Research · Corporate Wellness

Does Corporate Wellness Actually Pay Off? What the Largest Trial Ever Run Found

By Muhammad Dzulhisham · MSc Sports Coaching, NSCA-CSCS

Ask a company why it runs a wellness program and the pitch is usually some version of: healthier employees means fewer sick days, lower healthcare costs, and better retention. It's a reasonable-sounding chain of logic. It's also one that's now been tested directly, at a scale most corporate-benefits decisions never get — and the answer is more complicated than either the vendors selling these programs or the skeptics dismissing them as "perk theater" usually let on.

The largest trial ever run on this question

In 2019, Harvard researchers Zirui Song and Katherine Baicker published the first peer-reviewed, large-scale, multisite randomized controlled trial of a workplace wellness program, in JAMA.¹ This wasn't a survey or a before-after comparison at one company — it was a clustered randomized trial across 160 worksites at a large US warehouse-retail chain, running from January 2015 through June 2016: 20 worksites (4,037 employees) offered the wellness program, 140 worksites (28,936 employees) didn't, and outcomes were tracked for both groups.

At 18 months, the program worked exactly as intended on the things employees self-reported:

But on every outcome the program was actually supposed to move — clinical health markers (BMI, blood pressure, cholesterol), healthcare spending and utilization, absenteeism, and job performance or tenure — the study found no significant differences between the program and control groups after 18 months.

Three years later, the same answer

Song and Baicker followed the same worksites for a third year and published the extended results in Health Affairs in 2021.² More time didn't change the conclusion. Randomization into a program worksite still showed no detectable effect on total medical spending (a −$298 per-employee difference that wasn't statistically significant), no effect on absenteeism, and no effect on employee tenure. The self-reported behavior gains held up — employees at program worksites were still more likely to report exercising and managing their weight — but nothing downstream of that moved.

"We did not observe a financial return on investment in the employment and claims measures we examined." — Song & Baicker, Health Affairs, 2021

That's about as direct a null result as health-services research produces. If the largest, most rigorously designed test of this question found no ROI after three years, "wellness programs pay for themselves" is not a claim the best available evidence actually supports — at least not for a program built the way this one was.

So is it all perk theater? A broader look says: not exactly

One landmark trial is strong evidence about one program, run one way, at one type of company. A 2020 systematic review and meta-analysis in the International Journal of Environmental Research and Public Health pooled 47 randomized controlled trials of workplace interventions — 19 of them (11 on absenteeism, 7 on productivity, 5 on work ability) had data usable for meta-analysis.³ The picture that comes out of pooling many smaller, more varied trials is genuinely more mixed than the single large trial's clean null result:

• Absenteeism: across moderate-quality trials, interventions reduced sickness absence by roughly 2.65 days on average (95% CI, −4.49 to −0.81; p<0.001) — a real, statistically significant effect. The review specifically flags that individualized interventions with fewer than 10 sessions were the most effective format.
• Productivity: moderate-quality studies showed a significant improvement (effect size 0.33; 95% CI, 0.07–0.59; p=0.01) — but the result became inconsistent once lower-quality studies were folded in, meaning study design quality itself was doing real work here.
• Work ability: no significant effect (0.48; 95% CI, −0.47–1.43; p=0.32) — the one outcome where the broader evidence agrees with the large trial's null finding.

Put the two pieces of evidence together and a specific pattern emerges, rather than a contradiction: a broad, low-touch program offered to an entire large workforce — gym subsidies, health coaching calls, biometric screenings, the standard corporate-wellness package — is the format the large RCT tested, and it's the format that produced a clean null. The interventions that showed a real, significant absenteeism effect in the meta-analysis were smaller and individualized, not broad and generic.

What this means for choosing — or building — a program

The honest conclusion isn't "wellness programs work" or "wellness programs don't work." It's that the evidence draws a real line between two different things that get sold under the same label. A generic, mass-distributed perk — the kind most large-employer programs default to because it's the easiest to administer at scale — is exactly the design that's now been tested most rigorously and come back with no measurable return. A smaller, individualized program is the format the broader evidence actually associates with a real absenteeism effect.

That's not a reason to assume any individualized program automatically works either — nobody has run a large RCT on ours, and I won't claim one has. But it is a reason to ask a sharper question before buying into corporate wellness than "does the vendor have a program." The research says the design of the intervention — individualized versus generic, structured versus a menu of perks — is doing more of the work than whether a wellness line-item exists on the benefits budget at all.