66% of office workers fall short of “extremely satisfied” with their jobs, compared with 35% of remote workers — a near-twofold dissatisfaction gap, according to a Global Work-Life Survey conducted by Remote, the Future Workforce Alliance, and IE Business School across five continents. Pew Research Center corroborates the asymmetry from the employee side: 46% of US remote workers say they would likely leave their job if forced back to the office full-time, with women and workers under 50 expressing the strongest resistance. The pattern is consistent with Gilovich and Medvec’s long-term inaction-dominance finding — staying tethered to the office when an alternative exists generates more dissatisfaction than the recognized downsides of remote work.
The employer side tells a parallel story. An Envoy survey of over 1,100 US executives found that 80% would have handled their return-to-office mandates differently — a figure widely reported as outright regret. The nuance matters: the original research used the word “differently,” not “regret,” and the dissatisfaction was largely procedural (insufficient data, not enough employee input) rather than substantive opposition to office work per se. Still, the fact that four in five executives second-guessed their own RTO decisions suggests that mandated office presence generates friction and regret on both sides of the org chart, not just among the workers subject to it.
The main caveat is measurement. Neither “extreme satisfaction” rates nor hypothetical quitting intent are direct regret measures. The 35% and 66% figures used here invert satisfaction top-box scores, which means a “very satisfied” office worker is lumped in with a genuinely unhappy one. The 31-point gap is directionally robust across multiple independent surveys, but the absolute magnitudes should be read as upper bounds on true regret. The Remote.com survey sample is modest (1,000) and was fielded by a company with obvious commercial interest in remote-work adoption. The Pew data is methodologically stronger but asks about hypothetical behavior, not recalled experience — and people are notoriously bad at predicting what they would actually do.