The Performance Paradox.

For over a decade, employers have invested in mental health benefits under the assumption that the problem is access. More providers, faster booking, better apps. Utilization hasn't moved.

The real problem is a structural mismatch: the benefits were built for a different kind of employee than the one carrying your business today. This report examines why high-output, globally distributed workforces are systematically underserved by traditional mental health programs — and what the financial and operational consequences look like when the gap goes unaddressed.

The average EAP utilization rate sits at around 3–5% globally — and in some industries as low as 2%. Despite over a decade of serious investment, that number has barely moved. Meanwhile, everything else has. Global anxiety and depression rose 25% in the first year of COVID alone and has not corrected.

Mental health-related leaves of absence have surged 300% between 2017 and 2023, with a further 22% increase in 2024 alone — and they now represent 1 in every 10 leaves taken. For HR leaders, mental health topped the list of reasons employees requested leave in 2024 — above illness, injury, and even parental leave.

The structure of traditional EAP support requires an employee to self-identify as struggling, seek out a resource, navigate a directory or phone tree, schedule an appointment, and follow through. That sequence was designed for someone who has already decided they need help. For the majority of today's workforce — particularly high-performing, globally distributed employees — that threshold is never crossed.

Across industries, companies investing in AI and automation tools are achieving real efficiency gains — fewer people handling more complex work, faster cycles, leaner operations. What is less visible is what it costs the people left carrying the load.

According to Mercer's Global Talent Trends report, 82% of employees worldwide are now at risk of burnout — yet fewer than half of employers have redesigned work with wellbeing in mind. In the U.S. alone, burnout costs businesses an estimated $322 billion annually in lost productivity. The per-employee burden is $4,257 for salaried staff, $10,824 for managers, and $20,683 for executives.

The employees who remain after automation aren't doing less. They're absorbing more — and the benefits built to support them were designed for a different era of work.

For globally distributed teams, the pressure compounds. Traditional support exists in business hours, in scheduled appointments, in a model that assumes the stress will wait. Burned-out employees are 63% more likely to take a sick day — and mental health claims now last 24% longer than the average claim.

There is a well-documented profile of the most underserved employee in any enterprise mental health program. They are high-performing. Self-reliant. They take pride in handling pressure. They would describe themselves as stressed, not struggling — and that distinction matters enormously to how they engage with support.

McKinsey research found that 37% of employees with high self-stigma missed at least a day of work due to burnout — yet those same employees were the least likely to seek help. Two in five Americans worry they would be judged if they shared about their mental health at work.

Why high-output employees don't use traditional EAPs

• Self-stigma. High-performers often internalize the belief that needing support signals weakness — seeking help conflicts with their professional identity.
• Timing mismatch. Stress doesn't arrive during business hours. Support that requires scheduling a future appointment doesn't meet the moment it's needed.
• Friction is a hard stop. A number to call, a form to fill, a wait for an appointment — any friction is enough to abandon the attempt.
• Global employees face compounding barriers. Language, cultural norms around help-seeking, and inconsistent provider availability across countries mean EAP coverage on paper often translates to minimal real access.

A 2025 study in Practice Innovations (APA) found that 48.7% of U.S. adults with mental health challenges who use AI are already turning to general-purpose LLMs like ChatGPT for emotional support or mental health-related conversations — because the threshold to start is low enough that they don't need to have already decided they have a problem.

Early engagement data from Thoughtful's enterprise pilots tells a consistent story: the barrier to mental health support was never willingness. It was format. When employees can access support immediately — no appointment, no waitlist, available at 6am or 11pm — they engage. And they come back.

Thoughtful has also completed a randomized controlled trial with 229 participants, finding statistically significant improvements across five metrics directly relevant to workplace performance: productivity, focus, confidence, engagement, and help-seeking behavior. Effect sizes were meaningful — emotional wellbeing improved at d=0.31, loneliness decreased at d=0.36, and self-efficacy increased at d=0.23.

Where traditional EAPs sit at 3–5%, modern engagement-first platforms report utilization exceeding 40%. The population hasn't changed. The access model has.

Mental health benefits have historically been difficult to defend to finance because the ROI case relies on soft metrics. The more productive framing starts from costs already sitting in the business, unattributed to mental health because no one has connected the dots.

The WHO estimates that depression and anxiety alone cost the global economy $1 trillion annually in lost productivity. In the U.S., untreated mental health conditions cost employers over $105 billion per year. Workers affected by mental health challenges perform at an average of 72% of their full capability.

Early intervention at scale isn't a wellness expense. It's performance infrastructure — and it has a defensible ROI that belongs in the same conversation as headcount planning.

The design requirements for mental health support that actually reaches a high-output, globally distributed workforce are not complicated. But they are fundamentally different from the requirements that shaped the EAP category.

Support needs to be available when the stress actually happens — not the following Tuesday at 2pm. It needs to fit into five minutes or thirty. It needs to work in London, Singapore, and Chicago without meaningfully different experiences. It needs to not require someone to identify themselves as struggling before they can access it.

Over 90% of leaders in one Unmind survey said mental health solutions should balance AI and human support — not rely on one or the other. AI expands reach. Providers deliver depth. Neither works as well without the other.

If 95% or more of your employees aren't using what you're offering, the right question isn't how to get them to change. It's whether what you're offering was ever built for them.

The employees carrying the most — the high performers, the globally distributed, the ones who would never describe themselves as struggling — are also the ones most likely to quietly degrade and eventually leave. The cost of reaching them earlier is a fraction of the cost of replacing them. The barrier isn't budget. It's format.