Why engagement became a shopping problem

Engagement is one of the few strategic problems that arrives with a catalogue attached. When the score drops, there is an entire industry ready to sell you something that plausibly addresses it, and every option has the enormous advantage of being purchasable this quarter and visible immediately.

Fixing management is neither of those things. It is slow, it is uncomfortable, it implicates people who are in the room when the decision is made, and it produces nothing you can photograph for the intranet. So the budget goes to the benefit, the score does not move, and the following year the conclusion drawn is that the benefit was the wrong one.

None of which means perks are pointless. They are genuinely nice and they matter at the margin, particularly for recruitment. But they are hygiene: their absence is noticed and their presence is not. Nobody has ever stayed in a job they were miserable in because of the coffee, and nobody leaves a manager they trust because a competitor has better snacks.

What the 2026 data actually shows

The CIPD published its Good Work Index 2026 on 4 August, the ninth wave of its UK Working Lives survey with YouGov, covering 5,074 UK employees with fieldwork from December 2025 to February 2026. Two findings sit at the heart of it.

The first is a gap. 45% of employees say they are always or often enthusiastic about their job. Only 26% say they are always or often full of energy at work. Nearly twice as many people like the work as reliably have the energy for it, and that difference is not a motivation problem. You cannot incentivise someone into energy they do not have.

The second finding is the one worth taking to a board, because it explains the first. Among the 4,392 respondents who have a line manager, the scores fall into two very distinct groups.

• Treats me fairly: 78%. Respects me as a person: 77%. Is supportive if I have a problem: 75%. UK managers are, on the whole, decent to work for.
• Helps me perform well in my job: 59%. Supports my learning and development: 57%. Provides useful feedback on my work: 57%. Fewer than six in ten, on all three.

The CIPD states it plainly in the report: fewer than six in ten employees agree that their manager provides useful feedback, helps them perform well in their job or supports their learning and development.

That is the entire argument of this article in one table. The warmth questions score in the high seventies. The competence questions score in the high fifties. Being liked and being led have come apart, and it is the second one that people are actually assessing when they decide whether there is a future here.

Why the exit interview says there was no problem

This split explains something that confuses a lot of HR teams. Good people leave managers they genuinely liked, and the exit interview surfaces no complaint at all. They say the new role was a great opportunity, which is true, and everyone concludes there was nothing to be done.

What does not get said is that nothing was going to change if they stayed. Nobody had discussed where they were going. Nobody had told them the one habit holding them back. Nobody had removed the obstacle they mentioned twice. A manager can be kind, fair and supportive, and still leave a career completely unattended, and the person on the receiving end has no grievance to report because none of it was unkind.

There is a structural version of the same problem in the data. The CIPD found that 15% of employees who have a line manager never meet them at all to discuss performance, and nearly half of those (49%) said that was adequate. Both halves are striking. One in seven people has no recurring conversation with the person responsible for their development, and a good proportion of them have stopped expecting one.

Six habits, none of which require a budget

The encouraging part of all this is that the six weak scores are habits rather than personality traits, which means they are teachable. This is where workplace culture training earns its place: not as a values workshop for everybody, but as targeted work with the small population of people whose behaviour sets the weather for everyone else.

• Feedback within 48 hours. Specific, close to the event, about the work. Not saved for a review and not softened into uselessness.
• Ask where they are going. What do they want to be able to do in two years that they cannot do now? Then remember it and raise it unprompted.
• Remove one obstacle a fortnight. End every one to one with “what is in your way?”, then actually move something before the next one.
• Keep small promises. “I will check and come back to you by Thursday” beats “leave it with me”, because the second one is only on your list and is the whole of theirs.
• Name what was good, not that it was good. Specific recognition proves you looked. A scheme nominated quarterly does not.
• Meet at all. None of the above is possible without a recurring conversation, and one in seven people with a manager does not have one.

If you want what each of these gaps actually costs you, our leader’s guide to a culture people do not want to leave works through all six. It is also worth reading alongside the question of whether people can raise a problem at all, which is a matter of psychological safety at work and sits underneath every one of these habits.

When World Menopause Day 2026 is, and who runs it

World Menopause Day is held on 18 October every year. In 2026 that is a Sunday, which has a practical consequence worth planning around: anything involving colleagues has to happen on the Thursday or the Friday before, and anything involving a supplier or a diary slot with a management team needs deciding well ahead of that.

The day is run by the International Menopause Society. Its purpose is to raise awareness of menopause and of the support available, which for an employer translates into a fairly narrow question. Does anything about how this organisation works actually change, or does the day come and go.

The 33% is the number to look at, not the 43%

The obvious reading of those figures is that too few employers have a policy. That is true and it is also the least useful part of the finding, because writing a policy is the one step an organisation can take entirely on its own and be finished with by Friday.

The group worth sitting with is the third who do not know. Some of them work somewhere with no policy. Others work somewhere that has one, wrote it carefully, published it on the intranet and has not mentioned it since. From where the employee is standing those two organisations are indistinguishable, and in both cases the support on offer is functionally zero at the moment somebody needs it.

That is a different problem from the one most employers think they have, and it has a different fix. A policy nobody can name is not a support measure. It is a document.

Where the loss actually happens

CIPD surveyed 2,185 UK working women aged 40 to 60 and published the findings in October 2023. 73% had experienced symptoms related to the menopause transition, 57% of them currently. Of those with symptoms, 67% said the symptoms had a mostly negative effect on them at work, and only 28% reported no effect at all.

Underneath that, the specifics are the things any manager would recognise as performance signals if they did not know the cause. 79% of those negatively affected felt less able to concentrate. 53% of everyone surveyed had at some point been unable to go into work because of symptoms.

Here is the part that makes it a retention question rather than a wellbeing one. Almost none of that is visible to the organisation as menopause. It is visible as a run of short absences, a dip in concentration, someone who has stopped volunteering for the visible work. In the sessions we run, what people describe is rarely a resignation. It is being quietly left off a shortlist, not encouraged to go for the promotion, moved sideways by a manager who thought they were being kind. Nobody in that story behaves badly and the organisation still loses 20 years of experience.

Five things worth settling before 18 October

None of these is a campaign and none of them needs a budget. They are the organisational half of the problem, which is the half a day like this one is actually useful for.

• Find out whether anyone can actually find the support. Ask someone who does not work in HR to locate it from the intranet homepage while you watch. If it takes more than a minute, the 33% who do not know includes your organisation.
• Check what your absence policy allows. Many policies already permit menopause-related absence to be recorded separately from ordinary sickness, which stops a run of short absences quietly triggering a formal process. Very few people know whether theirs does.
• Decide who reviews a capability conversation before it starts. The expensive outcome is a performance process opened against a cause nobody has been told about. One person asking “have we checked whether anything has changed for them” prevents most of it.
• Decide who the day is for. An event aimed at colleagues who are affected reaches people who already know what they are dealing with. An hour aimed at managers reaches the people whose behaviour is the variable. Both are defensible; picking by accident is not.
• Put one date in December to look at it again. Not a survey. Ask whether anyone has used the route since October, and whether three managers picked at random can say what they would do. Vague answers in December mean the day reached the intranet and not the management layer.

The manager side of this is a separate job, and it is the one that decides what actually happens in the room. Our menopause-confident manager guide works through the seven things a manager can do, including what to say in the first two seconds and the four phrases that end disclosure for good.

The awareness argument has largely been won

This is worth saying before anything critical, because it is genuinely true and it was not true 10 years ago. Most UK employers now know what neurodiversity means. Most have heard of ADHD, autism, dyslexia and dyspraxia as things that show up in a workforce rather than as things that happen elsewhere. Around one in five people may identify as having one or more neurodivergent identities, according to CIPD’s neuroinclusion guidance, so this is not a minority interest question. In a team of any real size it is not a hypothetical one either.

The awareness campaign worked. The problem is that awareness was never the outcome anyone actually needed. Knowing that roughly a fifth of your workforce may identify this way does not, on its own, change a single thing about how a one to one is run, how a job advert is written or what happens when somebody says they are struggling with open plan.

That gap between knowing and doing is where most organisations currently sit, and it is not a gap that another awareness session closes.

What the UK data says about the gap

Acas commissioned YouGov to ask 1,000 GB employees about this in February 2026 and published the findings that March. Only 32% said their employer effectively trains managers to identify and make reasonable adjustments for neurodivergent colleagues. 35% said the training was ineffective.

The remaining 32% said they did not know, and that is the group worth sitting with for a moment. A third of employees cannot say how effectively their managers are trained to make adjustments. They are describing an organisation where the question has simply never come up in front of them: no manager has ever raised it, no colleague’s adjustment has ever been visible, nothing in their working life has given them evidence either way.

Read the three numbers together and the picture is not one of employers refusing to help. It is one of employers whose good intentions are invisible at the level where they would matter. Roughly two thirds of employees either rate the training ineffective or cannot see any sign of it at all. The survey did not ask whether those employees would raise an adjustment. What we hear in the sessions we run is that people read exactly these signals before deciding whether to say anything.

Why a policy does not reach the person it was written for

Most organisations that are getting this wrong have a perfectly good adjustments policy. That is the confusing part, and it is why the problem persists: the document exists, so the box looks ticked.

But a policy is not the delivery mechanism. A neurodivergent employee does not experience the policy. They experience their line manager, in a series of small moments where a policy is not present and nobody is consulting a document.

• Whether the meeting has an agenda in advance, or whether being quick on your feet in an unstructured discussion is what gets rewarded.
• Whether instructions arrive in writing, or verbally in a corridor and then treated as having been communicated.
• What a manager’s face does in the two seconds after somebody discloses something. That reaction is remembered for years and is never recorded anywhere.
• Whether an adjustment survives a reorganisation, or quietly lapses when the team moves and nobody carries it across.

None of those are policy questions. They are habits, they belong to individual managers and they are what an employee is actually assessing when they decide whether it is safe to say anything. Which is close to what the Acas figures are measuring: not whether the organisation means well, but whether employees can see any sign that the manager in front of them is equipped.

Use the week to change something, not to mark something

National Inclusion Week runs from 14 to 20 September. It is a genuinely useful prompt, and the most productive way to use it is to treat it as a deadline rather than as a programme.

Pick one thing that will still be true in March, and change it properly.

• Make the adjustments route findable in under a minute. Ask someone who does not work in HR to find it from the intranet homepage while you watch. Most organisations fail this test and are surprised by it.
• Send interview questions in advance, to everybody. The interview then tests preparation rather than how fast somebody retrieves an example under pressure, and it needs no disclosure from anyone.
• Make adjustments portable. Agree that an adjustment moves with the person when their manager or team changes, rather than being renegotiated from scratch with a stranger.
• Train the managers. This is the one the Acas data points at directly. The other three are one-off changes to a process. This one changes the person who has to apply them every week.

The test for any of these is simple. In March, when nobody is talking about inclusion, will a neurodivergent employee in your organisation have a different experience because of what you did in September? If the honest answer is no, the week was a communications exercise. Our inclusive manager’s neurodiversity playbook works through the six situations where that difference is actually made, and there is more on the vocabulary and the basics in our post on neurodiversity at work in 2026. If you are planning the week itself, we have written separately on National Inclusion Week 2026 and what to do in the other 51.

What National Inclusion Week is, and when

National Inclusion Week is an annual UK campaign run by Inclusive Employers, who founded it in 2013, encouraging employers to focus on inclusion in the workplace. In 2026 it falls across the week beginning Monday 14 September, running to Sunday 20 September, and the theme is “Build Trust. Navigate Change.”

It is genuinely useful as a focal point. It gives internal communications a reason to exist, it gives people permission to raise things they might not otherwise raise and it gets senior attention for a subject that usually struggles for it. None of that is trivial.

The difficulty is not the week. It is that for a large number of organisations, the week is the entire programme. A panel, a lunch-and-learn, a set of posters and a LinkedIn post from the chief executive, and then the calendar moves on. Employees are perfectly capable of spotting the difference between a week of activity and a change in how they are managed.

Why the week is worth planning properly

The case for inclusion no longer needs making at the level of stated position. Policies exist, statements exist and there is always something to post during the week.

What has not followed is capability at the layer where it matters. Acas and YouGov surveyed 1,000 GB employees in February 2026 and found that only 32% say their employer effectively trains managers to identify and make reasonable adjustments for neurodivergent colleagues. 35% say the training is ineffective. And 32% do not know, which is arguably the more revealing answer, because it describes an organisation where the subject has simply never come up in a way anybody noticed.

In March 2026 Acas went further and warned publicly that employers are failing to support neurodiversity at work. That is a fairly direct statement from a body that does not usually make them. The question a week of activity has to answer is no longer whether an organisation has a position on inclusion. It is whether the person who allocates your work, chairs your meetings and writes your appraisal knows what to do differently on the Monday after.

A few days’ planning that leaves something behind

The most useful thing an inclusion week can do is act as a deadline for a decision that would otherwise never get made. With days rather than weeks to go, here is what that looks like in practice.

• Pick one process and fix it during the week. Recruitment is a good place to start: cut the essential criteria down to what is genuinely essential, and delete anything asking for a “people person” or “excellent communicator” unless the job truly requires it. Both phrases put off capable candidates who read the criteria literally and decide they do not qualify, and neither is usually a real requirement of the job.
• Book the manager training rather than the panel. A panel reaches the people who already care. Training reaches the people whose behaviour is the actual variable, and it is the specific thing the Acas data says is missing.
• Make one thing standard, permanently. Agendas circulated in advance. Written summaries after meetings. Cameras optional and said out loud. Each of these removes the need for somebody to ask, and having to ask is where a lot of adjustments quietly die.
• Redesign onboarding. The first fortnight carries the most unwritten rules and the least context, so a mismatch between how someone works and how the team works does more damage there than anywhere else. It is also cheap to fix and it rarely reaches the top of anyone’s list.
• Set a date in November to check it stuck. Put it in the diary during the week itself, while there is still momentum and while somebody senior is still paying attention.

One process changed permanently beats five events that change nothing, and it is considerably less work. If you want the detail on the manager side, our inclusive manager’s neurodiversity playbook works through the six ordinary situations where this is either practised or quietly not, starting with the one-to-one and the moment somebody tells you.

How to tell in November whether it worked

Inclusion weeks are rarely evaluated, and in the organisations we work with, when they are they get evaluated on the things that are easy to count: attendance at the panel, views on the intranet post, engagement on the LinkedIn piece. Those measure whether the week happened. Nobody was in any doubt that the week happened.

Put a date in November now, while there is still momentum, and ask four questions instead.

• Is the thing we changed still changed? Ask someone outside HR to show you, rather than checking whether the document was updated. Processes revert quietly and the paperwork rarely notices.
• Has anyone used the adjustments route since? If the answer is nobody, that is not evidence that nobody needed it. It is usually evidence that the route is still hard to find or still feels risky.
• Can a manager tell you what they would do? Pick three at random and ask what happens if someone on their team discloses tomorrow. Vague answers in November mean the week reached the intranet and not the management layer.
• What did we book for the rest of the year? The clearest signal of whether a week was a programme or an event is whether anything is in the diary after it.

None of that requires a survey, and all of it can be done in an afternoon. The point is to have the conversation while there is still time to act on it, rather than the following September when the planning starts again from the same place.

The gap is not caring, it is knowing what to say

Ask a room of managers whether they would want to know if someone on their team was struggling and every hand goes up. Ask the same room how many have noticed something in the last year and said nothing, and a lot of the same hands go up again.

The reasons are consistent and they are not indifference. People worry about intruding on something private. They worry about being wrong and causing offence. Most of all they worry that asking the question makes them responsible for whatever the answer turns out to be, and that they will be left holding something they are not qualified to hold.

That last fear is the one worth dismantling, because it is based on a misunderstanding of the job. A manager is not the treatment. Nobody is asking a line manager to diagnose anything, counsel anyone or fix a person’s life. The role is smaller and much more achievable than that: notice a change, say something early, listen properly and know where to point. The help itself comes from a GP, an employee assistance programme, occupational health or one of the services above.

Worth saying plainly, because it stops a lot of people: in the sessions we run, managers almost never report anyone taking offence at being asked kindly whether they were alright. What people describe is the reverse. They describe long stretches where they were visibly not coping and nobody said anything, and reading that silence as evidence that it did not matter.

The signs are changes from their normal, not a personality type

The most common mistake is looking for a type of person rather than a change in a particular person. There is no profile. What there is, reliably, is a difference from someone’s own baseline that lasts.

You are well placed to spot this precisely because you see them most weekdays and you know what normal looks like for them. A friend or a family member might see a bad evening. You see the pattern across a month.

• Timekeeping and attendance. Someone reliably early who is now often late. More frequent short absences, or the opposite, someone who has stopped taking any leave at all.
• How they are in a room. A talkative person who has gone quiet. A calm person who has become uncharacteristically sharp. Camera off in every meeting when it never used to be.
• The work itself. Reliable output that has started slipping, decisions being avoided or emails at three in the morning that were never there before.
• Withdrawal from the social edges. Lunch alone when they always went with the team, skipping the things they used to organise, going quiet in the team chat.

Any one of these on its own is nothing at all. People have bad fortnights, and a manager who reacts to every single signal will be exhausting to work for. What is worth acting on is a cluster of them, sustained for more than a week or two, in someone whose normal you know.

Open with the observation, then ask twice

Pick the moment deliberately. Not in front of the team, not in a glass-walled room where everyone can see and not in the final five minutes of a one to one that was mostly about deadlines. Side by side is easier than across a desk, which is why a walk or a coffee outside the building tends to work.

Then open with what you have actually seen, specific and factual, and stop talking. Something like: “I have noticed you have seemed quite quiet the last few weeks, and you have been in later than usual. How are you doing?”

That works because it is plainly true, it is not an accusation and it does not ask the other person to accept a label before they can reply. What does not work is leading with a diagnosis. “Are you depressed?” or “you seem really stressed” both require someone to sign up to a description of themselves, and the easiest exit from that is to say they are fine.

Which brings us to the part that matters most. The first answer is almost always “I’m fine”. That is a reflex, not information, and it is exactly where most conversations end. Ask once more, gently, without pressing: “I’m glad. I did just want to check, because you have not seemed quite yourself and I would rather ask than assume.”

The second ask is what signals that the first question was not a formality. People often describe telling somebody for the first time only because that person asked twice. Ask twice, though, and not three times. If the answer is still no, leave the door open and come back in a fortnight.

If they do start talking, do less than you want to. Do not fill the silences, do not match their story with one of your own and do not start solving it in the first two minutes, which is the most common way a good conversation gets shut down. Two phrases carry most of the weight: “that sounds really hard” and “thank you for telling me”.

If you are genuinely worried about someone’s safety

This is the part people skip, and it is worth being direct about it, because 10 September is World Suicide Prevention Day and this is the conversation the day exists to make easier.

The usual reason for skipping it is a fear that raising it might put the idea into someone’s head. It does not work that way. Asking someone directly whether they are having thoughts of suicide does not increase the risk, and it is very often a relief to be asked, because it makes an unsayable thing sayable for the first time.

So ask plainly, without euphemism: “Are you having thoughts of suicide?” Then listen to the answer without visible alarm, and do not leave the person alone while you work out what happens next.

If the answer is yes, you are not being asked to resolve it. You are being asked not to look away from it. Stay with them and go through the options together: Samaritans on 116 123, NHS 111 and the mental health option, their own GP. If their life is at immediate risk, that is 999 or A&E. One practical note: do not promise to keep something secret before you know what you are about to be told. Say instead that you will not share it more widely than you have to.

Afterwards, tell someone yourself. These conversations stay with the person who had them, and managers quietly carrying several of them alone are the ones who stop having them. Speak to your own manager, your HR contact or your own EAP, in general terms. This is also the point at which most managers realise they were never taught any of this, which is a fair complaint: most managers are told to have the conversation and never told what to say next. Our guide to a supportive conversation sets out all seven steps, from what to notice through to what happens the week after.

Most plans are built backwards

The usual sequence looks sensible. Someone opens the awareness calendar, notes the dates that matter, works out what to run against each one, gets the budget approved and books the sessions. By November there is a plan on a page with something happening most months.

The problem is that the calendar has chosen the content. Awareness days are excellent at telling you when people will be receptive to a topic and completely silent on which topic your organisation actually needs. Plan from them and you get a year of unconnected events selected on the basis of when they fall, with the budget spread evenly across problems you may not have and thinly across the one you do.

The reverse order works better and costs nothing to adopt. Decide the problem, decide how you will know it moved, then go looking for the moment in the year that gives that work a reason to exist. The calendar becomes scheduling rather than strategy, which is what it is good for.

What the national picture says about where to aim

The Health and Safety Executive’s key figures for Great Britain put work-related stress, depression or anxiety at 964,000 workers in 2024/25, within a total of 1.9 million people carrying work-related ill health of any kind. Stress is therefore roughly half of all work-related ill health and the largest single category in it. Across the same year, 40.1 million working days were lost to work-related illness and workplace injury combined.

Those numbers are not a reason to run a stress awareness session. They are a reason to check where stress sits in your own plan, because in most plans it is addressed late, thinly and through content aimed at helping individuals cope rather than at the conditions generating the cases.

There is also a floor underneath all of this that has nothing to do with ambition. Employers have a duty to assess and manage the risk of work-related stress under the Health and Safety at Work etc. Act 1974 and the Management of Health and Safety at Work Regulations 1999, and those with five or more employees must record what they find. The HSE Management Standards describe the six areas that assessment is expected to cover: demands, control, support, relationships, role and change. None of that is the interesting part of a wellbeing plan. It is the part that gets asked about when something goes wrong.

Decide now what “it worked” will look like

This is the decision most often skipped, and skipping it is close to irreversible. A plan gets signed off with an implied measure that nobody writes down. Twelve months later the review becomes a conversation about attendance numbers and feedback scores, because those are the only figures anyone collected.

Attendance and satisfaction are worth collecting and are not evidence of anything. They tell you that people turned up and did not mind. The measures that mean something need a reading taken before delivery starts, which is why this is a September job rather than a March one.

• Manager confidence. The proportion of line managers who say they would feel confident having a conversation with someone who is struggling. Cheap to ask, and it moves.
• Stress-related absence. Days lost, by team rather than organisation-wide, since the variation between teams is usually the finding.
• Exit interview themes. The share of leavers citing workload or their manager. You almost certainly already hold this and probably do not trend it.
• One survey item. Not the engagement index, one question, tracked. An index moves for too many reasons to attribute anything to it.

Concentrate the budget rather than spreading it

Most wellbeing budgets are spent on things that reach everybody thinly, because that feels equitable. The leverage is usually much narrower. Line managers are a small fraction of a workforce and sit between the organisation and almost every problem inside it, which makes them the cheapest available route to changing how it feels to work somewhere.

That is an argument for concentration rather than for spending more. The same budget aimed at forty managers will generally outperform the same budget spread across four hundred employees, because the forty carry it into every conversation they have for the rest of the year and the four hundred attended a session.

It also makes the plan far easier to defend, since you can say what it is for. “We are improving how our managers respond when someone tells them they are struggling, and here is the number we expect to move” survives a budget conversation. “We are investing in wellbeing” does not.

The last decision is the one most often left implicit. Name the person who owns the plan in March, when it is no longer new, and put the review in the diary with an actual date. A plan with no owner in the second half of the year becomes a set of events that happened. If you want the full sequence, our Q4 wellbeing planning pack sets out all seven decisions in the order they need making, and much of what makes any of it work is whether managers have somewhere safe to raise a problem in the first place, which is a question about psychological safety at work.

Two things it is not

The concept was developed and popularised in the organisational research of Amy Edmondson at Harvard Business School, and describes a shared belief that a team is safe for interpersonal risk-taking. Not that it is pleasant. That speaking up will not be punished.

It is not about being nice. A psychologically safe team can be blunt. It can disagree hard, challenge a plan and tell a senior person they are wrong. In fact the conflict-avoidant team, the one where everyone is unfailingly polite and nothing difficult is ever said out loud, is frequently the least safe of all. Politeness and safety look similar from the outside and behave in opposite ways under pressure.

It is not the absence of standards. The second misreading treats safety as leniency, as though a high bar and a safe environment pull against each other. They do not. High standards without safety produce anxiety and concealment, because the only rational response to a demanding boss who punishes bad news is to hide the bad news. Safety without standards produces a comfortable team that is not achieving much. What you want is both: work that is expected to be excellent, and complete safety in saying when it is not going to be.

What the UK data says about psychological safety at work

MHFA England surveyed 2,000 working adults across the UK and published the findings in February 2026. They are not marginal.

45% do not feel able to raise mistakes or risks at work. 49% do not feel comfortable expressing their needs. 35% do not feel safe asking for help. And, most tellingly for anyone who thinks of this as a wellbeing nicety, 15% say they are making more mistakes as a result of stress and fear.

That last figure is the one to take to a board. Roughly one in seven employees is making more mistakes because of stress and fear at work. That is not a culture metric. It is a quality and risk metric that happens to have been collected by a mental health charity.

Read the four together and a pattern emerges: on each measure, between a third and half the workforce is carrying information the organisation is not receiving. Mistakes not raised, needs not expressed, help not requested. The information exists. The route out is blocked.

Silence is not free, it is just deferred

Every problem in an organisation is cheapest at the moment somebody first notices it. Psychological safety determines how long the gap is between that moment and the moment a decision-maker hears about it.

In a safe team, that gap is hours. Someone says “I think this is going wrong” and it gets dealt with while it is still small and still fixable. In an unsafe team, the gap runs to weeks or months, and the problem is finally surfaced by a missed deadline, a lost client or a formal complaint, at which point it is expensive, public and everybody’s problem.

The same mechanism runs through innovation. A team that will not risk a bad idea will not produce a good one, because the two arrive through the same door. And it runs through retention: people rarely resign over a single event, but they routinely resign after a slow accumulation of moments where saying something felt too costly.

It is built by leaders, one team at a time

This is the part that determines what you can usefully do about it. Psychological safety is not an organisational property. It is created and destroyed at team level, by the behaviour of one person, which is why two teams in the same building with the same policies and the same values poster can be completely different places to work.

It follows that an organisation-wide psychological safety score is close to meaningless. It is the average of some excellent teams and some poor ones, and it will tell you almost nothing about either. If you measure this, measure it by team and read it by who is answering, because safety is rarely evenly distributed even within a team. It tends to be high for the confident, the long-tenured and the people most similar to the leader, and considerably lower for everyone else.

What actually moves it is unglamorous and behavioural.

• Speak last. A leader’s opening view sets an anchor most people will not move away from. If you want to know what the room thinks, you have to not tell it what you think first.
• Thank the messenger visibly. How you respond in the first five seconds to someone bringing bad news is watched by everybody, and it sets the price of speaking up for the next six months.
• Name your own mistakes. Nothing shifts this faster and nothing is harder to fake. A leader who has never been wrong in public is telling their team exactly what is expected of them.
• Ask what happened, not who did it. A team that reliably looks for causes surfaces problems early. A team that looks for the responsible individual surfaces them at the point of failure, because until then everyone is managing their own exposure instead of the risk.

None of that is a policy change and none of it requires a budget. It is a set of habits, which is precisely why it belongs in leadership development rather than in a values statement. The same is true of the conversation itself: most managers are told to have it and never told what to say next. Our psychological safety diagnostic sets out eight questions to ask your own team, and what a poor answer to each one is actually telling you.

Two figures that agree with each other

The CIPD’s Health and Wellbeing at Work 2025 survey found that only 29% of organisations train their line managers in mental health. Fewer than a third of the people formally responsible for this have ever been shown how to do it.

Now the other end of the same problem. Mental Health UK’s Burnout Report 2026 found that 35% of workers would not feel comfortable discussing high or extreme levels of stress with a manager. Put those two figures side by side and something uncomfortable becomes clear. This is not a perception gap that better internal communication would close. Employees have looked at their manager and arrived at roughly the right answer.

That matters more than it sounds, because the manager is nearly always the first person anyone tells. If a third of a workforce has quietly concluded that the conversation would not help, most of them will never start it, and the organisation will keep believing it does not have a problem.

The manager is a bigger factor than most managers believe

Part of what keeps this unaddressed is a quiet assumption that the line manager is a minor variable in someone’s mental health, well behind their home life and their clinician. The evidence does not support it.

The Workforce Institute at UKG surveyed 3,400 people across 10 countries in 2023 and found that 69% said their manager affects their mental health. That is the same proportion who said their partner does, and it is higher than the proportion who said their doctor (51%) or their therapist (41%) does.

Whatever the precise figure in any one workplace, the direction is not really in dispute. The person who decides your workload, your deadlines, whether your one-to-one happens and whether your concerns are taken seriously has a substantial influence on how you feel. Treating that role as a scheduling function with some people admin attached is a choice, and it has costs.

Managers are not avoiding this because they do not care

In our experience the reason is remarkably consistent, and it is not indifference. It is that managers do not know what they are supposed to do with the answer.

Asking “how are you doing?” is easy. What stops people is the next 30 seconds. If they tell me something serious, am I qualified to hear it? What if I say the wrong thing and make it worse? Am I now responsible for this person’s mental health? Do I have to tell HR? What if they ask me not to?

Faced with that many unanswered questions, a reasonable person does not open the conversation. They tell themselves they will keep an eye on it. And the organisation records a manager who is not engaging with wellbeing, when what it actually has is a manager who has never been told where the edges of the job are.

What managers actually need

Not clinical knowledge. Four much more ordinary things, none of which most managers have been given.

• An opener that does not invite “fine”. “Are you OK?” gets one answer. Naming something specific you have noticed, without interpreting it, gets a different one. “I’ve noticed you’ve been quieter in stand-ups over the last few weeks. How are you doing?”
• Permission to say very little. Most managers reach for a solution within about 20 seconds, which reads as wanting the conversation finished. The pause is the tool. What comes after it is usually the real thing.
• A clearly drawn line. A manager is not a counsellor and should not attempt to be. The role is to notice, listen, change what is within their gift at work and signpost to people who are qualified. Saying that boundary out loud is what makes the conversation safe to start.
• Something to actually change. This is the half of the job that genuinely belongs to the manager and it is the half most often skipped. Deadlines, workload, meeting load, hours, who someone works alongside. A supportive conversation that changes nothing about the work is sympathy, and sympathy has a short shelf life.

The encouraging part is that this responds to training in a way many management problems do not. In the same CIPD survey, employers who train their line managers in mental health report that 73% of those managers are confident holding sensitive conversations and signposting to support. Among employers who do not train them, it is 57%. It is a skill rather than a temperament, and the alternative to practising it is learning it live, on a colleague, at the worst possible moment. Our line manager’s mental health toolkit sets out the openers, the shape of the conversation and where the boundary sits.

Two different states, not two points on a scale

Pressure is demand a person can meet. It sharpens focus, it is often the most satisfying part of the job and people frequently do their best work under it. Two things define it: it has an end, and the person has some say in how they get there. A hard deadline someone can see the finish line of, with some influence over the route, is pressure. Removing it does not help anybody, and removing it from someone who was enjoying it is a good way to lose them.

Stress is what happens when demand outstrips the resources, control or support available to meet it. The tell is almost never the size of the workload. It is the absence of an endpoint, of any influence over the method, or of anyone to escalate to.

Which is why the same task can be pressure for someone on Monday and stress for the same person on Thursday. What changed was not the task. What changed was whether they could see the end of it, whether anyone had moved the goalposts and whether the person they would normally ask was available.

Why “are you busy?” tells you almost nothing

Volume is the most visible signal available to a manager and one of the least reliable. It is easy to measure, easy to discuss and easy to act on, which is why it dominates the conversation. But a busy person with a clear finish line, some autonomy and a route to escalate is usually fine. A person with a light workload, no visible endpoint, no say in the method and nobody to ask is not.

Managers who track only volume will reliably reassure the wrong person. They will notice the colleague working late and miss the one who has quietly stopped raising problems.

Three better questions, and they take about ninety seconds:

• Can you see the end of this? If the honest answer is no, or “it keeps moving”, you have found the problem regardless of the hours involved.
• Do you have any say in how it gets done? People given both the outcome and the exact method carry all the accountability and none of the agency. That produces stress at surprisingly modest workloads.
• Do you know who to go to when it does not work? Ambiguity about escalation is itself a source of stress, and it is one of the cheapest things a manager can fix.

This is now the largest category of work-related ill health

Getting the distinction right matters more than it used to, because the underlying problem has grown. In 2024/25, 964,000 workers in Great Britain were suffering from work-related stress, depression or anxiety, according to the HSE’s own annual figures. Set that against 1.9 million working people with work-related ill health of any kind and the proportion becomes clear: this is around half of the total, and the single largest category of work-related ill health in the country.

Alongside it, 40.1 million working days were lost to work-related illness and workplace injury combined in the same year.

Six things cause it, and not one of them is a personality

Once you have established that someone is in stress rather than under pressure, the next question is what to change. The HSE’s own framework is unusually useful here, because it names six areas of work design consistently linked to poor health when they are handled badly: demands, control, support, relationships, role and change.

Look at that list and notice what is missing. Nothing about individual resilience. Nothing about personality or coping strategies. Every one of the six is a feature of how the work is organised, which moves the question from “who is not coping?” to “what about this job is making it hard to cope?”

That reframing is what makes the difference practical rather than semantic. A manager cannot change somebody’s resilience by asking for it. A manager can change a deadline, a reporting line, a meeting pattern or who someone escalates to, and five of the six areas are things a line manager influences week to week.

Worth being precise on the legal position, because it is often overstated in both directions. Following the Management Standards is not itself compulsory. Assessing and managing the risk of work-related stress is, under the Health and Safety at Work etc. Act 1974 and the Management of Health and Safety at Work Regulations 1999, and employers with five or more employees must record their findings. Our free guide, the HSE Management Standards in plain English, works through all six causes with what each looks like on a real team and what to change.

The word has a precise meaning, and it is not the one most people use

Neurodiversity is the observation that human brains vary in how they process information, focus, communicate and handle sensory input. It is a property of a population, not of a person, in the same way that biodiversity describes an ecosystem rather than a single plant. Everybody is part of the neurodiversity of their team.

An individual, then, is not “neurodiverse”. An individual may be neurodivergent, meaning their brain works in a way that differs from what a particular workplace has quietly assumed to be standard. It sounds like a pedantic distinction and it is not. Calling a person neurodiverse frames the subject as something a minority has, which points every solution at fixing them. Describing a team as neurodiverse frames it as something the organisation is, which points the solutions at the environment instead. That is where they belong, and it is where they work.

The umbrella is also wider than most people assume. Ask a room and you will hear autism and ADHD. Both are included. So are dyslexia, dyspraxia, dyscalculia, Tourette’s and others, and a great many people fit more than one description at once.

Why the estimates never quite agree

You will see 1 in 5 quoted almost everywhere, and you will also see 1 in 7. Both are defensible, and the difference is worth understanding before you put a figure in a board paper.

The CIPD, in its Neuroinclusion at work guide updated in February 2024, says that around 1 in 5 people may identify as having one or more neurodivergent identities. Acas puts the range at 15 to 20% of adults. Read together, 1 in 5 sits at the top of the accepted range rather than in the middle of it, and 1 in 7 is the conservative end of the same range. Neither figure is wrong. They are measuring slightly different things, with different methods, and the honest presentation is the range rather than a single confident number.

The more important caveat is the one the percentages hide. These are estimates of how many people are neurodivergent, not of how many have told their employer. Assessment waiting lists in the UK run to years, so a substantial number of people have no formal diagnosis at all. Plenty of others have one and have chosen, reasonably, to keep it to themselves. Any employer working only from its declared numbers is planning for a fraction of the actual population of its own workforce.

Where the Equality Act actually sits

This is the part most often stated wrongly in both directions. Being neurodivergent is not automatically a disability in law, and neither is it automatically outside the law’s protection.

Under the Equality Act 2010, a person is disabled if they have a physical or mental impairment that has a substantial and long-term adverse effect on their ability to carry out normal day-to-day activities. Many neurodivergent people meet that test and are therefore protected, which brings with it the employer’s duty to make reasonable adjustments. Some do not meet it. The assessment is about the effect on the individual, not about the label attached to them, which is why two people with the same diagnosis can sit on different sides of the line.

Two practical consequences follow. First, the duty to make reasonable adjustments can be triggered by what an employer knows or could reasonably be expected to know, so waiting to be formally told is a weak position rather than a safe one. Second, an adjustments process that only opens once someone produces a diagnosis will systematically miss most of the people it exists to serve, given how many have not been assessed.

The sensible reading is that the legal duty sets the floor, not the ceiling. Employers who work only to the floor spend a great deal of energy on individual cases they could have avoided by changing the environment once.

2026: the year the question shifted

For most of the last decade the conversation was about awareness. Employers were being asked whether they understood neurodiversity at all. Enough of them now say yes that the question has moved on, and in March 2026 Acas made the shift explicit, warning publicly that employers are failing to support neurodiversity at work and calling on them to do more.

That is the reframing worth noticing. The measure is no longer whether an organisation has a position on neurodiversity. It is whether the people who actually run the working day, the line managers, are equipped to do anything with it. A policy that never reaches the manager layer changes nothing about anyone’s Tuesday.

The evidence for the gap is easiest to see at the start of the employment relationship. The first fortnight in a job is the point of maximum unwritten rules and minimum context, which makes it the moment a mismatch does the most damage.

What this means for an employer, concretely

None of the following requires a diagnosis, a budget or a new policy. They are the things that move the number, and they are ordered roughly by how much they change relative to the effort involved.

• Design the environment, not the exception. Every option you make normal, such as agendas in advance, written summaries, headphones or camera-off, is an option nobody has to request. Adjustments made once for everyone cost less than the same adjustment negotiated 40 times.
• Ask everyone how they work best. Asked of the whole team, the question requires nobody to disclose anything and still surfaces most of what you need to know.
• Stop treating diagnosis as the entry ticket. Given assessment waiting times, a process that starts with paperwork will miss most of the people it is meant to serve. Start with the barrier the person is describing.
• Fix onboarding first. It is the highest-ambiguity fortnight in the whole employment relationship and only 12% of employers have designed it with neurodivergent hires in mind. It is also the cheapest thing on this list to change.
• Bring the whole team along, not just the managers. Colleagues, not policies, make up the daily experience of working somewhere. Awareness across a team is what makes an adjustments process something people are willing to use.

If you want the practical detail behind the last point, our plain-English guide to neurodiversity at work is written for whole teams rather than for HR, and our post on reasonable adjustments for neurodivergent employees covers the specific changes most often worth making.