People Copy What They See

Every organisation says the right things about balance. Behaviour above sets what people believe, and where the two disagree, behaviour wins.

What people read

When leave is taken, and whether it is interrupted.

When messages are sent.

Whether the person who worked the weekend is praised.

Whether anyone senior admits a mistake.

Whether anyone senior says they are at capacity, which sanctions others doing so.

Whether illness is worked through.

None of this is stated. All of it is observed and it forms the actual policy.

The specific signals that matter most

A leader on holiday who replies to messages has told the organisation that holidays are not real. Nothing in a policy document undoes this.

Late-night sending creates a response expectation whether or not one is intended.

Praising heroics specifies the standard. "Thanks to Sam for pulling this off over the weekend" tells everyone what pulling it off requires.

Never appearing to struggle makes struggling a private failure.

Working while ill makes rest conditional on being visibly incapable.

Cheap corrections

Schedule-send. If you work late, send in the morning. One setting, and it removes the largest single source of implicit expectation.

Take your leave, fully, and be unreachable. Then say afterwards that you were.

Say when you are at capacity, out loud. This gives permission that no policy can.

Say what you got wrong, at your own level.

Praise the outcome, ask about the weekend. "This landed well — what made it need a weekend?" changes the message entirely without withdrawing the recognition.

Take sick leave when ill.

The middle management squeeze

Worth naming because it is where modelling frequently breaks.

Middle managers model behaviour they do not control. They are pressed from above and observed from below.

They work late because their own workload requires it, and their team reads this as the standard.

They cannot fix it by pretending, and the organisation cannot fix it by asking them to model differently while keeping the load.

The intervention is at their workload, not at their behaviour. Telling a drowning manager to model balance is not a plan.

What does not work

A written statement about work-life balance, with no behavioural change above it. This is worse than silence, because it adds visible hypocrisy.

A policy on out-of-hours contact that senior people are exempt from in practice.

Wellbeing initiatives launched by someone who visibly does not observe them.

"Do as I say" in an environment where people are professionally attentive to what senior people actually do.

The test

Ask people, anonymously, two questions.

Do you believe you can take your full leave entitlement without any consequence?

Do you believe you would be judged for not replying to a message in the evening?

The gap between these answers and the stated policy is the size of your modelling problem, and it is usually large.

Schedule-send and other small settings

A short list of changes that cost nothing and alter what people believe is expected.

Schedule messages written outside hours to send in the morning. One setting, and it removes the single largest source of implicit expectation.

Set a status that states your actual working hours.

Say in your absence message that you will not be checking, and then do not.

Do not reply to non-urgent things at midnight even when you are awake.

Say out loud when you are at capacity, which grants permission nothing else does.

Take your leave in full blocks.

None of this requires a policy, a budget or anyone's approval, and collectively it does more than most published statements about balance.

The two questions that measure it

The gap between stated policy and observed behaviour is measurable in one anonymous survey.

Do you believe you can take your full leave entitlement without any consequence?

Do you believe you would be judged for not replying to a message in the evening?

Segment by level and by team.

Compare against the written policy, which almost certainly says the opposite.

The size of that gap is your modelling problem, and it is usually larger than leadership expects, because leadership is the group whose behaviour created it.