Does workplace mental-health training actually work?

We measure every talk and workshop we run. Here is what five years of it on one standard form added up to, told plainly, including the parts that are perception rather than proof.

• Since 2009, iGROW has worked with 700+ companies, delivered 7,000+ sessions and reached 70,000+ employees across Singapore and ASEAN. This paper zooms in on the slice we can measure the same way every time: 4,166 post-Covid responses on one standard form.
• Across 300-plus of those sessions, participants rated their overall experience 8.6 out of 10, and 83% gave it an 8, 9 or 10.
• Roughly 8 in 10 said they came away knowing more about their mental and emotional wellbeing (8.6/10), and about three in four said they picked up skills they could actually use (8.4/10).
• Facilitators scored highest of all, on knowledge (8.9/10) and communication (8.9/10). That matters more than it sounds: if a room doesn't trust the person at the front, nothing else lands.
• 8 in 10 would recommend the session to a colleague (8.5/10).
• These are perceptions, collected in the minutes after each session. They tell us the experience landed as useful, relevant and well delivered. They don't measure clinical change, and we won't pretend they do.
• The bigger point is a gentle challenge to a common industry habit: most wellbeing spend is never measured at all. We measure ours on one consistent form, and what you see here is only the part we can stand behind that way.

Ask most organisations what they spent on employee wellbeing last year and they'll tell you to the dollar. Ask what they got back, and you usually get a pause.

That pause isn't negligence. It's the nature of the category. Wellbeing programs tend to be bought on faith, run as one-off events, and judged, if at all, by whether the room looked engaged and nobody complained. Budgets renew or lapse on impressions. When money is tight, that's an uncomfortable place for a line item to sit, and an uncomfortable place for the person defending it in a budget meeting.

There's a second problem underneath the first. A lot of workplace mental-health content is lifted wholesale from other markets, translated lightly, and delivered into Singaporean and ASEAN workplaces as if culture, hierarchy, language and the local weight of "asking for help" were footnotes.

We're not publishing this because the numbers flatter us, though we'll admit they do. We're publishing it because the people who sign off on wellbeing spend deserve to see a provider show its working.

Everything below comes from post-session self-report, rated on a 0 to 10 scale. Next to each average we show the top-box figure: the share of people who gave an 8, 9 or 10. It's a stricter test than a plain average, because it counts only the genuinely convinced and ignores the polite middle.

The reach behind the sample

The experience landed, and kept landing

An overall mean of 8.6, with more than four in five people in the top box, is a strong signal, and it holds steady across five years and 300-plus sessions. This isn't one charismatic morning flattering the average. It's a pattern.

People felt they walked out with something

For a wellbeing program, the two questions that count are whether people feel they understood more and whether they feel they can do something differently. Both sat in the mid-8s. The "usable skills" item is the hardest thing we ask, because enjoying a talk is easy and leaving with a tool is not, and roughly three in four still put it in the top box.

The facilitators are the standout

Knowledge and communication both at 8.9, each with 86% in the top box, are the highest marks in the set. In mental-health work this isn't a vanity metric. Whether a room trusts the person at the front is often the whole difference between leaning in and quietly checking out. Credible, clear facilitation is what lets the content do anything at all.

They'd put their name behind it

Eight in ten would recommend the session to a colleague. Recommending something at work carries a small social risk that mere satisfaction doesn't, which is exactly what makes it a useful read on perceived value.

Numbers give you the shape of the response. The comments give you its texture. These are drawn from free-text feedback, lightly edited for readability, with names removed and each one attributed by sector rather than employer.

Every form ends with a quiet, revealing question: what would you like more of? We have asked it two ways as our form evolved, with a blank box and later with a fixed checklist of topics we teach. Each person saw one or the other. Read as expressed interest rather than a diagnosis, the answers tell us what is still on people's minds as they walk out.

Ask the same question with a blank box, and one word dominates

The checklist tells you what people recognise. The blank box tells you what they reach for on their own, and it is a blunter kind of honest. Across 1,712 written-in answers in three languages, the single most-written word was stress, the number-one free-text theme by a clear margin. What makes that striking is that stress is not one of the seventeen options on our checklist. We teach it, under other names. We simply never put the word in front of anyone.

In 1,712 free-text answers pooled across iGROW Pte Ltd sessions in Singapore and ASEAN, the single most-written word was "stress", named in 228 write-ins across English, Bahasa and Thai, even though stress was not one of the 17 topics on iGROW's checklist.

Read the two figures separately. The percentages above come from the 17-item checklist (1,723 respondents); the 228 write-ins come from the blank-box version of the question (1,712 answers). They are different questions, answered by different people in different years, and must not be combined.

Two patterns worth naming

The first is about vocabulary. Where we ran both versions of the question with the same client organisations, 14% of blank-box answers named stress and fewer than one in a hundred named burnout, while 55% of checklist respondents chose burnout once we offered it. In Thai, not one person in 430 wrote the word for burnout, though seventeen wrote the word for depression. The topic is the same. The word people reach for is not.

The second is quieter. A small but persistent thread of write-ins reaches for something the menu never offers: difficult workplace relationships, hard-to-manage bosses, awkward colleagues, the friction of simply getting along. It rarely appears as a formal option anywhere, yet people raise it unprompted. It is a real gap between what tends to be offered and what is sometimes actually on people's minds.

This is the section a clinical firm has to write, and the one most marketing skips.

What the data does tell us

Across a large, multi-year, multi-country sample, people consistently found our sessions valuable, relevant, well delivered and worth recommending. They felt more knowledgeable and better equipped afterwards. Engagement, perceived usefulness and facilitator credibility are real and they matter. They're the preconditions for anything downstream, and a program that failed on them would have no chance of helping anyone. On these measures, the sessions perform strongly and reliably.

What it doesn't tell us

It doesn't tell us anyone's mental health improved. It can't. These are perceptions captured in the minutes after a session, not clinical measurements, and certainly not clinical measurements taken before and after against a comparison group. We didn't track symptoms. We didn't follow people over time. We didn't compare them with employees who attended nothing.

We hold this line for one reason. The credibility of a psychologist-led firm rests on saying only what the evidence supports. A provider willing to overclaim on evaluation data is a provider you can't trust with anything harder.

Here's a confession. Left to our own devices, this is not the study we'd run. As psychologists our instinct runs hard toward method. Given a free hand, we'd want a validated instrument before and after each program, a follow-up months later, a comparison group who attended nothing, and a careful read across the participant, their manager, their peers and their leadership, ideally linked to real performance data. That's what solid evidence of impact takes, and we know it.

Five years inside real organisations has taught us how rarely those conditions show up. A few things get in the way, every time:

• Survey fatigue. Ask employees to fill in one more long before-and-after battery and your response rate quietly collapses.
• Management appetite. Most sponsors want a talk delivered, not a research program run.
• Champion bandwidth. Wellbeing usually rides on one committed HR person carrying it on top of a full job. They can't realistically marshal managers, peers and leaders into a multi-stakeholder data exercise.
• No access. Individual and company performance data sits behind walls we're rarely handed the keys to, often for good reason.

None of that is an excuse, and we won't dress it up as one. It's just the working reality of measuring anything inside a busy organisation. So we do what can be done consistently, session after session, without adding load to the very people we're there to support. It isn't the ideal study. It's a real, multi-year signal from thousands of participants, gathered under real constraints, and we'd rather show it to you with its limits marked than offer nothing while we wait for perfect conditions that never come.

To the sceptic, and we mean this warmly because scepticism is the right instinct here: you're correct that satisfaction isn't outcome. We agree, and we've said so. The question we'd put back isn't whether this is perfect evidence. It's whether disciplined, transparent, imperfect measurement beats the unmeasured faith that most wellbeing spend still runs on.

You don't need a research department to hold wellbeing spend to a higher standard. You need a few good habits and a provider who'll support them. Start with the checklist below, then read the note written for the seat you sit in.

If you like to check the working, this section is for you.

Three questions come up in nearly every conversation about this data. Here are the answers we give, with the same limits attached.