Building a workplace mental health policy: what to put in it

Most workplace mental health policies fail the same way: they are written once, signed off by legal, uploaded to a shared drive, and never opened again. A workplace mental health policy India organisations can actually rely on only earns its keep when an employee in distress, or a manager who has just been told something difficult, can find it and know exactly what happens next. The test is not whether the document exists. The test is whether it changes behaviour on a Tuesday afternoon when someone is not coping.

This guide walks through what actually belongs in the policy: scope and intent, confidentiality and data handling, what managers are and are not responsible for, leave and reasonable accommodation, crisis and escalation, and the review cadence that keeps it alive. It is structured so HR can lift each section, adapt the specifics to its own organisation, and end up with something usable rather than ornamental. None of it is legal advice; treat the regulatory points as prompts to check with your own counsel.

The first section should answer three questions in plain language. Who does this cover? What is the organisation committing to? And what is it explicitly not promising? Aim to cover everyone the organisation is responsible for: full-time staff, employees covered by PF and ESIC, contract and gig workers, interns, and remote or shift workers who rarely set foot in the office. Mental health risk does not respect employment category, and a policy that quietly excludes the night shift or the warehouse floor is not really a policy, it is a perk for head office.

Intent should be concrete. Instead of a sentence about valuing wellbeing, state what the organisation will provide: confidential counselling access, defined leave provisions, manager training, and a named escalation path. The World Health Organization and the International Labour Organization estimate that depression and anxiety account for roughly 12 billion lost working days and close to US$1 trillion in lost productivity globally each year, so the business case does not need inflating. The honest framing is that untreated distress is already a cost the organisation is paying, and the policy is how it decides to pay less of it.

Be equally clear about limits. The organisation is not a clinic and managers are not therapists. Saying so in the document protects employees from well-meaning but unqualified intervention, and protects the organisation from implying support it cannot actually staff.

Employees will not use any of this if they believe their manager, or a future appraisal panel, will learn they sought help. Confidentiality is the load-bearing wall of the entire policy, and it has to be written precisely rather than reassuringly. Spell out who can see what. Clinical conversations with a counsellor stay with the counsellor. HR sees aggregate, anonymised usage data, never individual records. A manager learns only what an employee chooses to share, plus any accommodation that has been agreed, and nothing about the underlying diagnosis.

India's Digital Personal Data Protection Act, 2023 treats data concerning health, including mental health, as personal data that needs a lawful basis, a clear purpose, and genuine consent. As a practical matter, the policy should state how such data is collected, where it is stored, how long it is retained, and when it is deleted. If counselling is delivered through an external provider, name the contractual safeguards: data minimisation, no sharing of identifiable records with the employer, and deletion timelines. Vague assurances are worse than useless here, because they invite a trust that the organisation has not actually engineered.

One specific trap: assessment tools. Validated screens such as the PHQ-9 for depression and the GAD-7 for anxiety are useful, but the results are clinical data. They belong with a qualified professional, not in an HR file or a wellbeing dashboard that a manager can browse.

Managers are where policy meets reality, and most have had no training for the conversations this policy will trigger. Give them a short, explicit brief. Their job is to notice changes in behaviour, ask without diagnosing, listen without fixing, and route the person to the confidential support that exists. Their job is not to assess severity, recommend treatment, or decide whether someone is faking it.

Psychological safety is what makes the rest function. Amy Edmondson's research on teams, and Google's Project Aristotle study of what makes teams effective, both point in the same direction: people tend to speak up and disclose when they believe they can do so without being punished or humiliated. A policy can mandate a counselling line, but a manager's day-to-day behaviour is what largely decides whether anyone walks through the door. Train for that, and write the expectation down so it is reviewable rather than assumed.

Give managers the language for boundaries too. It is legitimate, and kinder, for a manager to say I am not the right person to advise on this, but here is exactly who is, than to improvise amateur counsel. The WHO's ICD-11 describes burnout as a syndrome resulting from chronic, unmanaged workplace stress, with three features: exhaustion, cynicism or mental distance from the job, and reduced efficacy. That shared vocabulary is useful because it lets a manager name a pattern without reaching for a diagnosis they are not qualified to make.

Mental health leave should not require an employee to disclose a diagnosis to a manager to access it. Define how sick leave applies to mental health on equal footing with physical illness, and decide in advance how medical certificates are handled so that the requirement does not itself become a barrier or a privacy breach. If the organisation offers anything beyond statutory leave, state the eligibility plainly rather than leaving it to case-by-case discretion, which tends to favour the confident and penalise the quiet.

Reasonable accommodation is the practical half of the policy. This is the menu of adjustments that lets someone stay at work or return to it: temporarily reduced hours, a shift change for someone whose rotation is worsening their sleep, a phased return after extended leave, or a quieter workspace. Name a few concrete examples so managers know what is on offer and employees know what to ask for. Where relevant, align it with the organisation's obligations under applicable Indian labour and disability provisions, and keep the process for requesting an adjustment lightweight, because friction is where good intentions die.

Account for the texture of the Indian workforce. Shift workers, field staff, and gig contractors may have no fixed desk and no easy private moment to make a call. PF and ESIC coverage differs across these groups. A policy that assumes a salaried desk employee with a laptop will silently exclude a large part of many Indian organisations. Write the access routes for them explicitly.

Every policy needs a clearly marked path for acute risk, written before anyone needs it, because no one improvises well in a crisis. Define what counts as an emergency, who an employee or manager calls, and what the organisation does in the moments after. Include current, locally verified crisis and helpline numbers and the steps for a situation involving immediate danger to someone's life. This section should be findable in seconds, not buried in clause fourteen.

Escalation also covers the less dramatic but more common situations: an employee who discloses to a manager, a manager who is worried about someone and unsure what to do, a counsellor who flags a pattern within the bounds of confidentiality. Map each route. Who does a manager talk to, what can they share, and what stays sealed? The point is to remove the moment of paralysis where a well-meaning person does nothing because they are afraid of doing the wrong thing.

Decide and document the rare circumstances where confidentiality may be limited, typically imminent risk to life. Be honest about it rather than promising absolute secrecy you cannot legally or ethically keep, and explain the threshold clearly so the promise that remains is one employees can rely on.

A policy that is not measured is a policy that is decaying. Set a review cadence, annually at minimum, and look at signals the organisation can collect without compromising privacy: anonymised counselling uptake, leave patterns, exit interview themes, and anonymous pulse surveys. Industry analyses of the cost of poor workplace mental health, alongside the widely cited replacement-cost range of roughly one-half to two times an employee's salary, are a reminder that the cost of doing nothing tends to show up in attrition long before it shows up in a wellbeing report. Watch the leading indicators, not just the lagging ones.

Close the loop visibly. When the policy changes because of something employees raised, say so. The fastest way to kill trust in a wellbeing programme is to ask for feedback and then change nothing, because it confirms the quiet suspicion that the document was always for show. Platforms such as AhaTherapy can supply the anonymised, aggregate signal that makes this review possible, but the discipline of actually acting on it is the organisation's to keep.

The measure of a good workplace mental health policy is not its length or its legal polish. It is whether a struggling employee can find help quickly and trust that asking for it will not cost them, and whether a manager facing a hard conversation knows what to do. Write for that Tuesday afternoon, keep every promise the document makes, and revise it as the organisation learns. That is the difference between a policy that protects people and a file that protects no one.