28 October 2026 · 12 min read

Why good people leave, and what a team can do before the exit interview

The resignation is the last thing that happens. By the time it reaches your diary, whatever caused it has already done its work, and the meeting you are about to have can only describe the damage.

Staff turnover runs backwards: the part you could have influenced is over long before the part you can see.

The Tuesday meeting

You run a team of nine in a company of seventy. If somebody asked who holds the team together, you would say Callum before you said yourself. He has been there for years.

On Tuesday he asks for half an hour. You know what it is before he sits down. He has accepted an offer elsewhere, and you hear yourself ask the question every manager asks: what would it take? You are in your director's office that afternoon. By Thursday there is a better package on the table.

Callum thanks you for it and goes anyway. The exit interview, run by HR, records "a new challenge". It is polite, accurate and no use to anyone.

Months later a former colleague mentions, over coffee, that Callum had asked last spring about leading the platform migration. He was told it was a conversation for the year-end review. Nobody raised it again. What would have kept him cost nothing, and it was available in the spring.

Why do good people leave?

iHire's 2026 survey of 1,220 US workers asked about toxic workplaces, not about pay, so read this as one slice of why people go rather than the whole picture. Of US workers who had met a toxic workplace, 79.1% blamed unethical, unaccountable or unsupportive leadership; 68.9% had worked for an employer they consider toxic, and 47.6% had quit a job because of one (iHire, 2026 Toxic Workplace Trends Report release, PR Newswire, 25 June 2026).

The 79.1% is a share of the workers who had met a toxic workplace, not of everyone surveyed, and "toxic" is the respondent's own word. Even so, most of them pointed at the people running the place.

The CIPD's Good Work Index 2026, a weighted survey of working adults across the UK, says something similar. Line management, leadership, skills development and meaningful work have the biggest effect on all five business-related outcomes the CIPD measures (CIPD, Good Work Index 2026, July 2026). One of those five is how likely a person says they are to quit. These are associations in survey answers, which is how the report describes them, not proof that fixing one moves the other.

The same report puts a size on the leadership half. In the CIPD's 2026 survey, 85% of employees who trust their senior management to act with integrity were satisfied with their job, against 43% of those who do not; 78% would recommend their employer, against 23%; and 71% were unlikely to quit, against 48% (CIPD, Good Work Index 2026: Survey report, July 2026, Figure 3). Again that is association. People who are content may simply find it easier to trust.

Put the two surveys side by side and the same short list comes up: the manager, the leadership above them, a chance to grow, work that matters. Pay is not among the four, which does not make it irrelevant. It means the things a manager can change this week are on the list, and the reflex of reaching for a raise is not.

Why people stay: the web nobody wrote down

The useful answer to why people stay is that staying is more than the absence of reasons to leave. In 2001 Terence Mitchell and colleagues set out an idea they called job embeddedness: a person's links to other people and groups, how well the job, the organisation and the community around it fit them, and what they would have to give up by leaving. In their data it predicted both who intended to leave and who actually did, over and above job satisfaction, commitment to the organisation, the alternatives on offer and how hard people were searching (Mitchell, Holtom, Lee, Sablynski and Erez, Academy of Management Journal, 2001).

A person stays because of a web: the colleague they learn from, the work they would miss, a life the job makes possible. Strands go one at a time. A friend moves teams. A project goes elsewhere. A plan to grow is pushed to the year-end review. No single strand looks like the reason, and nobody sees the web thinning because it was never written down.

That is why satisfaction scores can flatter you. Callum would have ticked "satisfied" in the spring and meant it. A survey asks how the job feels today. The web asks what would be left if one more strand went.

Is staff turnover contagious?

Leaving also spreads. Will Felps and colleagues, working with data from a regional bank and a hospitality firm, found that what coworkers do, how embedded they are and whether they are looking, helps explain who goes on to quit, over and above the other individual and group-level predictors in their model (Felps, Mitchell, Hekman, Lee, Holtom and Harman, Academy of Management Journal, 2009). A resignation does not always stop at one. The people who worked with Callum are now doing a quiet sum of their own.

Take Sadia, who sat next to him for four years. She is not unhappy. She has started counting: who will she ask now, who else knew how the old system worked, and what did it mean that the company found a better package only once it was too late? On Friday she opens a recruiter's message she would have deleted in the spring. Nothing has gone wrong yet, so nobody says a word to her.

Some leaving is healthy, and some is pay or the job market, which no conversation on a team will change. But meaning, growth, the manager and the daily friction of the job were there before the notice, and they can be talked about before it.

The signs someone is working out whether to go

Hold these lightly. They are signs about a team, and hunting for who is next is a good way to create a next.

• The best people go quiet. Whoever used to argue in planning now agrees with everything, and it reads as contentment.
• Growth only comes up at review time. "A conversation for the year-end" is where ambitions go to wait.
• The exit interviews start rhyming. The same few reasons come back, each filed on its own.
• The plan talks more about backfill than growth. Next year is mostly a list of chairs to fill.
• A networking profile becomes team gossip. An update to one is a small fact. The talk about it says how people are feeling.
• After one goes, others ask about openings. That is Felps's finding, arriving as a corridor conversation.

What one resignation costs the people who stay

The visible cost is the one everybody counts: the search, the handover, the months before a replacement is any use. The rest goes unrecorded.

What Callum knew leaves with him, and most of it was never in a document. The people who relied on him lose a strand of their own web. And an exit nobody can explain makes those who stay recalculate, in private.

The package is the expensive part, in a way no spreadsheet shows. A team that has watched a better offer appear on Thursday has learned that asking did not work and leaving does.

Do exit interviews and counter-offers work?

There are two easy answers, and both arrive after the decision.

The exit interview happens, by definition, once the person has decided. Whatever it finds, it finds too late to use on them. It makes a good black-box recording and a poor steering wheel.

The counter-offer is an improved package offered to someone who has already decided to go. It answers the offer on the table, not the reasons the person went looking in the first place. It is a late, costly fix.

Neither is wrong to do. Both start at the resignation, and the causes started well before it.

How to improve employee retention this week

Three things to run this week, none of which needs a product.

Book a stay conversation with the person you would least like to lose

Half an hour, one to one, three questions: what keeps you here, what would make you look around, and what should I do more or less of? Listen first and talk second. Then send a written follow-up within the week that says what will happen next. Stay conversations are a format practitioners use, not one the research has tested, so treat it as practice. A conversation with no follow-up is a survey with better chairs.

Make the growth ask easy to say yes to

Ask what they want to do next, and agree a first step in the same conversation: second-chairing the next one, say, rather than "let's look at it at review time". If the step is not on your team, help them find it elsewhere in the company. A good person who moves across the organisation is a good person the company keeps.

When somebody leaves, check in with the people who worked closest to them

Within the week, informally: "how are things for you since Callum left?" Felps's finding is why. With the leaver, say thank you, ask what would have made a difference without trying to talk them round, and leave the door open.

None of it costs money. All of it depends on you remembering, in the middle of delivery, to hold a conversation nothing is forcing. That is where it fails. It happens once, in the month after the resignation that frightened you, and then the diary fills up again.

Questions about keeping good people

How is a stay conversation different from a one-to-one?

A one-to-one is about the work in front of you, and a stay conversation is about whether the person wants to be here next year. It has no agenda items and no status updates. Any notes are there to prompt what you will do next, never to build a file on anyone, and word of that will reach the rest of the team.

What if the reason really is pay?

Then say so plainly and send it up, because a team cannot fix a pay band. Sort what you hear into two columns: what is ours to change, and what is built into the organisation or the market. Take the second column to whoever sets it, while the people are still here to benefit. Pretending a pay problem is a culture problem wastes everybody's goodwill.

Is some staff turnover healthy?

Yes: some leaving is healthy, and a manager who treats every exit as a failure ends up clinging. Someone who has outgrown the role, or is moving for reasons that have nothing to do with you, is not a fault to fix. The useful test is narrower: could the reason have been talked about in the spring? If it could, that is the one to learn from.

How MeaningFuel helps with staff turnover

The conversations are easy to describe and hard to keep up.

What a reader meets on an ordinary morning

A company chooses the values and skills it wants to work on, and each working day everyone gets one short insight on one of them. Attrition is a problem, so it cannot be chosen for the daily culture. Purpose and Appreciation can, as values, and so can Leadership & Coaching as a skill. In an ordinary week a reader might meet a line on why the work matters on Monday and one on being noticed for it on Thursday. It does not stop anybody leaving. It keeps meaning and growth in the conversation in the months before a review, which is when they go missing.

A Place Worth Staying

A Place Worth Staying is a module for that: a short course of daily lessons over about two weeks of working days, taken on by a team together, with the stay conversations in the team lead's hands. Booked for a team, it invites the people on it, and those who join get one short lesson each working day.

Each person writes down what actually holds them to the team, which people, which work, which conditions, inside work and outside it. Then they change one small thing about their work so the strongest of those shows up by design rather than by luck, an idea drawn from Amy Wrzesniewski and Jane Dutton's research on job crafting (Wrzesniewski and Dutton, Academy of Management Review, 2001). Each makes one growth ask out loud, with a first step attached. The team lists everything that makes working there heavier than the work itself, votes for the top item, and at the debrief it gets an owner and a date. The lead holds a stay conversation with each person and sends the follow-up.

That list matters because few people resign over one irritation alone, and many small ones can tip the decision: a clumsy tool, a form filled in twice, an approval that waits for days.

What Insights can and cannot tell you about who is leaving

Insights shows how much of a team takes part and how the insights land, pooled, with nobody named, and a room going quiet is worth a look. It would not have told you that Callum asked about the migration. It will never tell you who is looking, because no screen shows one person's answers, and a lead sees that the team is doing the module, never what anybody wrote. That is by design: the map of what holds someone here is only honest if the person drawing it knows nobody is reading it.

What it does not do

It does not touch pay or the job market. It sorts what the team can fix from what is built into the organisation, and sends the second pile upward as pooled themes, with nobody named. It makes no promise about who stays.

Before the notice

Good people do not leave on the day they resign. The cheapest moment to act is an ordinary working day, while staying is still the default.

• Employee Retention
• Staff Turnover
• Management
• Workplace Culture
• Stay Conversations