Sooner or later, everyone who proposes a wellbeing budget meets the study. For Maria, who runs a doctoral school of sixty researchers, it arrived in a budget meeting, slid across the table by a dean who does his reading: a major trial found workplace wellness programmes do not work, so why are we funding one? The dean is not wrong about the trials. In a randomised study across one hundred and sixty worksites of a large US employer, published in JAMA, a multicomponent wellness programme produced some healthier self-reported behaviours and no significant differences in clinical health, absence or job performance after eighteen months [1]. The Illinois Workplace Wellness Study found no productivity effects however productivity was measured, with null results precise enough to rule out most of the positive findings the earlier literature had claimed [2]. A large UK study found participants in individual wellbeing offerings, mindfulness apps, stress classes, resilience modules, no better off than colleagues who never took part [3].
And the verdict closest to Maria’s world sits in PNAS. Feldon and colleagues followed 294 PhD students in the life sciences across 53 US institutions to test the boot camps and short-format trainings graduate programmes run to build research skills, formats backed at the time by nearly 28 million dollars in active National Science Foundation and National Institutes of Health funding. Across 115 measured variables, skill development, scholarly productivity, socialisation into the academic community, they found no evidence of effectiveness [4]. Note carefully what that study was not about: wellbeing. It was about research skills, the training universities believe in most. The format failed anyway.
Table Of Contents:
- What was actually convicted
- What behavioural change actually requires
- The question that settles the budget meeting
- FAQ
- References
What was actually convicted
Here is what I would have Maria say when the study comes back across the table: I believe every one of these trials, and they convicted the right defendant, which is not training. Look at what the tested programmes shared. Modules attended once or occasionally, alongside the working week. Apps used alone in whatever moments survived the day. Boot camps compressing everything into days. Participation thin, practice unspaced, application to real work absent, and follow-up measurement nonexistent. In the language I use when designing formats: these were awareness interventions, and they were being sentenced for failing to deliver behavioural outcomes, a crime they were never equipped to commit in the first place.
The reason is the oldest curve in learning science. Ebbinghaus mapped the decay of newly learned material in the 1880s, and a modern replication confirmed his curve holds [5]; I take it apart in my article on the Ebbinghaus forgetting curve. Most of a single exposure is gone within days. A lunchtime module against the forgetting curve is a snowball against summer, and an entire industry priced its products as if summer were negotiable. The null trials did not discover that wellbeing training fails. They discovered, at magnificent scale and public expense, that format mismatched to outcome fails, in wellbeing and in research skills alike. The common factor in every conviction is not the topic. It is the format.
What behavioural change actually requires
None of the requirements are secret and none are proprietary. Lasting change needs spacing, repeated distributed practice rather than one exposure, the best documented counter to forgetting there is, evidence in my post on spaced repetition. It needs application in context, the technique used in the person’s real working week the same week it is learned, so it attaches to daily structure rather than to a classroom. It needs weeks, because that is how long a behaviour takes to become self-sustaining. And it needs measurement before and after, because without it, neither provider nor buyer ever learns which condition was missing.
Those are the conditions Paul and I build our formats around, and they are why we publish our numbers. In our February 2026 cohort, measured with validated instruments in week one and week eight, every wellbeing scale improved with medium to large effects, the share of researchers in the at-risk range for anxiety roughly halved, self-rated work performance rose by more than a point on the ten point WHO scale, and working hours did not change. The full results, limits and non-movers included, are in my article on whether resilience training actually works. And two months in, the majority of techniques were still in daily use by the clear majority of participants, the measured opposite of the decay the null trials imply, with the application numbers in my piece on whether training sticks.
One frankness requirement before anyone quotes me against the JAMA trial: our cohort has no control group and theirs did, so I cannot and do not claim proven superiority. My claim is narrower and, in this industry, rarer: we meet the conditions the learning science demands, we measure with the same class of instruments the trials used, and we publish what we find, including our own nulls. The burden the big trials place on every provider is not despair. It is measurement.
The question that settles the budget meeting
The null studies hand Maria a sharper procurement question than does this work. The question is: what is this format built to achieve? If the objective is awareness, a strong short format is the right instrument and should be judged as one; I never rank formats, I match them to purposes, and the trials above are a museum of mismatches. If the objective is behavioural change, then the provider must show the mechanics every convicted programme lacked: spaced practice over weeks, application inside the participant’s real work, and before and after measurement they are willing to publish. A provider who has these will produce them in one meeting. A provider who does not will produce a satisfaction score, and Maria now knows what a satisfaction score is: the sound the forgetting curve makes when nobody is measuring. The fuller anatomy of a behavioural format is in my article on what resilience training involves.
FAQ
Do workplace wellness programmes work?
The straight reading of the big trials: the dominant model does not. The JAMA randomised trial across one hundred and sixty worksites found no significant effects on clinical health, absence or performance after eighteen months, the Illinois study found no productivity effects, a large UK study found participants no better off than non-participants, and a PNAS study found the same null for short-format research-skills training for PhD students. What every tested programme shared was format: occasional modules, apps and boot camps with no spaced practice, no application to real work, and no outcome measurement. Training built on the learning science, distributed practice over weeks applied in context, is a different intervention, and the fair test of any provider is whether they measure and publish.
Why do one-off wellbeing workshops rarely change behaviour?
Because the forgetting curve eats single exposures first. Ebbinghaus showed in the 1880s, and the Murre and Dros replication confirmed, that most newly learned material decays within days without reinforcement, and a PNAS study by Feldon and colleagues found short-format boot camps for PhD students ineffective across 115 variables for the same structural reason. One-off formats have a real purpose, creating awareness and opening a topic. Choose them for that purpose, and stop sentencing them for a different one.
How can I tell whether a training provider will actually deliver change?
Ask three things and count the hesitations. How is practice spaced across the programme, and where do participants apply the techniques inside their real work? What do you measure before and after, and with which instruments? Will you show me results including what did not move? The programmes convicted in the big trials could not have answered one of the three. A provider building for behavioural change answers all of them before the coffee is cold.
References
[1] Song, Z., Baicker, K. (2019). Effect of a workplace wellness program on employee health and economic outcomes: a randomized clinical trial. JAMA, 321(15), 1491 to 1501.
[2] Jones, D., Molitor, D., Reif, J. (2019). What do workplace wellness programs do? Evidence from the Illinois Workplace Wellness Study. The Quarterly Journal of Economics, 134(4), 1747 to 1791.
[3] Fleming, W. J. (2024). Employee well-being outcomes from individual-level mental health interventions: cross-sectional evidence from the United Kingdom. Industrial Relations Journal, 55(2), 162 to 182.
[4] Feldon, D. F., Jeong, S., Peugh, J., Roksa, J., Maahs-Fladung, C., Shenoy, A., Oliva, M. (2017). Null effects of boot camps and short-format training for PhD students in life sciences. Proceedings of the National Academy of Sciences, 114(37), 9854 to 9858.
[5] Murre, J. M. J., Dros, J. (2015). Replication and analysis of Ebbinghaus’ forgetting curve. PLOS ONE, 10(7), e0120644.
Last Updated on August 16, 2026 by Dr Nadine Sinclair
Dr Nadine Sinclair
Dr. Nadine Sinclair is a molecular biologist, strategy consultant and co-founder of Mind Matters. A scientist by training and at heart, she conducted her doctoral research at the Max Planck Institute for Biophysical Chemistry. She has 30,000+ hours of hands-on project management experience, built across 18 years as a strategy consultant at McKinsey and as an independent consultant, working with research institutions, foundations, pharmaceutical and biotech companies and governments. She is the co-creator of the Personal Resilience Indicator and the author of On Track. Since 2018, over 1,500 researchers from more than 30 institutions have trained with Mind Matters.