Two years of RTO mandates and the longitudinal research is in. The productivity gains executives expected haven't materialized, but something more nuanced has emerged, and understanding it may be the key to designing work arrangements that actually work.
Key Takeaways
The CEOs who drove aggressive return-to-office mandates in 2024 and 2025 were, in most cases, acting on a hypothesis: that proximity would restore the cultural cohesion, collaborative energy, and performance intensity of the pre-pandemic office. Two years of longitudinal data have now accumulated on the outcomes of those mandates , and the picture is considerably more complicated than the hypothesis suggested. Individual productivity, the primary stated justification for most RTO policies, has not measurably improved. But something else has, and the nuance matters for every HR leader trying to navigate this terrain.
A comprehensive analysis published by Stanford's Institute for Economic Policy Research in Q1 2026 reviewed 47 natural experiments created by companies that implemented staggered RTO mandates between 2023 and 2025, covering approximately 380,000 workers across the United States, United Kingdom, Germany, and Australia. The study's central finding: full five-day RTO mandates produced no statistically significant improvement in individual output productivity as measured by project completion rates, quality scores, or manager performance assessments. In technology and knowledge work functions specifically, individual productivity declined modestly in the 90 days following mandate implementation , likely reflecting disruption and elevated stress rather than any inherent output differential favoring remote work.
"The productivity argument for RTO was always on shaky empirical ground," says Dr. Emily Chang, a labor economist at Stanford who co-authored the study. "Individual knowledge work productivity is difficult to measure, and the studies that seemed to show remote work productivity declines were almost universally conducted during the chaotic early pandemic period , not in mature remote work arrangements. The evidence that bringing people back to the office makes individuals more productive is, at best, weak."
The Stanford analysis did identify meaningful differences in collaborative outcomes between fully remote and hybrid arrangements. Informal cross-team knowledge transfer , the spontaneous interactions that generate novel solutions and accelerate problem-solving , occurred at measurably higher rates in hybrid environments than in fully remote ones. New employee integration was faster and more effective when new hires had structured in-person time in their first 90 days. And in roles requiring intensive real-time collaboration , creative teams, trading desks, surgical units , in-person co-location showed genuine performance advantages.
The costs of full five-day mandates, however, were equally measurable. Organizations that implemented mandatory full-time office requirements in 2024 saw voluntary attrition spikes of 12 to 18% in the six months following implementation. The departures were not evenly distributed: they were concentrated among high performers, employees with long tenure, and senior women , precisely the cohort whose loss is most expensive to an organization. In technology companies with fully remote-capable workforces, the talent substitution problem was acute: the workers most likely to leave over an RTO mandate are also the workers with the most marketable skills and the most employment alternatives.
"The companies that implemented blanket five-day mandates in 2024 lost their most talented people to companies still offering flexibility — and then spent the next 12 months trying to hire their way back to the capability level they started at. The math was never close to working." — Dr. Emily Chang, Stanford IEPR
The most robust finding across multiple independent research streams is that structured hybrid arrangements , typically two to three designated in-office days, coordinated by team so that co-location is intentional rather than coincidental , produce the strongest combined outcomes on productivity, retention, and employee experience. The key word is "structured": hybrid models that leave in-office attendance to individual discretion produce most of the costs of remote work (reduced co-location, lower spontaneous collaboration) without producing most of the benefits (schedule flexibility, commute time savings). The value of hybrid depends entirely on whether in-office time is designed to take advantage of physical proximity or simply replicates remote work patterns in a different building.
The most important single variable in work model outcomes, multiple studies now confirm, is manager quality , not physical location. Teams with high-quality managers perform well in any work arrangement. Teams with poor managers underperform in any work arrangement. The manager variable swamps the location variable in virtually every longitudinal study that has controlled for both. This finding has profound implications for how organizations should prioritize their investments in the work model debate: the ROI on manager development almost certainly exceeds the ROI on real estate policy.
The return-to-office wars are not over , they're entering a more data-informed phase in which the organizations designing work models based on actual evidence are separating from those still relitigating the debate on intuition and ideology. The research is now sufficient to say with confidence: the question is not whether employees should work in an office, but what work benefits most from physical co-location and how to design arrangements that deliver those benefits without sacrificing the flexibility that the most capable workers have come to value and expect.
Amid political headwinds and programme rollbacks, CHROs are finding new ways to advance belonging and equity.
What separates organizations where employees genuinely feel they belong? We studied six of the best.
Companies with genuinely accessible mental health programs are seeing 24% lower voluntary turnover.
