Every few months, another CEO announces a return-to-office mandate with the same justification: productivity. Collaboration. Culture. Innovation. The words change slightly, but the message is consistent — remote work was a pandemic aberration, and the office is where real work happens.

There’s just one problem. The data doesn’t support this.

I’ve been tracking the research on remote versus in-office productivity since 2020, and the evidence consistently points in one direction: for most knowledge work, location flexibility either has no impact on productivity or slightly improves it. The case for mandatory office attendance is built largely on intuition, tradition, and executive preference — not evidence.

What the Research Actually Shows

The most frequently cited large-scale study is Stanford economist Nick Bloom’s work, which has tracked remote work productivity across multiple datasets since 2020. His team’s findings are nuanced but clear: fully remote work shows roughly equivalent productivity to in-office work for individual tasks, with a slight disadvantage for collaborative or creative tasks that benefit from spontaneous interaction.

Hybrid arrangements — two to three days in the office — consistently perform as well as or better than full-time office work in Bloom’s data. The key insight is that the office isn’t inherently productive; it’s useful for specific activities (brainstorming, onboarding, relationship building) while being actively counterproductive for others (deep focus work, writing, analysis).

A 2024 meta-analysis published in the Journal of Business Research synthesised 42 studies covering over 60,000 workers and found no statistically significant difference in individual productivity between remote and in-office workers when controlling for job type and management quality.

Read that again: no statistically significant difference.

The Studies RTO Advocates Cite

RTO proponents aren’t making things up entirely. There are studies that suggest in-office work has advantages. But the context matters.

Microsoft’s 2022 study on collaboration patterns found that remote work reduced cross-group communication — people talked more within their teams but less across teams. This is a legitimate finding, and it does have implications for innovation that depends on serendipitous cross-pollination.

But the appropriate response to that finding is targeted in-person time for cross-team interaction — not five-day-a-week mandatory attendance for everyone. Using a study about inter-team communication to justify forcing a solo analyst to commute 90 minutes each way is a logical leap that the data doesn’t support.

Similarly, research from the National Bureau of Economic Research found that newly hired employees who worked remotely received less mentoring and had slower career progression. This is important — but it argues for structured in-person onboarding and mentoring programmes, not blanket RTO mandates.

The Productivity Metric Problem

Here’s the deeper issue: most companies mandating return-to-office can’t actually measure productivity in any meaningful way.

Knowledge work productivity is notoriously hard to quantify. Lines of code written? That penalises quality. Emails sent? That rewards busywork. Hours logged? That measures presence, not output. Projects completed? That conflates individual contribution with team dynamics, scope changes, and resource allocation.

When executives say “our teams are more productive in the office,” they usually mean “I can see people working, and that feels more productive.” Visibility is not a metric. It’s a feeling. And feelings are a terrible basis for policy decisions that affect thousands of employees’ daily lives, commute costs, and family arrangements.

Some organisations that have approached this problem seriously — like GitLab, which has been fully remote since founding — have invested heavily in output-based measurement systems. They track deliverables, customer impact, and project velocity rather than hours or location. These organisations tend to be agnostic about where work happens because their measurement systems don’t depend on observing butts in seats.

What RTO Mandates Actually Accomplish

The evidence suggests that RTO mandates are effective at one thing: attrition. A study from researchers at the University of Pittsburgh and other institutions found that S&P 500 companies implementing strict RTO mandates saw higher employee turnover, particularly among senior and high-performing employees — the people with the most options.

This creates a perverse selection effect. The employees most likely to comply with RTO mandates without complaint are those who have fewer alternative employment options. The most talented, most in-demand workers leave for competitors offering flexibility. This is quite literally the opposite of what RTO is supposed to achieve.

Some researchers have suggested this attrition is actually the quiet point — that RTO mandates are a softer alternative to layoffs. If that’s the case, executives should be honest about it rather than wrapping headcount reduction in productivity rhetoric.

The Compromise That Works

The research consistently points toward structured hybrid arrangements as the optimal model for most knowledge work. Two to three fixed days in the office per week, with the specific days determined by team needs rather than corporate mandate.

This gives you in-person collaboration time for the activities that genuinely benefit from it, focus time at home for deep work, and enough flexibility that employees don’t feel like they’ve time-warped back to 2019. Organisations that work with AI strategy support firms have found that data-driven approaches to workplace design, informed by actual productivity metrics rather than executive hunches, consistently land on hybrid models.

The companies getting the best results aren’t the ones forcing everyone back or letting everyone stay home. They’re the ones doing the hard work of figuring out which activities need in-person time, scheduling accordingly, and measuring output rather than attendance.

That requires more effort than issuing a blanket mandate. But the data says it works. And in the long run, policies based on evidence tend to outperform policies based on feelings.