure reads to the retained population as a series of rolling threats, each one of which raises the question of who is next. The cost of the approach is the cost of the anxiety it generates in the population that has not yet been phased.
A single announcement covering everything at once
The whole restructure is announced in a single event, with a single timeline, and a single set of selection criteria. The motion earns its place when the volume of change is small enough to be absorbed in one motion, when the criteria across the structure are genuinely the same, and when the legal obligations are best discharged by a clean, single event. The motion falls short when the announcement lands before the layer above the affected population has been personally briefed. A single announcement is only as good as the conversations that have already happened. The cost of the approach is the cost of getting the conversations wrong, because the announcement is irreversible in a way that a phased approach isn't.
Letting attrition reshape the organisation without an announcement at all
The restructure happens through the natural attrition of the population over a fixed period, with no formal announcement. The motion earns its place in exactly one situation: when the strategic direction of the organisation is genuinely uncertain, and the leadership wants to preserve optionality. The motion falls short in every other situation. The cost of the approach is the cost of the silence. A restructure that happens by attrition is a restructure that the retained population is left to interpret without help. The cost of interpretation is always higher than the cost of communication.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Retained leadership stable, design settled | Under two hundred | Tight team, recent honest one-to-ones, no signals of drift | Anxiety that is not visible in the data | Quiet sensing only; do not over-engineer the motion |
| Layer above the cut not yet briefed | Two hundred to six hundred | Mid-cap, single function cut, leaders above the cut uninformed | The story is being written without you | Personal conversations with named individuals above the cut, before the announcement |
| Cross-functional restructure with named retentions | Six hundred to fifteen hundred | Multiple functions affected, top performers identified, terms not yet discussed | The strongest people reading the silence as a signal | Proactive retention of named individuals, structurally framed |
| Strategic restructure with retained leadership risk | Over fifteen hundred | Divestment or transformation, leaders two levels above the cut unsettled | The restructure is the wrong vehicle for the conversation | Pause the reorganisation and have the strategic conversation first |
| Phased restructure mid-flight | Any | Function-by-function, current phase done, next phase announced | The retained population reading each phase as the next threat | Brief the layer above the next phase personally, before the next announcement |
| Edge case: top performer resigns mid-process | Any | Single named individual, signal of intent, no formal resignation | The signal of intent is itself a contagion event | Same-day conversation, personally, by the leader they most trust |
| Quiet sensing surfaces a recalculating individual | Any | Behavioural signal, no formal signal, individual is in a layer above the cut | The individual will write the story without you | One conversation, this week, by a named person they already trust |
| Whole restructure at risk of being the wrong vehicle | Over fifteen hundred | Strategic context unsettled, restructure forced by timeline | The structure will not solve the question being asked | Defer the restructure until the strategic question has a settled answer |
Naming Who Is Affected, and When
| Group | What they need to know | When they need it | What happens if they hear it elsewhere first |
|---|---|---|---|
| The leadership group above the affected population | The direction, the criteria, what is off the table, what their own role looks like in the new structure | Before the design is final; updated as the design settles; personally before the announcement | They write the design themselves, badly, and the work of the next quarter is undoing what they built |
| The layer immediately above the affected population | What the criteria are, how the selection will be made, what their role is in the conversations with their own teams | Two to three weeks before the announcement, in a setting where they can ask questions | They have to choose between telling their teams the truth and protecting the plan; they will protect the plan, and the trust cost will be paid later |
| The affected population | The structure, the criteria, the timeline, the support available, the consultation process if applicable | At the announcement, with a private conversation on the same day | The plan is declared illegitimate and the consultation process, where one exists, becomes adversarial |
| The retained population below the affected layer | What the structure means for them, what their role looks like, what the criteria were | At the announcement, followed by team conversations within forty-eight hours | They apply their own criteria, get the wrong answer, and start looking |
| The retained population in adjacent functions | What the structure means for the function, who their new counterparts are, what changes in the cross-functional working | At the announcement, followed by function-to-function conversations within a week | They write a story about why their function was not affected, and the story is worse than the truth |
| The board and the executive committee | The full picture, including the retention risk, the criteria, the timeline and the named retentions | In advance of the announcement, with a private session to test the assumptions | The board learns the cost of the plan after it has been committed to, and the conversation is about damage rather than design |
The Retention Risk You Create by Saying Nothing
The people who are most likely to leave because of a restructure are almost never the people the restructure is about. They're the people in the layer above the cut, the people in adjacent functions whose role is unchanged but whose context isn't, and the people who carry the institutional knowledge that doesn't appear on any org chart. They're not affected by the plan. They're affected by the absence of a plan to tell them about it.
The strongest of them won't wait for the announcement. They will hear a rumour, or they will notice that their manager has been in meetings they're not in, or they will see a calendar invite they were not copied on. They will do one of two things. They will either go to their manager and ask, or they will go to their network and ask. The first is the path you want them to take. The second is the path you've allowed by leaving the first one closed.
The way to identify these people before the announcement is to walk the layer above the cut and write down, for each individual, three things: how recently they have been in a room with the leader they most trust, how visible their next role is inside the new structure, and how many options they have in the external market. The first two you can answer from inside the building. The third you can only answer by asking, and the asking is itself the conversation.
A senior manager in this position isn't asking for reassurance. They're asking whether the plan is the kind of plan they want to be part of. The answer isn't always yes. Sometimes the right answer is that this isn't the kind of plan they should be part of, and the right motion is to help them find a role elsewhere. The retention motion isn't about keeping everyone. It's about keeping the people whose retention would change the kind of business you're running.
What to Put in Writing
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The criteria for selection into the new structure | The CHRO or design lead | Before the selection grid is final | The selection being challenged as arbitrary, and the consultation process becoming adversarial |
| The retention risk assessment for the layer above the cut | The CHRO or design lead | Before the announcement | The strongest people in the retained population leaving because the plan never named them |
| The briefing record for each leader above the cut | Each leader, with a copy held centrally | Within forty-eight hours of the conversation | The leader having to choose between honesty and loyalty six months later, when the conversation has been forgotten |
| The criterion the retained population can apply to test whether they are affected | The CHRO or design lead | Before the announcement | Every individual in the retained population applying their own criterion and getting the wrong answer |
| The timeline, including the consultation process where one exists | The CHRO, with legal review where applicable | Before the announcement | The timeline being interpreted as either rushed or deliberately slow, both of which damage trust |
| The list of named retentions, with the rationale and the motion | The CHRO or design lead | Before the announcement | The retention conversations being treated as transactional, by the people involved and by the people who observe them |
| The communication to the affected population, including the support available | The CHRO or design lead, with legal review | At the announcement | The support being discovered to be thinner than promised, after the consultation has started |
| The communication to the retained population, including the structure and the criteria | The CHRO or design lead | At the announcement, with team conversations within forty-eight hours | The retained population writing the story without you |
| The post-announce sensing plan, including the signals to watch for | The CHRO or design lead | Before the announcement | The post-announce attrition being discovered three months later, when it is too late to address |
Questions to Ask Before You Commit
On the criteria. Ask: what is the explicit criterion the retained population can apply to test whether they're affected. A bad answer sounds like "it depends" or "we will work it out function by function". A good answer sounds like one sentence that produces the same answer for three hypothetical individuals.
On the named retentions. Ask: who, by name, is the retention motion built around. A bad answer sounds like "the senior leadership group" or "the high-performers in the affected functions". A good answer sounds like five to eight names, with a one-line rationale for each.
On the conversations above the cut. Ask: who has personally briefed each of the named individuals in the layer above the cut, and when. A bad answer sounds like "they have been told the restructure is happening" or "their manager has been briefed". A good answer sounds like a name, a date, and a one-line summary of what was said.
On the criterion for the retained population. Ask: what story is the layer above the cut telling itself right now, and who has checked. A bad answer sounds like "we've not seen any signals". A good answer sounds like the two or three senior leaders in that layer, the question they were asked, and the answer they gave.
On the timeline. Ask: what is the consultation process where one applies, and what is the local advice that has been taken. A bad answer sounds like a timeline without a legal sign-off, or a sign-off without a documented conversation. A good answer sounds like a timeline, a legal review, and a documented record of the advice taken. Note that the rules differ by jurisdiction, and you should confirm locally before any timeline is set.
On the post-announce sensing. Ask: what signals will you watch for in the first thirty days, and who will own the response. A bad answer sounds like "we will see how it lands". A good answer sounds like three or four specific signals, the named person who will watch for them, and the motion that will be triggered if they appear.
On the worst case. Ask: if you lose one of the named retentions in the first six months, what does the plan look like. A bad answer sounds like "we would replace them". A good answer sounds like the specific consequence, and the motion that would have prevented it.
On the success measure. Ask: how will you know, two quarters from now, whether the plan worked. A bad answer sounds like "the new structure is in place". A good answer sounds like a measure of the kind of business you're running, including the retention of the people you needed to keep.
The cost of getting this wrong
The cost of a reorganisation that loses the people it was meant to keep isn't a recruitment line. It's the cost of the project that runs late because the person who was going to lead it has gone. It's the cost of the customer who was the contact for the senior manager who resigned in the second week. It's the cost of the technical knowledge that walked out of the building and into a competitor who is now hiring against you. None of these costs appear on an invoice. All of them are real, and all of them compound. The plan that read as a success two weeks after the announcement will read very differently two quarters later, when the retained population has settled into its post-restructure shape and the shape isn't the shape the plan described.
So the question to ask, before the plan is locked, isn't whether the design is right. The question is whether the motion you've built to keep the people the design needs is as rigorous as the motion you've built to select the people the design doesn't need. If it isn't, the design won't survive its own success.
When you are ready to go further
HROpsLab is a review publication, not a vendor. We don't sell software, we don't supply payroll services, and we don't give legal advice. What we do is compare the categories of tool that HR teams use to make decisions like this one, and we do it independently. If you're weighing a workforce planning platform, an HRIS, or an HR analytics product that will help you surface the signals described above, our comparison work is the place to start. It isn't a substitute for the conversation you need to have with the people in your own building. It's a way to make that conversation easier.
Frequently Asked Questions
How do I plan a restructure without losing the people I needed to keep?
Start with the names, not the structure. Write down the five to eight people whose departure would change the kind of business you're running, and design the retention motion around them as individuals before you design anything else. The structure can be rebuilt if it doesn't work. The people can't. The motion needs a senior leader, a personal conversation, and a clear statement of the role, the scope and the trajectory. The conversation needs to happen before the announcement, not after.
Who do I tell first?
Tell the layer above the affected population first, and tell them personally. The structure of the conversation is more important than the content. Each individual needs to hear it from a named person they trust, in a setting where they can ask questions, and with enough information to test the plan against their own situation. The mechanism is the relationship, not the message.
How much should I say before the design is final?
Say the direction, the criteria, and what is explicitly off the table. Don't say the conclusions. The detail that's not yet ready isn't the part that needs to be heard. The part that needs to be heard is the principle that the plan is being built around, and the principle that some outcomes aren't being traded. A senior manager who knows the principle can hold the uncertainty. A senior manager who only knows the structure will fill the silence with the worst version of the uncertainty.
How do I handle the people who are not affected?
Brief the layer above the cut personally. Brief the retained population at the announcement, and follow it with team conversations within forty-eight hours. The conversation with each team needs to be specific to that team, not generic to the population. Generic reassurance produces generic anxiety. Specific reassurance produces specific commitment. The criterion the retained population can apply to test whether they're affected needs to be articulated before the announcement, not after.
Should I announce everything at once, or phase it?
Phase it if the volume of change is too large to be absorbed in a single motion, or if the interdependencies between functions are limited. Announce it all at once if the criteria across the structure are genuinely the same, and if the legal and consultation obligations are best discharged in a clean single event. Either way, the conversations with the layer above the affected population need to happen before the announcement, not at it.
What do I do when my strongest performer resigns mid-process?
Have the same-day conversation, personally, with the leader they most trust. The signal of intent is itself a contagion event, and the response needs to be visible to the population within twenty-four hours. The conversation isn't about retention. It's about understanding what changed, and whether the change is reversible. If it isn't reversible, the right motion is to manage the departure well, and to communicate to the retained population that the change is being absorbed.
How long should I wait before judging whether the restructure worked?
At least two quarters. The plan will read as a success two weeks after the announcement. It will read differently one quarter later, when the retained population has had time to settle. It will read differently again two quarters later, when the projects that depend on the retained leadership have either landed or not. The test isn't whether the new structure is in place. The test is whether the kind of business you're running is the kind of business you designed it to be.
How do I rebuild trust afterwards?
The work is in the next restructure, not this one. Trust is rebuilt by what you do when the next difficult decision lands. The retained population will be watching how the criteria are applied, how the leaders above the cut are briefed, and how the named retentions are treated. The signal isn't what you say about trust. The signal is what you do about it. The motion to rebuild is the motion to do the next decision well, and to do it visibly well, so that the people who stayed have a reason to believe the next decision will be the same.
Make the next decision the best one yet.
