59% of workers regret not prioritizing work-life balance, according to Resume Now’s International Career Regrets survey of 1,000 workers across four countries. On the other side, roughly one-third wish they had taken more career risks. The resulting 1.8:1 ratio favors action-regret — an unusual outcome in the regret literature, where Gilovich and Medvec’s temporal pattern predicts that inaction regrets dominate over the long term. Career overinvestment appears to be one of the rare domains where the pattern inverts, likely because the opportunity costs are uniquely irreversible: you cannot retroactively attend your child’s first steps.
Deloitte’s workplace burnout survey quantifies the mechanism. Seventy-seven percent of professionals report experiencing burnout at their current job, and 83% say it has negatively impacted their personal relationships. The APA’s 2024 Work in America survey corroborates: 45% of workers report working more hours than they want to, and 57% report burnout-related stress. The Harvard Study of Adult Development — an 87-year longitudinal study — found that relationship satisfaction at age 50 predicted physical health at 80 more reliably than cholesterol levels. When participants reached their 80s, the dominant regret was having prioritized career achievement over relationships. These converging datasets, from different methodologies and timeframes, all point the same direction.
The main caveat is measurement asymmetry. The 59% and 33% figures come from different survey questions, not from a clean forced-choice between career and balance. A respondent could plausibly endorse both regrets simultaneously. The Harvard Grant Study tracked an elite, all-male cohort that entered Harvard in the late 1930s — a population with more career optionality than most. The Deloitte burnout figure (77%) measures stress prevalence, not regret. The directional finding (career-overinvestment regret exceeds career-underinvestment regret) is robust across all sources; the precise 1.8:1 ratio is an artifact of combining non-identical survey items and should be treated as indicative, not calibrated.